{"idea":{"id":"gapfill","title":"Gapfill — building-product attributes the supplier never sent","summary":"An enrichment layer for building-product catalogues: pull supplier data from websites, inventory systems and PDFs, then fill the fields the supplier never sent. Five research passes refuted the original moat — vision-enrichment vendors DO serve building materials, including an Australian one — but confirmed a narrow documented gap. Full measurements now live in four linked research annexes. The binding constraint is no longer capability, it is whether anyone pays.","stage":"researching","category":"ecommerce / product-data-automation","next_step":"Stop researching capability; test demand. Ask one tile merchant and one architect which fields they query per-SKU versus accept as a PDF datasheet. Three conversations, and it decides everything. In parallel: can Cin7 Omni's editable field mappings target Shopify metafields, and does BIMobject have a working write API (one pass found an active OAuth2 publish endpoint, another found 410 Gone)?","risk":"Three independent absences point the same way: no forum thread or review documents the Cin7 spec-data pain; five years of ETIM permission for flooring plus a 1,020-member merchant federation produced no tile data; ETIM AU/NZ launched 2021 and its site is still a parked domain. Either greenfield or nobody pays. The market is small too, roughly 2,000 serious AU/NZ targets, and most of the largest run Magento rather than Shopify.","created_by":"profile-serge-ivo","status":"active","pro_candidate":0,"created_at":"2026-07-29 09:25:06","updated_at":"2026-08-05 06:09:30","preview":"Suppliers omit half the fields. Some are derivable by rule, some need an experienced eye on the photo. Researched hard across five passes: the capability is commodity, the Cin7-to-Shopify gap is real, Australia's schema is GTIN-anchored rather than taxonomy-anchored, and demand is the open question. Detail in Annexes A–D.","signal":"Cin7 Core's Shopify integration is 17 fixed fields with no attribute row at all, and metafield returns zero results across its entire help centre, while the same product carries 10 attributes to WooCommerce and BigCommerce. An unbuilt integration, not an architectural limit. NZ has mandated product-information fields in law since Dec 2023 with no mandated machine-readable format. And API/PIM integration is the top-tier upsell at two independent spec-library vendors — bimstore at GBP 4,800/yr, BIMstreamer at EUR 16,000/yr.","body_md":"","body_key":"ideas/gapfill/body.md","source_url":"","visibility":"public","parent_id":"","body_words":16979,"chapter_count":13,"support":0,"trash":0,"pivot":0,"contribution_count":45},"body":"Supplier product data arrives as a mess and has to leave as a catalogue. This is the research record for an idea about closing that gap for building products — tiles and lighting first — and it is mostly a record of the idea getting harder, not easier. Three proposed moats were researched and all three collapsed. The claim the page leaned on hardest was refuted by its own evidence. What survives is narrower, better evidenced, and still gated on one question nobody has answered.\n\n## Where this came from, and what it actually is\n\nThe origin is first-hand observation from a digital marketing strategy conversation, not desk research. A business owner supplying Italian tiles to architects has to specify 14–20 meta fields per product. Dimensions arrive in metres from one supplier and millimetres from another, sometimes within the same country. Surface finish is described inconsistently — polished versus not, plus supplier-specific finish names. Colours are given as marketing names such as \"olive\" rather than a canonical \"green\". Each supplier ships its own free-text product description.\n\nThe immediate trigger: a tile store on Shopify needed roughly 200 new products created, and the entire path from supplier documents to Shopify listings was manual data entry.\n\nThat is the whole of the validated pain. No third-party sizing, pricing, or willingness-to-pay evidence existed at the point the idea was written down, and the rest of this document is the attempt to find some.\n\n### The scope correction that changed the product\n\nThe idea owner corrected an early framing, and the correction matters more than it first appears. The output is **not** Shopify-specific. The goal is normalise-once, publish-many — the same unified product record publishing to Shopify, Magento, and WordPress/WooCommerce. Shopify was the example that surfaced the pain, not the target.\n\nThis changes the shape of the thing. It is no longer an importer that fills a Shopify store; it is a canonical product record plus per-platform publishing adapters, where each adapter owns that platform's schema quirks — Shopify variants and metafields, Magento attribute sets and attribute-set inheritance, WooCommerce product attributes and custom fields. The unified format becomes the actual asset, the thing a supplier owns and re-publishes from, and the storefront becomes a render target rather than the system of record.\n\nTwo consequences follow immediately, and both are unwelcome.\n\nFirst, **the buyer may change.** Normalise-once/publish-many is most valuable to whoever runs multiple storefronts or is replatforming — an agency, a distributor with several brand sites, a supplier mid-migration — not necessarily the single-showroom owner originally described.\n\nSecond, **the competitive framing hardens.** Multi-channel publishing from a central record is the textbook definition of PIM plus syndication. That puts the idea head-on against Akeneo, Salsify, Sales Layer, Plytix, Productsup, Channable and Feedonomics rather than beside them. The wedge narrows to exactly one thing: the building-materials spec vocabulary and the supplier-format reconciliation that generic PIMs make you configure by hand.\n\n### Ingest is three source classes, so the record is a merge\n\nThe owner specified three ingest sources, and they are not variations of one pipeline. They carry different halves of the record.\n\n**Supplier website (crawl).** Structured HTML spec tables plus images. Cheapest, needs no supplier cooperation, and uniquely enables continuous monitoring — re-crawl to catch new products, discontinued lines, spec revisions. Costs: terms of service, brittleness, politeness limits.\n\n**Inventory / ERP** (Cin7 Core or Omni, Unleashed, DEAR, NetSuite). Clean and API-accessible, but carries *commercial* fields: SKU, barcode, cost, price, stock, supplier. Not the 14–20 architectural spec fields — nobody types slip rating or PEI class into an inventory system.\n\n**PDF / Excel** (price lists, spec sheets, brochures). Messiest, highest extraction cost, but where spec data actually lives: thickness, finish, slip rating, frost resistance, rectified edge, application.\n\nThe consequence is structural: the canonical record is a **merge, not an import**. Commercial truth from inventory, spec truth from documents and website, joined on SKU or supplier code — which will not always match cleanly. Merge policy becomes first-class: per-field source-of-truth rules, conflict resolution when the ERP and the PDF disagree, and provenance on every field.\n\n## The category is occupied at every price tier\n\nThe first competitor scan found the idea is not novel, and the market is served end to end.\n\n**Tier 1 — enterprise PIM with a building-materials vertical already built.** Pimberly runs dedicated \"Building Materials Manufacturers\" and \"PIM for Construction\" pages. Sales Layer runs \"PIM for the Building Materials industry\" and explicitly markets handling of building-materials attributes — dimensions, density, fire and water resistance, SKUs — plus DAM for technical layouts, 3D models and certifications. Bluestone PIM and Apimio publish construction content too. Pricing anchor: Pimberly starts around USD 30,000/year (Capterra). \"Building materials\" is not an unserved vertical; it is a named, marketed segment.\n\n**Tier 2 — AI extraction and supplier onboarding.** SKU Launch is close to the idea as written: it pulls structured values from PDF spec sheets including tables, bullet lists and multi-column layouts mapped to your attribute schema; reads specs off product images; runs web-search agents to find authoritative product data; and replaces spreadsheet collection with a portal where suppliers confirm pre-filled data, claiming completion rises from about 40% to over 80%. Also here: dataX.ai, Klizer, Hypotenuse AI, getcatalog.ai.\n\n**Tier 3 — SMB Shopify apps doing the exact described workflow.** Shopif-AI takes a supplier PDF catalogue up to 300 pages and extracts names, prices, variants, images and specs, then publishes to Shopify. Importier imports CSV/Excel/PDF, auto-detects headers and auto-maps 40+ Shopify fields. Apport does AI bulk PDF/CSV import. Automated Commerce imports from supplier CSVs, PDFs and ERPs and generates metafield values in bulk.\n\nThe \"200 products by hand into Shopify\" pain is real enough that a competitive market already serves it from free-ish apps to USD 30k/year platforms.\n\n### A direct vertical competitor, and vertical ERP underneath it\n\nA second, deeper pass found what the first missed. **MindHarbor** serves \"Tile, Flooring & Building Materials Distribution\" specifically, and states it normalises vendor price lists and product catalogues, with supplier price and product list management, multi-UOM handling and variant management, delivered as ERP integration. That is this idea's core value proposition, already named, already sold into this exact vertical. It appears positioned as an ERP integration services firm rather than a self-serve product — a meaningful difference in shape, but not in claim.\n\nThe surrounding tier is crowded with vertical ERP. Epicor runs a dedicated Tile Distribution solution. Klipboard ERP covers flooring and surfaces including supplier catalogue integration and multiple price lists. Comp-U-Floor automatically exchanges price catalogues, purchase orders and vendor invoices between manufacturers, distributors and flooring retailers. Acctivate does tile distribution on QuickBooks. Catalogue and price-list ingestion is table stakes in flooring ERP.\n\n### Named building-products customers, with published outcomes\n\nThe strongest refutation of any \"absent vertical\" claim is not a logo wall but case studies with numbers.\n\n- **Syndigo → UFP Industries** (lumber and building products): 60% faster launches, 90%+ content accuracy, 60% sales increase. Also Kohler, Klein Tools.\n- **Sales Layer → Frans Bonhomme** (40+ years in building materials, supplies ACCIONA) and **buildingmaterials.co.uk** (500 suppliers, 537 categories, 50,000 SKUs across drylining, timber, insulation, SFS steel framing).\n- **Pimberly → Timco** (fixings), **Yesss Electrical**, **Highbourne Group** (plumbing and heating), **IronmongeryDirect**. Its logo wall includes **Headlam**, the UK's largest floorcoverings distributor, and **Marshalls** (paving, hard landscaping, stone).\n- **Bluestone PIM → Neumann** (major Norwegian B2B construction distributor), **Saint-Gobain Distribution Norway**.\n- **inriver** runs a page titled \"The PIM for tools and construction\"; case study WernerCo (ladders), plus Kohler, Prysmian, Atlas Copco, Lincoln Electric.\n- **Salsify** ships a Home Improvement Growth Kit, a dedicated Home Depot connector, and a \"State of Product Syndication for Tools and Hardware\". Named: Moen, UMA Home Décor.\n- **Feedonomics → Stone Coat Countertops.** **Klizer (DCKAP)** does AI extraction from supplier PDFs for B2B distributors, core market plumbing/HVAC/industrial.\n\nHeadlam matters most: Pimberly simultaneously sells to the UK's largest flooring distributor *and* ships ImageAI. That is the closest anyone has come to assembling the product proposed here.\n\nDocumented as empty rather than assumed: Akeneo has no building-materials case study; Constructor.io names about 30 logos, none in building, with no building or DIY mention; nothing found for Bunnings, Reece, Beaumont Tiles, Tile Cloud, Ferguson, Build.com or Floor & Decor; Vue.ai and Mad Street Den are fashion-only.\n\n### The integration reality: what Cin7 actually carries\n\nAn early competitive worry was that Cin7 already publishes to Shopify, WooCommerce and Magento, so for a Cin7 customer publish-many already exists and is paid for — removing publishing as a reason to buy. The honest positioning was therefore a *spec layer enriching an existing sync*, not replacing it. Researching the field-level detail confirmed the gap is real and sharper than assumed.\n\n**Cin7 Core to Shopify is 17 fixed fields:** SKU, Name, Category, Description, Weight, Barcode, Price Tier, Images, Tags (first 256 characters only), Option 1/2/3, Variation, Brand, HS code, Country of origin.\n\nWhat it does not carry:\n\n- No Additional Attributes row at all.\n- **No metafields.** \"metafield\" returns zero results across the entire Cin7 Core help centre, verified via the Zendesk search API.\n- No dimensions — present in the WooCommerce and BigCommerce mappings, absent for Shopify.\n- **No unit of measure** — fatal for tile, which is stocked by box and sold by square metre.\n- No product attachments: no spec sheets, no Declarations of Performance, no slip certificates.\n\nA correction to an earlier entry: the manual-sync limitation is **not** WooCommerce-specific. Cin7 Core's Shopify documentation says \"Product information must be manually updated to be exported to Shopify, it will never be updated automatically.\" It had been recorded as a WooCommerce quirk; it applies to Shopify too.\n\nThe channel asymmetry is the crux. Cin7 Core **does** carry its 10 additional attributes to WooCommerce and BigCommerce, but not to Shopify and not to Magento. That is not an architectural limit — it is an unbuilt integration. And Cin7 knows the data matters: Core's own B2B Portal has a \"Show additional attributes\" setting, so Cin7 renders spec attributes on its own storefront while having no pipe to push them to Shopify.\n\nCore's attributes are too small to be a source of truth in any case: a maximum of 10 per set, one set per product, text capped at 256 characters, and plan-gated — \"Ignored when current Cin7 Core Subscription plan is Small or Medium\". A tile spec sheet needs 20–40 fields.\n\nUnleashed is narrower still: 9 fields, with Tags and Vendor explicitly not synced, attribute *values* capped at 50 characters — unusable for spec prose — and attribute sets appearing nowhere in its Shopify mapping.\n\nThere is also a connector tier that the first pass missed entirely. Stacksync markets a Shopify-and-Cin7 integration stating both standard and custom objects can be synchronised, and Make.com offers Cin7-to-WooCommerce automation. Beyond Cin7's native connectors there is a generic iPaaS layer able to move custom fields between systems with no purpose-built product, which weakens the multi-platform-publishing claim further.\n\n## Competitors and integration — the field-level evidence\n\n### The refutation that killed the original moat\n\nThis document's most load-bearing claim — *\"vision-enrichment vendors do not serve building materials; they target fashion, electronics and automotive\"* — was **wrong in its strong form**. Two statements withdrawn:\n\n1. \"Building materials are absent from these vendors' industries\" — **false**.\n2. \"Its application to tile and lighting vocabulary is not evidently occupied\" — too strong; **partly occupied**.\n\nWorse, the confidence register had called this an *unverifiable* absence. **It was verifiable, and verifying it broke it.** It had been an **inference from an absence of evidence** — vendor pages listing fashion, electronics and automotive — not a positive finding.\n\n**And the counter-example was already half-visible in this page's own source index.** [SKU Launch](https://skulaunch.com/platform/product-data-extraction) is a product of **Start with Data, a Melbourne company** (+61 3 8376 1218). Its industry list names **Building Supplies** and **DIY & Tools**; it does AI document parsing of supplier PDFs and spec sheets. Named case studies: **Huws Gray** (\"product data enrichment for a leading builders merchant\"), **MKM Building Supplies**, **Castorama**, **Kingfisher Group**. The page had **twice** cited startwithdata.co.uk — including an article titled *\"the 5 best PIMs for builders merchants\"* — without connecting it.\n\n**The \"unexplored Australian home turf\" is already occupied by an Australian vendor selling to builders merchants.**\n\n### PIM / syndication vendors with named building-products customers\n\n- **Syndigo** at UFP Industries\n- **[Sales Layer](https://www.saleslayer.com/pim-for-building-materials)** at BuildingMaterials.co.uk\n- **Bluestone PIM** at Saint-Gobain Distribution Norway\n- **[Pimberly](https://pimberly.com/building-materials-manufacturers/)** — **the acute threat.** It ships **ImageAI** (\"identifying materials, recognizing shapes or patterns\") *plus* **ColorAI**, and its customers include **Headlam**, the UK's largest floorcoverings distributor. **Both halves of the thesis, uncombined.** If Pimberly joins them, the corpus lead is worthless.\n- Also **inriver**, **Salsify**\n\n### Image-sourced attribute extraction already shipped\n\n- **[Hypotenuse](https://www.hypotenuse.ai/features/ecommerce-product-data-enrichment)**\n- **[Velou](https://www.velou.com/post/product-attribute-enrichment-powering-discoverability-and-personalization-at-scale)**\n- **[Envive](https://www.envive.ai/post/ai-product-feed-enrichment-attributes-taxonomy)**\n\n### The missed axis: visualisers, and they are the more dangerous set\n\n- **Renoworks** — 200+ building-product manufacturer customers\n- **Floori** — extracts **UV and PBR maps from ordinary photos of tile samples**\n- **Roomvo / Leap Tools** — 250+ brands, 7,000+ dealers\n- **Consumer tile-ID apps** already classify finish, look, pattern and size from a single photo — precisely the task this document called its moat\n\n**Warning sign:** Lily AI ran a home vertical doing material and finish tagging in 2023. **That page now 404s and the company lists only apparel and beauty.** Someone tried this and retreated.\n\n### The three-moats verdict\n\n- **Moat 1 — building-materials taxonomy: GONE.** ETIM is an open standard; ISO 13006 already standardises ceramic tile classification. A free public good.\n- **Moat 2 — AI extraction from messy supplier formats: GONE.** Commodity, shipping in cheap Shopify apps (**Shopif-AI, Importier, Apport**) and mid-market **SKU Launch**. LLM improvement makes it cheaper every quarter, compressing margin.\n- **Moat 3 — multi-platform publishing: GONE.** Cin7 Core documents Shopify, Magento and WooCommerce integrations; Cin7 Omni exposes editable field mappings; PIMs treat syndication as table stakes.\n\n**The squeeze:** this sits in the valley between a cheap single-platform Shopify importer and a **~USD 30,000/yr enterprise PIM** (Pimberly, Capterra, medium confidence). The buyer there — a showroom with a few thousand SKUs across twenty suppliers and two storefronts — is **real but small, hard to reach, low willingness to pay.** Both sides are expanding into it.\n\n**Honest read: as venture-scale software, weak.** The differentiated work remaining is integration and mapping labour — **a services shape, not a licence shape.**\n\n**But one caveat in that verdict proved decisive:** a Cin7 source noted that WooCommerce syncs stock automatically while descriptions and other product details need manual sync. *\"If shallow field coverage is typical, the enrichment gap is wider than this verdict assumes. Verify before killing.\"* It was verified. It is wider.\n\n### The surviving positioning\n\nReplace the \"vision enrichment is shipped — but not here\" framing entirely; it is refuted and cannot be softened into survival.\n\n**Honest restatement:** the vertical is **well served on the PIM/syndication axis** (Syndigo, Sales Layer, Pimberly, Bluestone, inriver, Salsify, SKU Launch — all with named building-products customers) **and well served on the visualiser axis** (Renoworks, Floori, Roomvo, tile-ID apps). **Nobody has joined them.**\n\nAcross every vendor taxonomy checked, **slip rating, PEI class, water absorption, frost resistance and rectified edge returned zero hits.**\n\nSo the thesis is no longer \"we infer where others cannot\" but **\"we infer into the compliance-grade schema the specification channel needs, which merchandising-focused vendors have no reason to build.\"** Consequences: **the visualiser companies are the more dangerous competitor set, not the PIM vendors**, and **Pimberly is the acute threat.**\n\n### Adjacent findings\n\n- **Vertical ERP / direct competitors** exist in tile and flooring — a services-and-software tier already selling to this exact buyer.\n- **iPaaS / connector layer** (previously missed) competes on the plumbing.\n- **Cin7's own integration directory: 700+ integrations, no PIM category, no PIM vendor.** Nobody occupies the space.\n- **Matrixify correction** — the one non-desk-research finding, from the owner's direct product knowledge.\n- **Supplier-confirmation pattern:** SKU Launch replaces spreadsheet collection with a portal where suppliers **confirm pre-filled data rather than typing** (claims ~40% → 80%+ completion, vendor claim). **Same confirm-not-type principle this document reached independently — evidence the pattern works, and that a competitor already applies it supplier-side.**\n- **Akeneo CE** feature freeze and **EOL September 2026**; running cost **EUR 1,500–3,000/month** (aggregator, medium confidence).\n\n### Cin7 Core → Shopify: the field-level evidence\n\n**17 fixed fields**, enumerated from vendor docs: **SKU, Name, Category, Description, Weight, Barcode, Price Tier, Images, Tags (first 256 chars only), Option 1/2/3, Variation, Brand, HS code, Country of origin.**\n\nNot carried:\n\n- **No Additional Attributes row at all**\n- **No metafields** — \"metafield\" returns **zero results across the entire Cin7 Core help centre**, verified via the Zendesk search API (count: 0)\n- **No dimensions** — present in the Woo and BigCommerce mappings, absent for Shopify\n- **No unit of measure** — fatal for tile, stocked by box and sold by m²\n- **No product attachments** — no spec sheets, no DoPs, no slip certificates\n\n**The channel asymmetry is the crux:** Core **does** carry its 10 additional attributes to **WooCommerce and BigCommerce**, but **not to Shopify and not to Magento**. **An unbuilt integration, not an architectural limit.**\n\n**Cin7 knows the data matters:** Core's own **B2B Portal has a \"Show additional attributes\" setting** — it renders spec attributes on its own storefront with no pipe to push them to Shopify.\n\n**Correction to an earlier version of this document:** the manual-sync limitation is **not WooCommerce-specific**. Cin7 Core's Shopify doc: *\"Product information must be manually updated to be exported to Shopify, it will never be updated automatically.\"*\n\n**Core cannot be source of truth:** **max 10 attributes per set, ONE set per product, text capped at 256 chars, plan-gated** — *\"Ignored when current Cin7 Core Subscription plan is Small or Medium.\"* **A tile spec sheet needs 20–40 fields.**\n\n**Unleashed is narrower still:** **9 fields**; Tags and Vendor explicitly not synced; **attribute values capped at 50 characters** (unusable for spec prose); attribute sets appear nowhere in its Shopify mapping.\n\n**Cin7 Omni** exposes an *Update Recently Modified Products Field Mapping* page mapping Omni fields to Shopify, with Omni data replacing Shopify data on upload. It has **unbounded custom fields to 2048 characters** and user-editable mappings — but **metafields appear only as a blocking error.** For WooCommerce, only stock availability syncs automatically.\n\n**The bigger-than-a-connector conclusion:** because Core caps attributes so hard, this cannot be \"read Cin7, write Shopify metafields\". **The enrichment layer must own its own datastore and use Cin7 only for SKU, stock and price identity — materially more product than a sync tool, which raises the bar on how strong the demand signal must be before building.** And it has to be a global product with an AU beachhead.\n\n### Shopify platform mechanics\n\n- **Metaobject limits: 128 metaobject definitions per app, 30 fields per definition.**\n- **Bulk Operations API**: async GraphQL returning **JSONL**, tens of thousands of records without hitting standard rate limits; **one bulk operation per type at a time per shop**; mutations can **partially fail**, so retry logic is the builder's.\n- **Embedded vs standalone conversion**: installed-to-first-use **70–85% embedded vs 45–60% standalone** (aggregator, medium confidence).\n- **Shopify's own guidance names four cases where standalone beats embedded, and all four apply**: UI needs more room than admin-chrome width; used by non-merchant roles (suppliers, agency staff, architects); spans multiple storefronts; is a full tool where Shopify is one integration among many.\n- **Shopify B2B gap:** Shopify **lacks native functionality for sales reps to take orders at trade shows or in the field** — documented against purpose-built B2B wholesale platforms. The storefront may not be the centre of this customer's world.\n- **Unresolved:** off-platform billing for connector apps where the vendor is a standalone SaaS acquiring **zero merchants through Shopify** — whether an exemption applies, or split billing. Flagged for whoever builds.\n\n### Ingest architecture: a merge, not an import\n\nThree source classes:\n\n1. **Supplier website crawl** — structured HTML spec tables plus images. Cheapest, needs no supplier cooperation, and **uniquely enables continuous monitoring** (re-crawl for new products, discontinued lines, spec revisions). Costs: ToS exposure, brittleness, politeness limits.\n2. **Inventory / ERP** — Cin7 Core or Omni, Unleashed, DEAR, NetSuite. Clean and API-accessible, but carries **commercial fields only** (SKU, barcode, cost, price, stock, supplier). **Nobody types slip rating or PEI class into an inventory system.**\n3. **PDF / Excel** — price lists, spec sheets, brochures. Messiest and dearest to extract, **but where spec data actually lives** (thickness, finish, slip rating, frost resistance, rectified edge, application).\n\n**The canonical record is a merge:** per-field source-of-truth rules, conflict resolution when Cin7 and the PDF disagree, and provenance on every field.\n\n### PDF extraction: an architectural fork\n\n- **LlamaParse 8.54 seconds; Marker 47 seconds; Docling 1 minute 50 seconds** (same test).\n- Why: **Docling** runs one large vision-language model reading each page as an image, generating table structure token by token — inherently slow. **Marker** runs a **5-stage pipeline** of smaller specialised models doing mostly classification and detection, avoiding token-by-token generation.\n- **Docling separately reported at 97.9% accuracy** on complex table extraction. **Unstructured** has the strongest OCR.\n- **The constraint that decides it: LlamaParse is cloud-hosted on LlamaCloud, so documents leave your infrastructure.** Supplier price lists carry confidential trade pricing. **That disqualifies the fastest option** and pushes the build to **Marker or Docling running locally**. An architectural fork, not a preference.\n\n### Open-source stack — build cost is assembly\n\n- **`wbsg-uni-mannheim/ExtractGPT`** — LLM attribute value extraction\n- **`wbsg-uni-mannheim/wdc-pave`** — extraction **and normalization**, the exact two-step needed\n- **`google/langextract`** — structured extraction with **precise source grounding**, matching the provenance requirement\n- Structured output: **Instructor, Outlines, PydanticAI, Guidance**\n- **Splink** — probabilistic record linkage, **a million records on a laptop in about a minute**, 100M+ on Spark. Also **Zingg**, **dedupe**\n- Papers: `arxiv.org/pdf/2501.01237` (Self-Refinement for LLM-based Product Attribute Value Extraction), `arxiv.org/html/2310.12537v4` (ExtractGPT)\n\n**Implication: build cost is assembly, not invention — cheap for us, equally cheap for anyone else. The corpus, not the pipeline, is the only asset.**\n\n### Form factor: the review surface IS the product\n\nIt requires a **durable record where every field carries extracted / derived / inferred / human-asserted status**, corrections persist, and corrections accumulate into a labelled corpus. It is **simultaneously the safety mechanism for the never-infer compliance rule and the collection mechanism for the corpus.** Every human correction is a **labelled example linking image + partial fields to a correct value.**\n\n**Verdict: standalone web app at the core, thin connectors at the edges. Not a Shopify app, not an Excel plug-in.**\n\n- **Office.js add-ins** are web apps in a sandboxed iframe — deployable across Office on web, Windows, Mac and iPad, able to read/write worksheets and add task panes and custom functions — **but strictly sandboxed and unable to access the local drive**, so bulk ingestion of local supplier files from inside an add-in is constrained.\n- A spreadsheet **cannot carry per-field confidence, provenance or extracted/derived/inferred status**, cannot support multi-user review, and does not persist corrections. **Excel belongs as an import/export surface only.**\n- A Shopify app should still exist — as a **distribution and publish channel, a thin client onto the standalone record, not the product.**\n- Review queue UI: image, source snippet, confidence and canonical value, side by side.\n\n### The demand hole\n\n**The weakest part of the whole dossier.** Searched specifically: **no forum threads, no reviews, no vendor complaints** documenting anyone suffering from Cin7's inability to carry rich spec data to Shopify. **The only signal is agency marketing copy.**\n\n## The standards layer, measured rather than assumed\n\nThe page originally claimed the defensible asset was \"a maintained, opinionated taxonomy of tile and lighting spec fields with the synonym maps that make it work\". That taxonomy exists, is international, and is free.\n\n**ETIM** (Electrotechnical Information Model) is the open international classification standard for technical products. It defines classes, codes and standardised technical attributes precisely so manufacturers, distributors and retailers can exchange product data unambiguously. Its stated scope is electrical, HVAC, plumbing **and building materials**. Lighting sits at its historical core.\n\n**BMEcat** is the paired XML exchange format: manufacturers structure catalogues as BMEcat with ETIM classification embedded, giving a standard machine-readable route from supplier to distributor. ETIM is increasingly the feature-data backbone for BIM objects, linking specify–procure–install.\n\nFor ceramic tiles, **ISO 13006:2018** already defines classification and characteristics — water absorption groups, extruded versus dry-pressed, dimensional tolerances, surface quality — plus canonical terminology for glaze, engobed and rectified. That is a substantial share of the \"14–20 fields\" already standardised.\n\nThe conclusion was: do not build a proprietary taxonomy. Adopt ETIM plus ISO 13006 and compete on mapping *into* them.\n\n### ETIM tile coverage, enumerated\n\nThat conclusion was then tested properly, and the answer is **partial** — better than assumed on fields, worse on the one field that matters most.\n\nThe access path is solved. The ETIM API returns 401 without credentials, but **bSDD serves all of ETIM 10.1 free under ODC-By with no authentication** — 5,640 classes, 17,377 features, public swagger spec. The class pages at viewer.etim-international.com prerender and are enumerable via a 5,680-URL sitemap. The taxonomy is machine-readable today at zero cost and zero paperwork.\n\nThe classes that exist:\n\n- **EC011340 Ceramic tile** — 28 features, group EG020056 (Masonry and tile material), published since ETIM 8.0. Across all ~200 bSDD dictionaries this is the *only* real ceramic tile class; \"porcelain stoneware\" returns zero.\n- EC011341 Mosaic tile (21 features), EC010182 Natural stone tile (16), EC012349 Concrete tile (30), EC011708 Tile profile (11).\n- Flooring group EG020068 holds 13 classes: EC004040 Parquet (18), EC004041 Laminate (15), EC004085 Solid wood (12), EC004931 PVC flooring (14, added only December 2024), EC012273 Screed (17), EC012271 Rubber (4), EC012272 Plastic (3).\n\nThe field mappings are real and usable, which corrects the earlier pessimism. Thickness EF020850; nominal dimensions EF020848/49 plus production dimensions EF020855/56; finish EF020841/46, EF024897, EF020838; **slip EF020842** (DIN 51097 A/B/C) and **EF020843** (R9–R13); **PEI-equivalent EF020844**; frost EF020845 (boolean only); rectified EF020839/40, EF022394; colour EF020851.\n\nLighting is roughly four times deeper, as predicted: EC001744 downlight/spot/floodlight carries 118 features, EC002892 ceiling and wall luminaire 115, EC000062 street 85, EC001957 emergency 68.\n\nOne telling detail: ETIM models the tiling **consumables** better than the tile. Tile adhesive EC011707 has 42 features and tile grout EC011709 has 28, against 28 for the ceramic tile itself. The standard grew out of the wholesale channel and it shows. Soft flooring is a genuine taxonomy gap — no carpet, textile, linoleum or LVT class exists at all.\n\n### The derivation rule breaks inside the standard\n\nThe page's worked example of rule-based gap-filling was: ISO 13006 ties water absorption to group, porcelain is definitionally below 0.5%, therefore porcelain-versus-ceramic is *derived*, never asked for.\n\n**EC011340 Ceramic tile has no water absorption feature.** That derivation cannot be expressed in the standard's own tile class.\n\nIt is not a one-off. Across all 17,377 ETIM features there are zero occurrences of ISO 13006 / EN 14411 group, DCOF / ANSI A326.3, modulus of rupture, breaking strength, crazing resistance, stain resistance, or thermal shock. Frost resistance exists only as a boolean, not the tested class. It is a *modelling* gap rather than a vocabulary limit — EC012349 Concrete tile does carry water absorption — and nothing relevant sits in the dynamic pipeline, so no fix is imminent.\n\nNor can the hole be filled from another free standard:\n\n- **eCl@ss 16.0** has the better taxonomy — 22-47 Tile/panel is 76 classes split by body type, roughly matching ISO 13006 bands, and 22-46 Flooring is 39 classes including carpet, linoleum and cork. But Advanced 16.0 tile classes expose only 7 properties, all REACH/SVHC or commercial, with no technical aspect. ETIM wins on physics, eCl@ss on classes, neither on both.\n- **ISO 13006:2018** is the real tile data model — A/B forming by Ev groups Ia–III, porcelain at Ev ≤ 0.5%, Annex N carrying the PEI scale, ISO 10545 test methods — but it is paid, prose, and not machine-readable.\n- **EN 14411:2012** remains the CE and Declaration of Performance basis. CPR (EU) 2024/3110 will not displace it for years, proceeding family by family with standardisation requests in 2027/2028. No tile digital product passport model exists or is commissioned.\n\nSo \"adopt ETIM, don't rebuild it\" was too glib. The honest position: adopt ETIM as the spine, then extend it with water absorption, ISO 13006 group, breaking strength and DCOF, which no free standard models for tiles. That extension is a defensible asset *and* unavoidable work.\n\n### The standard exists; the data does not\n\nThis was established by enumeration rather than by failing to find things, which makes it unusually solid.\n\nAll **1,383 registered data suppliers in the Dutch 2BA datapool** were enumerated. Absent: Tarkett, Gerflor, Interface, Desso, Unilin/Quick-Step, Kährs, Bolon, Rockfon, Armstrong, Uzin, Bostik, Forbo/Eurocol, Mapei, Ardex, Omnicol, Knauf, Saint-Gobain, Schlüter. Among the top 25 data suppliers in the \"Bouw\" sector, **not one is a tile or flooring manufacturer** — they are bathroom, electrical, HVAC, fastener and tool companies.\n\nForbo Flooring Systems: all 96 child sitemaps enumerated, and exactly one URL in the entire Forbo estate matches `/etim/` — the Eurocol *adhesives* page. Uzin/Uzin Utz: zero ETIM URLs across full sitemap crawls. Germany's 2022 building-materials launch plan recorded \"Fliesen und Natursteine\" as the one assortment group with no industry participants at all. Mosa's only traceable ETIM data was applied downstream by a wholesaler rather than published by Mosa. Villeroy & Boch's 2BA presence is its bathroom division, not tiles. Marazzi is nowhere.\n\nAnd yet flooring is officially in scope, correcting an earlier \"could not establish\". ETIM UK states its Building Materials sector includes \"Walling, Flooring & Roofing Materials, Structural steel, Civil infrastructure, Doors & windows, Finishing & Decoration Products\". The Builders Merchants Federation has been Sector Lead since 1 January 2020 — 1,020 members, GBP 51.6bn combined sales. ETIM Netherlands created a \"Tegels\" (tiles) commission after mid-2020. Eleven of eighteen national chapters run building materials.\n\n**The problem is participation, not permission.**\n\n### Adoption cuts both ways, and the unflattering read is at least as plausible\n\nThe flattering read: the normalisation work is genuinely undone. A free, machine-readable tile schema exists with 28 usable fields including R-value and a PEI equivalent, and not one tile brand has populated it. The gap between \"standard exists\" and \"data exists\" is exactly what this idea proposes to close.\n\nThe unflattering read: **adoption is absent because demand is absent.** ETIM has admitted flooring since January 2020. The Builders Merchants Federation — 1,020 companies, GBP 51.6bn of sales, precisely the channel that would consume tile spec data — has led the sector for five and a half years. ETIM NL stood up a dedicated commission. Germany ran a launch in which tiles and natural stone attracted zero industry participants.\n\nFive years of permission, an organised merchant constituency and a purpose-built commission produced no tile data. That is not a story about missing tooling. If merchants were losing money for want of structured tile attributes, a brand would have published or a merchant would have paid someone to do it. The more likely explanation: tile selection is driven by appearance, price, availability and range, while compliance fields are supplied once as a PDF datasheet or Declaration of Performance at specification time — not needed per-SKU in a catalogue.\n\nIf that is right, it inverts the central split described in the next chapter. The compliance-grade schema may have no commercial buyer at all.\n\nThe cheapest disconfirming test is not a search: **ask a tile merchant and an architect which fields they query per-SKU versus accept as a PDF.** If the answer is \"appearance and availability, plus a datasheet on request\", the standards work is a cost, not a moat.\n\n## Standards layer — the full ETIM measurements\n\n### Access and scale\n\n- [ETIM is free under the Open Data Commons Attribution Licence](https://www.etim-international.com/classification/license-info/) (ODC-By), downloadable in full.\n- Its [own API](https://etimapi.etim-international.com/) is Swagger-documented but **returns 401 without credentials** (OAuth2 client_id/secret, Release and Details services with filtering and paging).\n- **bSDD serves all of ETIM 10.1 free, with no auth: 5,640 classes, 17,377 features.** Swagger spec public. `viewer.etim-international.com` class pages prerender and are enumerable via `/sitemap.txt` — **5,680 URLs**.\n- Conclusion: **machine-readable today at zero cost and zero paperwork.** No licence negotiation is needed to adopt ETIM as a spine.\n\n### The tile classes, with feature counts\n\n| Class | Name | Features |\n|---|---|---|\n| **EC011340** | **Ceramic tile** | **28** |\n| EC012349 | Concrete tile | 30 |\n| EC011341 | Mosaic tile | 21 |\n| EC010182 | Natural stone tile | 16 |\n| EC011708 | Tile profile | 11 |\n\nEC011340 sits in group **EG020056** (Masonry and tile material), published since **ETIM 8.0, v5**. Across all ~200 bSDD dictionaries it is the **only** real ceramic tile class — **\"porcelain stoneware\" returns zero**.\n\n**Flooring group EG020068, 13 classes:** Parquet EC004040 18; Laminate EC004041 15; Solid wood EC004085 12; PVC flooring EC004931 14 (added only **December 2024**); Screed EC012273 17; Rubber EC012271 4; Plastic EC012272 3.\n\n**Soft flooring is a genuine taxonomy gap: no carpet, textile, linoleum or LVT class exists at all.**\n\n### Feature codes that DO exist on EC011340\n\n- Thickness **EF020850**\n- Dimensions, nominal **EF020848 / EF020849**; production **EF020855 / EF020856**\n- Finish **EF020841 / EF020846**, **EF024897**, **EF020838**\n- **Slip modelled twice** — **EF020842** (DIN 51097 A/B/C) and **EF020843** (R9–R13)\n- **PEI equivalent EF020844**\n- Frost **EF020845** — **boolean only**, not the tested class\n- Rectified **EF020839 / EF020840**, **EF022394**\n- Colour **EF020851**\n\nThat is a usable spec spine: slip in both common notations, a wear class, thickness, rectified edge, finish and colour.\n\n### Properties that are ABSENT\n\n- **Water absorption is absent from EC011340** — which breaks this document's showcase derivation, since ISO 13006 ties absorption to group and porcelain is definitionally <0.5%.\n- **Across all 17,377 ETIM features there are ZERO occurrences of:** ISO 13006 / EN 14411 group, **DCOF / ANSI A326.3**, modulus of rupture, breaking strength, crazing resistance, stain resistance, thermal shock.\n- It is a **modelling gap, not a vocabulary limit** — **EC012349 Concrete tile does carry water absorption.**\n- **Nothing relevant sits in the DYNAMIC pipeline**, so there is no imminent fix.\n\n**Honest position:** adopt ETIM as the spine, then **extend** it with water absorption, ISO 13006 group, breaking strength and DCOF. That is a defensible asset *and* unavoidable work.\n\n### The revealing detail: consumables beat the tile\n\nTile adhesive **EC011707 = 42 features**; tile grout **EC011709 = 28**; ceramic tile EC011340 = 28. Lighting is roughly four times deeper: downlight/spot/floodlight **EC001744 = 118**; ceiling/wall luminaire **EC002892 = 115**; street **EC000062 = 85**; emergency **EC001957 = 68**.\n\n**The standard grew out of the electrical and plumbing wholesale channel, and it shows.** ETIM models what merchants stock and reorder, not what architects specify.\n\n### eCl@ss: better classes, no physics\n\n- **eCl@ss 16.0**: **22-47 Tile/panel = 76 classes**, split by body type and roughly matching ISO 13006 bands. **22-46 Flooring = 39 classes**, including carpet, linoleum and cork — the exact classes ETIM lacks.\n- **But ADVANCED 16.0 tile classes expose only 7 properties, all REACH/SVHC/commercial. No technical aspect at all.**\n\n**Verdict: ETIM wins on physics, eCl@ss on classes, neither on both.**\n\n### ISO 13006 / EN 14411 / CPR\n\n- **ISO 13006:2018 is the real tile data model** — A/B forming, **Ev groups Ia–III**, **Annex N PEI scale**, **ISO 10545 test methods** — but **paid, prose, and not machine-readable**. ([2018](https://www.iso.org/standard/63406.html), [2012](https://www.iso.org/standard/51063.html))\n- Water-absorption bands: **BIa <0.5%, BIb 0.5–3%, BIIa 3–6%, BIIb 6–10%, BIII >10%.** Porcelain is definitionally <0.5%, i.e. impervious.\n- It also defines extruded vs dry-pressed, dimensional tolerances, surface quality, and the canonical terminology for **glaze, engobed, rectified**.\n- **EN 14411:2012** remains the **CE / DoP basis**. **CPR (EU) 2024/3110** will not displace it for years — family by family, with standardisation requests in **2027/2028**.\n- **No tile Digital Product Passport model exists or is commissioned.**\n\n**PEI caution.** PEI grades the wear of a **glazed** surface; unglazed and through-body porcelains receive no PEI rating, so an *absent* PEI is evidence of an unglazed or through-body product. Do not over-derive: sources treat water absorption, PEI and COF/slip as distinct complementary attributes, and **slip rating is a measured test result that is not reliably derivable from finish or appearance.**\n\n### Adoption: an enumerated absence, not a failed search\n\nThis is the most important finding in this section, because it was established by enumeration rather than by failing to find something.\n\n- **All 1,383 registered data suppliers in the Dutch 2BA datapool were enumerated.** Absent: **Tarkett, Gerflor, Interface, Desso, Unilin/Quick-Step, Kährs, Bolon, Rockfon, Armstrong, Uzin, Bostik, Forbo/Eurocol, Mapei, Ardex, Omnicol, Knauf, Saint-Gobain, Schlüter.**\n- **Among the top 25 \"Bouw\"-sector data suppliers, not one is a tile or flooring manufacturer** — they are bathroom, electrical, HVAC, fastener and tool companies.\n- **Forbo Flooring Systems: all 96 child sitemaps enumerated. Exactly ONE URL in the entire Forbo estate matches `/etim/` — the Eurocol adhesives page.**\n- **Uzin / Uzin Utz: zero ETIM URLs** across full sitemap crawls.\n- **Mosa's** only traceable ETIM data was applied downstream **by a wholesaler**, not published by Mosa. **Villeroy & Boch's** 2BA presence is its **bathroom** division. **Marazzi is nowhere.**\n- **Germany's 2022 building-materials launch plan: \"Fliesen und Natursteine\" was the one assortment group with no industry participants at all.**\n\n#### And yet permission has existed for years\n\n- **Flooring is officially in scope** (this corrects an earlier \"could not establish\"): ETIM UK's Building Materials Sector includes \"Walling, **flooring & roofing materials**, Structural steel, Civil infrastructure, Doors & windows, **finishing & decoration products**\".\n- **The Builders Merchants Federation has been Sector Lead since 1 January 2020 — 1,020 members, GBP 51.6bn combined sales.**\n- **ETIM Netherlands created a \"Tegels\" (tiles) commission after mid-2020.**\n- **11 of 18 national chapters run building materials**; Belgium does not.\n\n**The problem is participation, not permission.**\n\n### The double read — and why it matters more than the schema\n\n**Flattering:** the normalisation work is genuinely undone. A free, machine-readable tile schema exists with 28 usable fields including R-value and a PEI equivalent, and not one tile brand has populated it.\n\n**Unflattering, and at least as plausible:** adoption is absent **because demand is absent.** Five and a half years of permission, an organised merchant constituency of 1,020 companies, a purpose-built tiles commission, and a national launch that attracted zero tile participants. That is not a story about missing tooling.\n\nMore likely: **tile selection is driven by appearance, price, availability and range**, while compliance fields are supplied **once, as a PDF datasheet or DoP, at specification time** — not needed per-SKU in a catalogue.\n\n**If that holds it inverts this document's central split**: the compliance-grade schema would have no commercial buyer.\n\n**Test before building:** ask a tile merchant and an architect **which fields they query per-SKU versus accept as a PDF.** If the answer is \"appearance and availability, plus a datasheet on request\", the standards work is a cost, not a moat.\n\n### Exchange format and cross-taxonomy mapping\n\n- **ETIM xChange 1.1** is the current JSON master-data exchange standard, replacing ETIM's earlier BMEcat recommendation; a **published xChange-to-BMEcat field mapping** exists. **BMEcat** remains the paired XML format, with ETIM classification embedded, and a published ETIM BMEcat guideline.\n- Sources note ETIM is increasingly the **feature-data backbone behind BIM product objects**, linking specify–procure–install.\n- **Licence trap on the mapping.** NBS publishes a native Uniclass REST API (`GET definitions/uniclass2015/{notation}/{depth}`, JSON/XML, registration-gated, quarterly downloads) — but **Uniclass is CC BY-ND**. An ETIM-class-to-Uniclass-notation crosswalk is a **derivative work**: internal use is probably fine, **publishing or selling the mapping likely is not.** Get legal advice.\n- **bSDD gives mapping capability but not the dataset** — it contains **no OmniClass and no MasterFormat at all**, so a mapping to the US taxonomies cannot be sourced from it. Cross-dictionary `ClassRelation` exists (`IsEqualTo`, `IsSimilarTo`, `HasReference`) and Molio's CCI shows full coverage is achievable, but Uniclass 2015 in bSDD carries **zero outbound relations** in sampling, sits at Preview status, `isVerified: false`, and was last touched August 2024.\n\n### Unverified — do not rely on\n\nA claimed list of tile-chemistry ETIM adopters (**Mapei, Ardex, Weber, Sopro, PCI**) **conflicts directly with the 2BA enumeration above.** Treat as unverified.\n\n### ETIM AU/NZ\n\nAnnounced 26 November 2021 by the Australian Industry Group, electrotechnical-only, and `etim-aunz.com` is still a parked NameBright page nearly five years later. Covered in full in the AU/NZ section below.\n\n## Filling the gaps: derivation, inference, and the line that must not be crossed\n\nSuppliers omit fields. Missing values get filled either by deduction from other fields or by an experienced person looking at the product. These are three mechanisms with different trust levels, and they must never be blurred.\n\n**Derived — a rule from a standard.** ISO 13006 maps water absorption to group to porcelain/ceramic; an absent PEI implies an unglazed or through-body product. Deterministic and auditable. (Noting the previous chapter's finding: ETIM's own tile class cannot express the water absorption half of this.)\n\n**Inferred — vision or category norms.** Colour family, look and style (wood-look, concrete-look, terrazzo), texture, pattern, room suitability. Probabilistic. This is the tacit industry judgment.\n\n**Asked — chase the supplier.** The only route for anything unsafe to guess.\n\n### The hard rule: split the fields in two\n\n**Compliance and safety fields** — slip COF or R-value, fire rating, frost resistance, load ratings, PEI — are *measured test results*. Never infer, never publish a guess. Leave blank and chase. A wrong value here is the liability event on the risk register.\n\n**Merchandising and discovery fields** — colour family, look, style, texture, room, finish family, application tags — can be inferred freely. Wrong costs a slightly worse filter.\n\nThe reframe is that merchandising fields are exactly what suppliers never provide and what stores most need for filtering, SEO and comparison. Inference targets precisely that gap, which lands the idea back in digital-marketing value rather than data entry.\n\nMissing fields also carry signal. PEI grades the wear resistance of a *glazed* surface; unglazed tiles, including many through-body porcelains, receive no PEI rating at all. An absent PEI is therefore evidence about the product, not merely a hole in the data.\n\nAnd there is a limit on over-deriving: water absorption, PEI and COF/slip are distinct complementary attributes. Slip rating is a measured test result and is **not** reliably derivable from finish or appearance — which is exactly why the liability risk stays live.\n\n### The claim this page leaned on hardest, refuted\n\nThe original position was that vision-based enrichment is shipped but not in this vertical: the vendors target fashion, electronics and automotive, so applying it to tile and lighting vocabulary was open ground. The confidence register itself flagged this as \"the single most load-bearing and least verified claim on the page\" and classified it as an unverifiable absence.\n\nIt was verifiable, and verifying it broke it.\n\n**SKU Launch is a product of Start With Data, a Melbourne company.** Its industry list names Building Supplies and DIY & Tools, and it does AI document parsing of supplier PDFs and spec sheets. Named case studies include Huws Gray (\"product data enrichment for a leading builders merchant\"), MKM Building Supplies, Castorama and Kingfisher Group.\n\nThe evidence was already in hand. startwithdata.co.uk had been cited twice in the source index — including an article titled \"the 5 best PIMs for builders merchants\" — without the connection being made to the same company that makes SKU Launch. Worse: the Australian angle, which this research treated as unexplored home turf, is already occupied by an Australian vendor selling into builders merchants.\n\nTwo statements had to go: that building materials are absent from these vendors' industries (false), and that the application to tile and lighting vocabulary is unoccupied (too strong — partly occupied).\n\nImage-sourced attribute extraction is likewise already shipped:\n\n- **Pimberly ImageAI**: \"tagging objects in photos, identifying materials, recognizing shapes or patterns, assigning accurate metadata\", with ColorAI standardising colour names plus hex. Its worked examples are fashion and homewares, but its customer list is construction — including Headlam.\n- **Akeneo** Supplier Data Manager, AI Extraction: \"Image assets can also be used as sources, provided they are of type media.\" Shipped, no building examples found.\n- **Catsy**: \"Computer vision to extract attributes directly from your product images\", aimed at industrial manufacturers with thousands of technical SKUs.\n- **Lucidworks** (20 January 2026): multimodal generative AI over product images plus text, generating categories, keywords, synonyms and descriptions, claiming an 8.66% conversion lift and USD 25M annualised for one retailer. A Lowe's association appears only in a third-party source and is unconfirmed.\n- **Hypotenuse AI** enriches from images, spec sheets, PDFs and supplier feeds. One unnamed case study — a home-goods marketplace, USD 120M GMV, 80,000 SKUs — records 40%+ of SKUs missing material or dimension attributes, a useful data point on how severe supplier omission is in an adjacent vertical.\n\nThe primitives are sold generically too: Google publishes a Gemini retail product-attribute-extraction notebook, and Vertex AI Search for Commerce treats \"materials\" as a default system attribute. No building-materials-branded example from Google, AWS or Azure — the primitive is marketed, the vertical framing is not.\n\n### A whole competing axis that had been missed\n\nBeyond the PIM vendors sit companies already machine-processing building-product imagery, purpose-built for it.\n\n- **Renoworks** (TSXV-listed): 200+ building-product manufacturer and distributor customers, claiming \"the most extensive catalog of digital building products and colors\", each integration adding materials, colours and configurations that power its AI visualisation. An entire public company doing this for building products.\n- **Floori**: flooring, tile and wall only. Digital twins of physical samples; extracts UV and PBR maps from ordinary photographs; upscales wood, epoxy, LVT and ceramic textures; exports BIM-ready.\n- **Roomvo / Leap Tools**: automated flooring and tile catalogue across 250+ brands and 7,000+ dealers — porcelain, ceramic, glass, mosaic, floor, wall, backsplash. Four US patents.\n- **Tile-ID apps** (TilesView, imageidentifier.ai, TilesDisplay) classify from a single photo: material (ceramic/porcelain/stone/mosaic), finish (glossy/matte/textured/polished), look (stone-like, wood-look), pattern, size, grout width. Consumer and dealer facing rather than PIM facing — but this is precisely the inference task called the surviving moat, and it is commodity.\n- Also Toolbx, B2Sell (\"AI Product Data Enrichment for Distributors\"), Ximilar (Home Decor vertical), and Domus Image Search (ML image search for tiles and finishing materials, aimed at architects).\n\nOne caution worth weighing: **somebody tried an adjacent market and retreated.** Lily AI ran a home vertical in 2023 built on exactly this — vision tagging Fabric/Material (glass, wood, metals), Wood/Tone, and Material/Treatment/Finish (shiny, glossy, frosted, matte), naming Arhaus and Bridge Furniture & Props. That post now 404s, and lily.ai lists only apparel, footwear, beauty and luxury, with no computer-vision mention. Understand why before assuming the economics work.\n\n(Material Bank does *not* do automated extraction — its tagging is user-applied and SmartMatch is curation assistance.)\n\n### What actually survives\n\nThe refutation cannot be softened into survival, because the claim's function was to establish that the inference layer was unoccupied. It has to be replaced.\n\nThe honest restatement: the vertical is well served on the **PIM/syndication axis** (Syndigo, Sales Layer, Pimberly, Bluestone, inriver, Salsify, SKU Launch — all with named building-products customers) *and* on the **visualiser axis** (Renoworks with 200+ manufacturers, Floori, Roomvo/Leap Tools, tile-ID apps). **Nobody has joined them.**\n\nThe defensible gap is that intersection — narrow but real. No vendor was found combining vision-derived surface, finish and look tagging with a **standards-grade building-product attribute taxonomy**. Across every vendor taxonomy checked, slip rating (R-value/PTV), PEI abrasion class, water absorption percentage, frost resistance, rectified edge and V-shade variation returned **zero hits**. The deepest published building attributes found anywhere were density, fire resistance, water resistance and dimensions (Sales Layer).\n\nThis makes the ETIM depth question the most important open item rather than a nice-to-have. The surviving thesis is no longer \"we infer where others cannot\" but \"we infer *into the compliance-grade schema* the specification channel needs, which merchandising-focused vendors have no reason to build\" — a thesis the adoption evidence in the previous chapter puts under direct threat.\n\nTwo positioning consequences follow. The **visualiser companies are the more dangerous competitor set**, not the PIM vendors: Renoworks owns the largest digital building-product library and Floori already extracts material properties from photographs. And **Pimberly is the acute threat** — it ships ImageAI *and* sells to Headlam, so it holds both halves and has merely not combined them.\n\n## The review surface is the product\n\nAn early competitor entry overstated Matrixify. From direct product knowledge: Matrixify provides no user interface into the data. Its configuration is over CSV/Excel only. It is a bulk import/export engine, not a place where product data lives and can be inspected or corrected.\n\nWhy that is structural rather than a missing feature: the surviving thesis is inference — filling fields the supplier never sent, some derived by rule and some inferred from imagery. **An inferred value is epistemically different from an extracted one.** It needs a confidence score, a source pointer, and a human confirm or reject before it is safe to publish. A CSV round-trip cannot express \"this colour family was inferred from the product image at moderate confidence, click to confirm\". The moment the artefact is a spreadsheet, provenance and confidence collapse into flat cell values and the review step becomes unauditable.\n\nSo file-based tools are not merely behind on UI — they are **architecturally excluded** from the inference product.\n\nWithout overclaiming: some app-tier competitors do have review surfaces. Importier generates descriptions and then lets you review everything in a bulk editor. The honest distinction is not \"UI versus no UI\" but **persistent canonical record versus import batch**. A bulk editor reviews one import into one platform and is then discarded. The inference thesis requires a durable record where every field carries extracted / derived / inferred / human-asserted status, where corrections persist, and where those corrections accumulate.\n\nThe consequence for positioning: the review interface is not a nice-to-have layered on the pipeline. **It is the product surface**, because it is simultaneously the safety mechanism for the never-infer compliance rule and the collection mechanism for the corpus.\n\nAnd that corpus is the one asset that compounds. Every human correction in the review queue is a labelled example linking an image plus partial fields to a correct value, in a vertical the enrichment vendors serve fashion instead of. The taxonomy is public and extraction is commodity; this corpus is neither.\n\n### Form factor follows from that\n\n**A standalone web app at the core, thin connectors at the edges.** Not a Shopify app, not an Excel plug-in.\n\nShopify's own guidance names four cases where standalone beats embedded, and all four apply here: the UI needs more real estate than admin-chrome width (the review queue puts image, source snippet, confidence and canonical value side by side); it is used by non-merchant roles (suppliers confirming data, agency staff, architects); it spans multiple storefronts; and it is a full tool where Shopify is one integration among many.\n\nThe cost is known and should be accepted knowingly: installed-to-first-use runs 70–85% embedded versus 45–60% standalone. So a Shopify app should still exist — as a **distribution and publish channel**, a thin client onto the standalone record, not as the product.\n\nExcel is the wrong home but the right doorway. Office.js add-ins are web apps in a sandboxed iframe across web, Windows, Mac and iPad, so it is technically viable — but a spreadsheet cannot carry per-field confidence, provenance or extracted/derived/inferred status, cannot support multi-user review, and does not persist corrections. That is the same structural disqualification established for Matrixify. Add-ins are also strictly sandboxed by the browser engine and **cannot access the local drive**, so bulk ingestion of local supplier files from inside an add-in is constrained. Excel and Sheets belong as import/export surfaces only.\n\nOn the publish path, Shopify's Bulk Operations API is the right mechanism: async GraphQL returning JSONL, handling tens of thousands of records without hitting standard rate limits. Caveats: one bulk operation per type runs at a time per shop, mutations can partially fail so retry logic is yours, and there are limits of 128 metaobject definitions per app with 30 fields per definition.\n\nThere is also an unresolved commercial question. A Shopify developer-community thread covers off-platform billing for connector apps where the vendor is a standalone SaaS acquiring zero merchants through Shopify, and whether an exemption applies versus split billing. If the Shopify app is only a publish channel onto a standalone product — the recommended architecture — billing needs deliberate handling and is not automatic.\n\n### Build shape: assembly, not invention\n\nNearly every component already exists as open source, which cuts both ways.\n\n**Taxonomy, the big one.** The ETIM Classification Model is published under the Open Data Commons Attribution Licence, free for everyone, with downloadable full-model files plus a Swagger-documented REST API using OAuth2. As established above, bSDD serves the same content with no authentication at all. The canonical schema is not merely public — it is programmatically consumable today at zero licence cost.\n\n**PIM backbone.** Pimcore is open source and self-hostable, spanning PIM, DAM and MDM, and one comparison scores it 35/40 across eight PIM capabilities against Akeneo's 31/40, leading on data modelling, governance, digital asset linkage, localisation and interoperability. Critically, **Akeneo Community Edition has had no new features since 2023** — last major CE release v7 in March 2023, support only to September 2026, with development moved to paid editions. That is a live gap. Alternatives: AtroPIM, UnoPIM, PixeePIM.\n\nThe economics are not free, though. Akeneo CE is free to licence but budget EUR 15,000–40,000 for initial setup via a partner, with an operational instance at EUR 1,500–3,000 per month all-in (server, maintenance, plugins), excluding in-house developer time. Pimcore commercial deployments start at Professional, USD 9,900/yr; Pimcore Community has no cloud-hosted option, so you self-host and need a PHP/Symfony developer to configure the server and build connector logic per channel. And a material omission now corrected: **Akeneo leads on supplier onboarding**, via its Supplier Data Manager — directly adjacent to this idea's ingest half, and missed in the first competitor scan.\n\n**Document extraction**, with a precise benchmark on the same test: LlamaParse 8.54 seconds, Marker 47 seconds, Docling 1 minute 50. The architectural reason matters — Docling runs one large vision-language model reading each page as an image and generating table structure token by token, which is inherently slow, while Marker runs a five-stage pipeline of smaller specialised models doing mostly classification and detection, avoiding token-by-token generation. Docling was separately reported at 97.9% accuracy on complex table extraction. Unstructured has the strongest OCR.\n\nThere is a constraint here that forks the architecture rather than expressing a preference: **LlamaParse is cloud-hosted on LlamaCloud, so documents leave your infrastructure.** Supplier price lists routinely carry confidential trade pricing, which disqualifies the fastest option and pushes the build toward Marker or Docling running locally.\n\n**Attribute extraction**, directly on point. `wbsg-uni-mannheim/ExtractGPT` covers LLM attribute value extraction; `wbsg-uni-mannheim/wdc-pave` covers extraction *and normalisation* — the exact two-step needed. `google/langextract` does structured extraction with precise source grounding, which is this idea's provenance requirement. For structured output: Instructor, Outlines, PydanticAI, Guidance.\n\n**Deduplication.** Splink does probabilistic record linkage — a million records on a laptop in about a minute, 100M+ on Spark. Also Zingg and dedupe.\n\nThe implication is uncomfortable: build cost is **assembly, not invention** — cheap for us, and equally cheap for anyone else. That reinforces the conclusion that the correction corpus, not the pipeline, is the only asset that compounds.\n\nOne more scope note on the standard: products outside electrical, HVAC, plumbing and building materials — consumer electronics, food, fashion, industrial machinery — have limited or no ETIM coverage. ETIM is the right standard for this vertical and would not generalise if the idea expanded beyond building products.\n\n## Where normalised data could actually go\n\nIf the thesis is normalise-once/publish-many, the destinations matter as much as the source. Nine specification platforms were researched independently across two continents, and the finding is blunt: **not one of nine accepts a product feed.**\n\n- **Material Bank**: no API (api.materialbank.com returns 403, no developer subdomain). Ingest is an \"MB New SKU Template.xlsx\" emailed in, after which their staff enrich it. Updates mean requesting an export, editing it, and sending it back, taking one to five days. On the physical side, ship 30–40 samples per SKU to their hub with an advance shipping notice.\n- **Archello**: a manual self-serve wizard, one product at a time. No API, no feed.\n- **Archiproducts**: managed only; robots.txt disallows /api. Furniture-weighted — a design marketplace with a building-products wing.\n- **NATSPEC, Masterspec, EBOSS, ArchiPro, Archify**: all account-managed or done-for-you.\n- Only **ARCAT** (MasterFormat) and **NBS Source** (Uniclass) are standards-based. Every AU/NZ platform uses a proprietary taxonomy.\n\nThe incumbent already sells the normalisation layer. Material Bank offers manufacturers \"PIM Lite\" — automatic attribute mapping plus a per-product data health score — with a next-generation PIM upsell to push enriched data to your *other* channels. That is this idea's product, sold by the platform.\n\nMaterial Bank's economics are order-of-magnitude only. Forbes (2021) reported manufacturers \"pay a monthly fee in the thousands… and additional fees when their items are ordered\". Business of Home records that Material Bank declined to disclose, with sources suggesting \"in the realm of $25 per order\" — trade-press estimate, not a rate card. The only published numbers are penalties: barcode $0.15 per piece, missing ASN $250, missing packing list $125. And it does not reach this market at all: Material Bank ships to the contiguous US and Canada only, with hubs in Olive Branch, Mississippi and Crépy-en-Valois, France.\n\n### The load-bearing unknown\n\n**BIMobject's write API is unverified, and the evidence conflicts.** One thread documented an active OAuth2 Publish API at publish-publicapi.bimobject.com behind gated approval. Another found the public content API returning 410 Gone / permanently shut down, while developer.bimobject.com still advertises \"the most powerful BIM content API\" and serves no endpoints. These are plausibly two different APIs — a withdrawn read API and a live partner write API — but nobody completed an OAuth handshake against either.\n\nThis is *the* load-bearing assumption in the spec-library feasibility verdict. If BIMobject has no working write API, then not one of nine platforms accepts programmatic ingest and the specification-channel pivot is a services business by construction. **Test this before any build decision:** request BIMobject partner API access and attempt an authenticated publish.\n\nRelatedly, whether ARCAT, NBS, Archbase, BIMsmith or Archify run private partner APIs under NDA is not researchable by search. That is a sales call. Absence of public documentation is not proof of absence.\n\n### Consolidation, and which company actually matters here\n\nAn earlier entry recorded Anguleris acquiring Concora Spec in February 2025, adding it to a portfolio holding BIMsmith, Swatchbox and Modlar — one owner across research, content, sampling and specification, meaning fewer integration targets and an incumbent with obvious motive to build ingestion itself.\n\nThat framing is right in kind but wrong in company for this market. **The consolidation that matters for AU/NZ is Hubexo**, which owns NBS UK, NBS Australia, Archify across APAC, *and* Uniclass. The local channel and the only free, API-accessible classification system have the same owner. A dependency on Uniclass in a product competing for the same manufacturer budget as Archify is a single point of commercial failure.\n\nTwo further constraints on any \"normalise between taxonomies\" play:\n\n**Uniclass is CC BY-ND — no derivatives.** NBS publishes a native REST API returning JSON/XML notation, title and children trees, registration-gated, with quarterly table downloads — better access than previously recorded. But the licence is Creative Commons Attribution-NoDerivatives 4.0. A cross-taxonomy mapping table (ETIM class to Uniclass notation) is exactly the artefact this idea would build and might sell, and it is a derivative work. Publishing or distributing it is likely a licence breach. Internal use is probably fine; the mapping as a product is not. Get legal advice before treating taxonomy mappings as a saleable asset.\n\nAnd bSDD gives mapping *capability* but not the dataset — it contains no OmniClass and no MasterFormat at all, so a mapping to the US taxonomies cannot be built from it. Only 2 of 14 platforms accept data programmatically.\n\n### The compliance layer is where the published prices are\n\nThe one part of this research with fully published prices and a mandated machine-readable format is the environmental-declaration layer.\n\n- **EPD Australasia** fees are entirely public: joining AUD 1,075–1,600, annual AUD 720–4,300, and **the first 150 EPDs are free**. Critically, publication *requires* \"an approved MS Excel template for machine-readable EPDs\" — a structured-data obligation with real consequences, unlike anything the spec libraries impose.\n- A live destination with a price: EPD Australasia's July 2026 partnership with Rosella Street puts verified Australasian EPDs into the AU/NZ DPP database at AUD 105 per EPD for members or AUD 300 standard, with no ongoing cost. Caveat: that platform's APIs are described as \"upcoming\", not shipped.\n- **Eco Choice Aotearoa** is fully public: NZD 1,000 application plus NZD 1,000–61,000/yr scaled by turnover.\n- **CodeMark** cost is explicitly \"a commercial arrangement between the applicant and the CodeMark Conformity Body\" — unquotable by design.\n- **GBCA does not certify products at all.** It recognises third-party initiatives and assigns a Responsible Products Value your product inherits. Anyone pitching \"GBCA product certification\" has misread it.\n- **NABERS** is buildings and tenancies only — nothing a manufacturer can obtain.\n- **The NCC itself** is published as XML with a schema on data.gov.au under CC BY 4.0. The code is machine-readable; the products are not.\n\nThat asymmetry — machine-readable code, unstructured products — is the gap the Australian common data template exists to close, and an EPD obligation that already demands a machine-readable Excel template, free for the first 150 products, is a cheaper wedge into a supplier's data than asking them to care about a taxonomy.\n\n### Australia already has an industry-agreed schema, and it is not in a BIM library\n\nThis inverts the premise of the specification-channel pivot and is the strongest strategic finding in the dossier.\n\n**The NBPC Common Data Template**, in the National Building Products Coalition's *Building Product Information Traceability and Digitalisation Guide* V4.1 (February 2026). Its lineage runs from the 2018 Building Confidence Report to the ABCB National Building Product Assurance Framework (2021), Element 3.\n\nWhat it mandates per product:\n\n- A globally unique identifier per AS ISO/IEC 15459.6 — in practice a **GTIN**.\n- Name, brand, description, model number, net content, unit of measure, variant.\n- Manufacturer and supplier contacts with URLs.\n- A **statement of application and intended use** naming the NCC edition, building classification, construction type and the specific NCC clause.\n\nCrucially, **product classification is explicitly optional**, with Uniclass, OmniClass and ETIM all named as acceptable. So in Australia the taxonomy war this research has been fighting is not the binding constraint — identity and intended-use are.\n\nAnd there is a programmatic destination waiting. GS1 Australia runs a Building & Construction programme partnered with the NBPC, buildingSMART Australasia (MOU), BPIC, the Australian Steel Institute, Standards Australia and the ABCB. Its National Product Catalogue is a **GDSN-certified data pool** — genuinely machine-to-machine. Construction is not yet a live vertical on it. GS1 membership fees are published (AUD 225 joining, 675–22,895/yr); NPC subscription is not.\n\nThe reframe: the spec libraries are manual, but Australia's identifier-and-compliance layer is standards-based, GTIN-anchored, and has a real data pool with an empty construction vertical. If the thesis is normalise-once, the schema to normalise *to* in this market is the NBPC template — not any platform's field list, and not necessarily ETIM. That makes the ETIM extension work optional in Australia rather than unavoidable.\n\nNew Zealand adds the cleanest gap-shaped finding in the entire dossier: the **Building Product Information Requirements have been mandated in law since December 2023**, specifying required fields with **no mandated machine-readable format**. A statutory obligation with no schema.\n\n## Specification channel — the 14-platform feasibility table\n\n### The table\n\n| Platform | Programmatic ingest? | Manufacturer cost | Taxonomy | AU/NZ |\n|---|---|---|---|---|\n| **BIMobject** | **Documented OAuth2 write API** (full CRUD on products/files/images/properties); access manually approved per brand | Not public — Lite/Essential/Pro all \"Custom price\" | Proprietary + 11 optional slots: ETIM, Uniclass 14/20/2015, OmniClass, MasterFormat 2014, UniFormat II, UNSPSC, NBS ref, IFC, COBie | **No office**; NATSPEC classes it \"European\" |\n| **ARCAT** | No API. Manufacturer-submitted portal + ARCAT-authored managed content | Not public (free to specifiers) | **CSI MasterFormat** div 02–48 | No |\n| **CADdetails** | No API. \"Onboarding Specialist\" + in-house design team | Not public; tier *names* only (10/50/100/100+ products) | Undocumented (403s) | No — North America only, 610+ mfrs |\n| **NBS Source / Chorus** | No public API. Partner Platform portal; NBS authors objects to NBS BIM Object Standard. **PIM streaming is explicitly roadmap** | Not public | **Uniclass 2015** + CAWS; NCS codes in AU. Aligns ISO 12006-2, 16739-1, 23386, 23387 | **Yes** — thenbs.com.au, absorbed SCL Schumann/EzySpec |\n| **SpecifiedBy** → CausewayOne | No API. Ingest is a **Causeway spiderbot scraping your own website** | Not public — \"fixed development fee\" | None published | No |\n| **Concora Spec** → **Archbase** | No API, no feed. Managed deployment + \"the Manager\" CMS. Integrations are Salesforce/GA, **not PIM** | **PUBLISHED: from USD 1,000/month** billed annually + one-time deployment fee | **Proprietary, customer-defined** (\"Category Genius\", paid extension) | No |\n| **BIMsmith** | No API, no self-serve. BIMsmith's own architects author every Revit family | Not public | Unresolved; MasterSpec-linked | Not evidenced |\n| **Modlar** | Self-serve web upload behind login. No API | Free 6-product page; paid \"unlimited product feed\", **amount not public** | Proprietary tags | NZ-origin investors; `bim-australia` tag **empty**; sitemaps stale since 2016 |\n| **Swatchbox** | Physical 3PL + embeddable storefront. Claims \"world's first sample API\" — **2019 marketing claim, zero docs** | Not public (free to A&D) | None | Structurally hostile (US warehouse consignment) |\n| **Unifi** | **Acquired by Autodesk, Mar 2023** → \"Content Catalog\". Real REST API exists but is **firm-side parameter CRUD**, not manufacturer publishing. UNIFI Connect has vanished from messaging | Not public / bundled | Firm-defined | Via Autodesk ANZ channel |\n| **Material Bank** | No API. **Emailed `MB New SKU Template.xlsx`**, then their staff enrich. Updates: request export, edit, send back (1–5 days). Physical: **30–40 samples/SKU** to the Hub + ASN | Forbes: brands \"pay a monthly fee **in the thousands**… and additional fees when their items are ordered\". Business of Home: declined to disclose, sources suggest **~USD 25/order** (estimate) | Proprietary, **35 material categories**. Only standard is **Common Materials Framework** (sustainability only) | **Effectively zero** — ships contiguous US + Canada only |\n| **Archello** | Manual self-serve wizard, one product at a time. No API/feed | Not public. Published: **EPD sharing free**; Awards **EUR 225/300/375 per product** | Proprietary, 14 top-level element-based. Only standards field is **GWP A1–A3 (EN 15804)** | **AU edition, no NZ**; AU brands listed |\n| **Archiproducts** | No — fully managed. Owner **Edilportale.com SpA** (Bari). `robots.txt` disallows `/api`. BIM hosted *or* authored by their team | Not public — every CTA is \"Request Media Kit\" | Proprietary ~20 categories, **furniture-weighted** | Locale only; `/en-au` renders in **EUR** |\n| **buildingSMART bSDD** | **YES — the only open one.** Free public REST + GraphQL, documented write endpoint `POST /api/UploadImportFile/v2`, org registration + review | **Free** for public dictionaries | Hosts IFC 4.3, Uniclass, ETIM 7.0–10.1, CCI, GS1, NL-SfB. **No OmniClass, no MasterFormat** | bSI Australasia chapter, but **zero AU dictionaries** |\n\n**Also found, not in the original brief:** **Deltek Specpoint** (AIA MasterSpec successor) documents a genuine **self-serve manufacturer upload** — \"a self-serve environment where manufacturers can upload, maintain and present products\" — over 50,000+ product listings, MasterFormat-organised. Pricing not public.\n\n### Why \"partly\" and not \"feasible\"\n\nTwo hard blockers, structural rather than missing features:\n\n1. **No destination accepts a standard feed.** ETIM xChange 1.1 (JSON) and BMEcat are open and exist, but they are wholesale/distribution formats. No architect-facing spec library ingests them. BIMstreamer states the constraint precisely: *\"The only requirement is that the system to which you want to publish the data provides an API or a place (network/cloud folder) to which this data can be sent.\"* Most do not.\n2. **bSDD is a capability, not a destination.** It holds **semantics only** — no geometry, no PDFs, no spec clauses — so it cannot host a BIM object. 406 dictionaries across 106 organisations, and **only 5 are verified**. Owners are national chapters, universities, contractors and demo companies; **almost no building-product manufacturers publish their own dictionaries.**\n\nSo a normalisation layer can **automate two destinations and prepare-and-package for the other twelve** — mapping to each library's schema, generating deliverables, driving human onboarding. Real product; not \"publish everywhere by API\".\n\n### Consolidation: three owners, not one\n\n- **Hubexo** (ex-Byggfakta) owns **NBS UK + NBS Australia + BCI/Archify across APAC + Glenigan** — *and delivers Uniclass* (`uniclass.thenbs.com` footers read \"© Hubexo North UK Ltd\"). Its Product Information page already markets the thesis: *\"for manufacturers, our platform simplifies data management across all channels, keeping everything consistent and easy to update\"* (marketing claim). **This is the owner that matters for AU/NZ.**\n- **Anguleris** acquired Concora on [18 February 2025](https://www.prnewswire.com/news-releases/anguleris-acquires-concora-extending-leadership-in-bim-technology-for-building-product-manufacturers-302377194.html), alongside BIMsmith, [Swatchbox](https://www.swatchbox.com/blog/Swatchbox-Combines-Technology-and-Design-with-Groundbreaking-Architecture-and-Design-Sample-Service) and Modlar. It does **not** publicly market syndication as a SKU — but it shipped the plumbing in 2019: BIMsmith pushes outward to **UNIFI Connect with documented \"live sync of product data\"**, and to Avail. That destination is now Autodesk-owned and Connect has disappeared from messaging.\n- **Autodesk** absorbed Unifi (Mar 2023).\n\n### The willingness-to-pay evidence\n\nThis is the strongest evidence in the dossier, because the numbers are published and the pattern repeats independently.\n\n- **An entire listed company is funded by manufacturers paying to publish.** BIMobject's audited filings: *\"The Group's revenue is primarily generated from manufacturers and distributors of building products\"*, with platform revenue defined as *\"recurring revenue from various subscriptions, such as product publishing\"*. **ARR SEK 145.7m** (YE 2024) → **147.5m** (Q1 2026), **81–85% recurring**, ~**2,400 brands listed**. *(Deliberately not divided into an ARPU — the filings disclose no customer count, and \"brands listed\" likely includes non-paying legacy listings.)*\n- **Published prices in the adjacent \"own your library\" segment, where integration is the upsell:**\n  - **bimstore (UK): GBP 3,120 / 3,600 / 4,800 per year** for 10 / 50 / 100 listings — and **CRM/PIM API integration is Enterprise-only**.\n  - **BIMstreamer (PL, since 2017): EUR 500 / 700 / 1,100 / 1,600 per month** (EUR 5k/7k/11k/**16k** per year, annual saves 17%), plus Enterprise. The **EUR 16k/yr Advanced tier is explicitly the one that \"integrate[s] your BIM libraries with external systems (e.g. PIM, CRM, ERP)\"**. It sells a product literally called **\"PIM to BIM\"**.\n  - **Archbase: from USD 1,000/month.**\n- **Two independent vendors gate exactly this capability behind their top tier.** That is the clearest signal in the dossier that the integration itself commands a premium.\n\n### Vendors already selling this wedge\n\n- **ProdBIM** distributes to *\"ProdBIM Website, Bim&Co, BimObject, and +20 product libraries across the Eurovent ecosystem\"*; ProdBIM + **ETIM** taxonomies; Excel/JSON export; no public API; no pricing.\n- **BIM&Co (Onfly)** sells a read API — *\"clients can connect their software to the BIM&CO API\"* — plus a **PIM connector**: *\"Connect your PIM to Onfly and refresh your BIM data from your tool without changing your internal processes.\"* Classifies by **OmniClass, UniFormat II, MasterFormat**.\n- **BIMstreamer** articulates the pain verbatim: *\"Manufacturers of building materials are increasingly making their products available as BIM objects, which must be distributed to multiple channels simultaneously — BIM marketplaces, websites, catalogs and e-commerce systems\"* and *\"Without a central PIM approach for BIM data and automatic synchronization of information, control over the distribution and timeliness of BIM objects quickly gets out of hand.\"*\n- **Distributor Data Solutions** syndicates to 475+ distributor endpoints — adjacent-market proof the model works.\n- **BIMobject itself** reports *\"growing demand for 'Content-maintenance-as-a-service', a subscription-based model in which BIMobject ensures that the customer's BIM portfolio is kept up to date.\"*\n\n**Read this two ways.** Demand is validated — and the wedge is already occupied by at least three vendors, one of which (BIM&Co) has both an API and a PIM connector.\n\n### Regulatory drivers\n\n- **EU CPR (EU) 2024/3110** — in force 7 January 2025, applying in stages from 8 January 2026, phasing through ~2032. Mandates Digital Product Passports and *\"a common data dictionary based on European standards\"* for machine readability.\n- **NZ BPIR** — law since 11 December 2023. See the AU/NZ section.\n- These force the normalisation work regardless of marketing ROI.\n\n### What could NOT be established\n\n- **Pricing for:** BIMobject, ARCAT, CADdetails, NBS Source, SpecifiedBy/Causeway, BIMsmith, Modlar (amounts), Swatchbox, Unifi, Material Bank, Archello, Archiproducts, Archify, Deltek Specpoint. All quote-based. **Not estimated.**\n- **Material Bank's rate card** — login-walled; the two figures above are trade press, order-of-magnitude only.\n- **Whether ARCAT, NBS, Archbase, BIMsmith or Archify run private partner APIs under NDA.** Not researchable — a sales call. **Absence of public docs is not proof of absence, and this is the highest-value unknown.**\n- **Whether Swatchbox's \"sample API\" exists** in any usable form.\n- **Whether Unifi's API still functions post-Autodesk**, and whether UNIFI Connect still exists.\n- **CADdetails' taxonomy** (403) and **BIMsmith's classification** (JS shell).\n- **CSI licensing terms for redistributing OmniClass / MasterFormat mappings** — csiresources.org returned no licence text. Live commercial risk for any US crosswalk.\n\n### The unresolved conflict\n\n**BIMobject's write API.** One research thread documented an active OAuth2 publish endpoint (`publish-publicapi.bimobject.com`) with full CRUD; another found the public **content** API returning **410 Gone**, with `developer.bimobject.com` advertising \"the most powerful BIM content API\" while serving no endpoints. Plausibly two different APIs — a read API withdrawn, a partner write API alive — but **nobody completed an OAuth handshake against either.**\n\n**This is the single load-bearing assumption behind \"partly feasible\" rather than \"blocked\". Test it first.**\n\n## The Australian market, sized from primary data\n\nThe geography gap — the idea originated in Australia and Australia had never been researched — was closed using ABS data cubes and IBISWorld public pages.\n\nThe structural problem comes first: **there is no ANZSIC class for tiles, and none for lighting.** Both trades are statistically invisible. ANZSIC 4212 Floor Coverings Retailing *explicitly excludes* ceramic floor tiles, which sit in 4231 Hardware & Building Supplies Retailing alongside Bunnings. Ceramic tile wholesaling sits in 3339 Other Hardware Goods Wholesaling. IBISWorld inherits the gap — there is no \"Tile Wholesaling in Australia\" report at all. Any tile count has to be built bottom-up.\n\nABS business counts at 30 June 2025 (8165.0, released 16 December 2025), shown as total and the $2m+ turnover cut:\n\n| ANZSIC class | Total | $2m+ |\n|---|---|---|\n| 4212 Floor Coverings Retailing | 2,005 | 614 |\n| 3731 Furniture & Floor Covering Wholesaling | 1,823 | 475 |\n| 3339 Other Hardware Goods Wholesaling (incl. ceramic tile) | 5,361 | 1,442 |\n| 4231 Hardware & Building Supplies Retailing (incl. ceramic floor tile) | 5,880 | 1,652 |\n| 4229 Other Electrical & Electronic Retailing (incl. light fittings) | 1,700 | 328 |\n| 3494 Other Electrical & Electronic Wholesaling | 4,379 | 1,285 |\n| 3243 Tiling and Carpeting Services | 20,560 | — |\n| 2029 Other Ceramic Product Manufacturing | 445 | — |\n| 2432 Electric Lighting Equipment Manufacturing | 378 | — |\n\nIBISWorld corroborates the ABS closely: Floor Covering Retailing AUD 3.7bn across 2,049 businesses; Furniture & Floor Covering Wholesaling 6.6bn across 1,832; Electrical & Lighting Stores 2.8bn across 1,802; Ceramic Product Manufacturing 791.5m across 452. Its note that \"imports account for a high share of industry revenue\" confirms the Australian tile trade is an **import trade** — exactly the situation this idea started from.\n\nApportioned from those inputs: roughly **1,500–2,000 Australian businesses at $2m+**, about 1,800–2,450 including New Zealand. Working number: ~2,000. The full universe including sub-$2m operators is roughly 6,000–9,000. NSW leads wholesale and import while Victoria leads retail, consistent with Sydney as the import gateway.\n\n### What those suppliers actually run\n\nMeasured by fetching the sites and matching platform markers, not assumed:\n\n- **Magento 2**: Amber Tiles, National Tiles, Beaumont Tiles, Beacon Lighting.\n- **Shopify**: TileCloud, Earp Bros, Cult Design.\n- **WooCommerce**: Signorino, Tile Factory Outlet.\n- **No storefront detected**: Artedomus, Academy Tiles, Di Lorenzo, Skheme, Surface Gallery, Living Edge — catalogue and brochure sites only.\n\nThere is a clean split by scale: large Australian chains run Magento/Adobe Commerce; newer DTC and mid-size importers run Shopify; WordPress/WooCommerce covers those that grew out of a brochure site; several premium importers sell entirely offline.\n\nThis matters more than it looks. The Shopify-running importer this idea started from sits in the **middle band**, and its peer group is genuinely mixed rather than Shopify-dominated. A Shopify-first product addresses a subset of a subset. It also means the Cin7-to-Shopify gap does not reach the biggest Australian players — they are on Magento, where Cin7 also drops attributes.\n\n### The organised supply side is tiny\n\nThe Australian Tile Council is the real body — 50+ years, \"Australia's only recognised tile industry representative\" — and claims members represent around **70% of all tiles sold in the country**. Its live directory shows 140 members: Tile Merchant 45, Contractor 31, Other 21, Consultant 14, Builder 11, Wholesaler 6, Manufacturer 5, Industry Supplier 4. So the entire organised supply side is about **51 merchants and wholesalers plus 5 manufacturers**. Membership is AUD 150–330/yr. By state: QLD 36, NSW 35, VIC 32, SA 28, WA 28.\n\nNew Zealand: New Zealand Tile Council 52 members; TANZ about 197 listings of which roughly 100 are genuine independents; FloorNZ 314 unique; Lighting Council NZ 25.\n\n(A correction from the research itself: ATTIA is the Australian *Tea Tree* Industry Association. There is no \"Australian Tile and Terrazzo Industry Association\" — the research brief invented it.)\n\n### Nobody occupies this space locally — which is both the finding and the warning\n\n- **Cin7's own 700+ integration directory has no PIM category and lists no PIM vendor.** Akeneo, Plytix, Salsify, Pimberly, inriver and Bluestone are all absent. The only Cin7-to-PIM offerings anywhere are agency consulting pages (Cogent2 selling Cin7 Core to Plytix / inriver integration work) — services, not products.\n- **Comestri**, the Australian PIM/commerce vendor, now redirects to ultracommerce.co. Absorbed.\n- **BCI Central** rebranded to LeadManager and is not a product-data business — it sells construction project leads.\n- No Australian digital agency publicly claims building-products ecommerce as a specialty.\n\nThe sharpest standards finding: **ETIM AU/NZ exists and is dormant.** ETIM International lists it, run by the Australian Industry Group, announced 26 November 2021 as a multi-country membership for Australia and New Zealand, with scope \"initially the electrotechnical sector\" and discussions to expand into HVAC and plumbing. Its own website, etim-aunz.com, is a parked NameBright \"coming soon\" page — verified by direct fetch, nearly five years after launch. It does not cover tiles, flooring or decorative lighting.\n\nNo Australian product data template regime exists either. NATSPEC's work is BIM-shaped: an Open BIM Object Standard v1.0 (2018, jointly with Masterspec NZ) plus a BIM Properties Generator, targeting geometry and object properties rather than commerce attributes. Its ~50 Product Partners get branded worksections, not a data feed. There is no Australian equivalent of the UK's NBS/CIBSE Product Data Templates. The ABCB runs conformity registers (NCBP, CodeMark, WaterMark), not product data. buildingSMART Australasia publishes no AU/NZ product data dictionary. New Zealand has miproducts — 1,161 suppliers, 9,590 listings, 115 added in 30 days — on its own CBI taxonomy with no published schema.\n\nNobody will standardise this out from under you. \"Nobody\" is also a warning.\n\n## AU/NZ — regulation, certification economics and bottom-up sizing\n\n### Australia already has an agreed product data schema\n\nIt just isn't in a BIM library. The **National Building Products Coalition's Common Data Template**, in the *Building Product Information Traceability and Digitalisation Guide **V4.1, February 2026***. Lineage: **2018 Building Confidence Report → ABCB National Building Product Assurance Framework (2021), Element 3.**\n\nMandated per product:\n\n- **A globally unique identifier per AS ISO/IEC 15459.6 — in practice a GTIN**\n- Name, brand, description, model number (Global Model Number), net content, unit of measure, variant, serial number where applicable\n- Manufacturer and supplier contact details **with URLs**\n- **A statement of application and intended use naming the NCC edition, building classification, type of construction and the specific NCC clause**\n- A conformity declaration with supporting evidence\n\n**Product classification is explicitly OPTIONAL**, with **Uniclass, OmniClass and ETIM** all named as acceptable (plus GPC, UNSPSC). It also names the Standards Australia + ICC **Universal Data Protocol** as emerging and the **UN Transparency Protocol** for verifiable exchange.\n\n**Strategic consequence: in Australia the taxonomy war is not the binding constraint — identity and intended-use are.** That makes the ETIM extension work described above **optional in Australia rather than unavoidable.**\n\n### GS1 Australia — the programmatic destination in waiting\n\n- Runs a **Building & Construction programme** partnered with the **NBPC, buildingSMART Australasia (MOU), BPIC, the Australian Steel Institute, Standards Australia and the ABCB**.\n- Its **National Product Catalogue is a GDSN-certified data pool** — genuinely machine-to-machine. **Construction is not yet a live vertical** (live verticals are food/grocery/liquor, healthcare, rail).\n- **Membership fees published:** individual GTIN membership from **AUD 79/yr** (turnover <$1M, one GTIN) to $485 for a 10-pack. Full membership **AUD 225 joining + 675/yr** (small) → 1,210 → 1,525 → 5,146 → 12,295 → 16,175 → **22,895/yr** (>$10B), ex-GST.\n- **NPC subscription fees are NOT public** — quote-tool lead form only.\n\n### New Zealand: mandated fields, no mandated format\n\nThe **Building (Building Product Information Requirements) Regulations 2022**, in force **11 December 2023**, oblige NZ-based manufacturers and importers to disclose, per product:\n\n- Product name + a product identifier distinguishing it from other products\n- Product line name + identifier where needed\n- A description\n- **The Building Code clauses relevant to the product within its intended scope of use, and how the product is expected to contribute to compliance with those clauses**\n- Intended scope of use and any supporting design requirements\n- Limitations or conditions on use within that scope\n- Installation requirements; maintenance requirements\n- A statement on any s26 warning or ban under the Building Act\n- Contact details: legal and trading name, address for service, website, public email and **NZBN** — for the manufacturer, and for imported product also the importer\n\n**Critically: \"BPIR regulations do not prescribe any specific layout or formatting of required disclosure information.\"** NZ mandates the *fields* but not a schema.\n\n**That is the cleanest gap-shaped finding in the entire dossier — a statutory obligation with no machine-readable format.** EBOSS and Building Confidence Ltd co-run a **free** self-assessment tool at bpir.nz, which is the obvious partner or the obvious incumbent depending on how it goes.\n\n### Certification and EPD economics — every published fee\n\nThe pattern: **environmental schemes publish their prices; conformity schemes do not.**\n\n- **EPD Australasia** — fully published, effective 1 April 2026. Joining **NZD 1,200 / AUD 1,075** (NZD 1,800 / AUD 1,600 if turnover >$100m). Annual from **NZD 800 / AUD 720** (Micro) to **NZD 4,800 / AUD 4,300** (Cat 4). **No fees for registration and publication of 150 or fewer EPDs per organisation**; updates within validity free. **Publication requires \"a final version of the dataset in an approved MS Excel template for machine-readable EPDs\"** — a structured-data obligation with real consequences, unlike anything the spec libraries impose. Construction datasets then flow to ECO Platform's digital product database (ILCD+EPD XML via the soda4LCA API — gated, **EUR 2,000–10,000/yr**).\n- **EPD Australasia × Rosella Street, July 2026** — verified Australasian EPDs extracted into Digital Product Passports on the **AU/NZ DPP Database** at **AUD 105 per EPD (member) / AUD 300 (standard)**, no ongoing cost. Used on ACT Light Rail Stage 2A. **Caveat: the platform's APIs are described as \"upcoming\", not shipped** — \"with upcoming capabilities for data input through upload, syncing, and barcode scanning… via APIs\".\n- **Eco Choice Aotearoa** (ex-Environmental Choice NZ) — fully published. One-off application **NZD 1,000**, then annual licence by turnover from **NZD 1,000** (≤$100k) through 3,500 / 5,000 / 7,000 / 9,000 / 12,000 / 15,000 / 18,000 / 22,000 / 26,000 / 30,000 / 35,000 / 40,000 / 45,000 / 50,000 / 55,000 to **NZD 61,000** (>$500m). Independent audit fees additional, quoted case-by-case. Ex-GST.\n- **CodeMark Australia** — JAS-ANZ is Scheme Administrator and Accreditation Body; certificates valid 3 years; the mandatory certificate field list is a published spec (CC BY-ND) with every field compulsory except manufacturer/plant details. **Cost is explicitly \"a commercial arrangement between the applicant and the CodeMark Conformity Body\"** — unquotable by design. Register is officially PDF-only, though the backing JASANZ JSON API is open and unauthenticated (388 certificates, 62 withdrawn) — scrape-grade, not a contract.\n- **WaterMark** — ABCB plumbing/drainage. Public product database searchable on licence, model, brand, product type, specification and status. **No JSON/CSV/API** — confirmed by probing. Schedule of Products is a PDF. Scheme fee not public.\n- **Global GreenTag** — Application for Quote → proposal → invoice → certification, on-site audit where required. **Cost not public.**\n\n#### Three corrections for anyone pitching compliance\n\n- **GBCA does not certify products at all.** You get certified by a recognised third-party **initiative**; GBCA issues that initiative a Letter of Recognition carrying a **Responsible Products Value (RPV)** which your product inherits. Product database is behind Green Star Manager login; no API. Recognition fees not public. GBCA's published fee schedule is for **buildings only**. Anyone selling \"GBCA product certification\" has misread it.\n- **NABERS is buildings and tenancies only** — sectors are building types (office, apartment, hotel, hospital, school, shopping centre, data centre, warehouse). **Nothing a manufacturer can obtain.**\n- **The NCC itself is published as XML with a schema on data.gov.au under CC BY 4.0**, consumable via the CKAN API. **The code is machine-readable; the products are not.** ABCB runs conformity registers (NCBP, CodeMark, WaterMark), not product data.\n\n**Sequencing note:** an EPD obligation that already demands a machine-readable Excel template, and is **free for the first 150 products**, is a cheaper wedge into a supplier's data than asking them to care about a taxonomy.\n\n### ETIM AU/NZ is dormant\n\nETIM International lists **\"ETIM AU/NZ\"**, run by the **Australian Industry Group**, announced **26 November 2021** as a multi-country membership for Australia and New Zealand. Scope \"initially the electrotechnical sector\", with discussions to expand into HVAC and Plumbing. An **en-AU** translation is documented, covering only one of ETIM's four sectors, access requiring national membership.\n\n**Its own website, `etim-aunz.com`, is a parked NameBright \"coming soon\" page — verified by direct fetch, nearly five years after launch.** It does not cover tiles, flooring or decorative lighting.\n\n### The AU/NZ specification channel\n\n- **NATSPEC** — national spec, not-for-profit, government + industry owned. **Product Partners** programme: NATSPEC *writes* your Branded Worksection, retains **full editorial control**, and requires NCC conformance. It states plainly that manufacturers **\"have paid us\"** to write one — **fee exists, amount not public**. Scale is small: **~82 branded worksections across ~50 brands** (BlueScope/Colorbond/Lysaght/Fielders/Truecore, Stramit, Stratco, Metroll, Capral, AWS, Breezway, Dulux, Haymes, Taubmans, Resene, Kingspan, Fletcher Insulation, Askin, Moddex, Assa Abloy, Interpon, Soprema, Solatube, Danpalon, Regupol, Getzner, Forbo, Evapco). Taxonomy is **NATSPEC 4-digit NCS** (1989). **NATSPEC TR02, April 2026: \"There is currently no unified construction information classification system, similar to Uniclass or OmniClass, used nationally.\"** It runs **no** national BIM object library — only a free **Open BIM Object Standard** and **BIM Object Properties Generator**.\n- **Archify** (AU/NZ/HK/ID/MY/PH/SG) — part of **Hubexo/BCI**. **200+ brands.** Managed BIM object creation (\"Get in touch\"), proprietary marketing categories, **cost not public** (\"solutions that fit your strategy and budget\"). **ArchifySpec** is an AU cloud spec platform on Australian Standards and the NCC. Its Sentiment Survey: **44% of architects say 10–30% of specified products get swapped**; a further **33% say 30–50%**.\n- **NBS Australia** — real: `thenbs.com.au`, NBS Chorus + Source, **SureSpec structured to NCS codes**, pushing Uniclass in. Owned by Hubexo.\n- **Masterspec (NZ)** — **450+ Product Partners**, 8,500+ specifiers, in-house editorial authors branded work sections, **8–10 week onboarding**, **CBI** classification, eCert for environmental certs. **You host your own CAD/BIM; Masterspec links out.** No API. Supplier cost not public.\n- **miproducts (NZ, Construction Information Ltd)** — 1,161 suppliers / 9,590 listings, **CBI** categories, and **published prices**: standard listing **free**; Product Technical Statement blocks **NZ$500 each for Masterspec Product Partners (first block free)** → $500/$1,000/$1,500; **non-partners NZ$1,200 for the first block** → $1,700/$2,200/$2,700. Ex GST, 12-month minimum. **Note the bundling incentive — cheaper if you already pay Masterspec.** The PTS field list is essentially the BPIR schema plus certificates, NZBC evidence, CodeMark/appraisals and a validation date.\n- **EBOSS (NZ)** — 5,300 products, proprietary categories (Structure/Enclosure/Interior/Finish/Services/External), quote-based, co-runs the free BPIR Ready tool.\n- **Productspec (NZ)** — supplier portal + \"BIM & CAD Services\", proprietary categories.\n- **Design Content (A2K)** — NATSPEC's named *\"most prominent Australian object library\"*. **`designcontent.com.au` is now a generic content-farm blog**; the corporate page 403s. Appears defunct.\n- **RubySketch (AU)** — PlusSpec / BIMBits, SketchUp-centric, managed manufacturer work. **ANZRS** — community Revit standard from RTC 2009, adopted by KarelCAD/DesignContent, Benn Design and Productspec; dormant.\n- **buildingSMART Australasia** — chapter since 1997, but **zero AU dictionaries in bSDD**. AU-linked owners are **RMIT University (11)** and **Lendlease (6)**.\n\n### Market size, bottom-up\n\n**Structural problem: there is no ANZSIC class for tiles, and none for lighting.** **ANZSIC 4212 Floor Coverings Retailing explicitly excludes ceramic floor tiles**, which sit in **4231 Hardware & Building Supplies** with Bunnings; ceramic tile wholesaling sits in **3339**. IBISWorld inherits the gap — **there is no \"Tile Wholesaling in Australia\" report at all.**\n\n**ABS Business Counts, 30 June 2025** (cat. 8165.0, released 16 Dec 2025) — total / $2m+ turnover:\n\n| ANZSIC | Class | Total | $2m+ |\n|---|---|---|---|\n| 4212 | Floor Coverings Retailing | 2,005 | 614 |\n| 3731 | Furniture & Floor Covering Wholesaling | 1,823 | 475 |\n| 3339 | Other Hardware Goods Wholesaling (incl. ceramic tile) | 5,361 | 1,442 |\n| 4231 | Hardware & Building Supplies Retailing | 5,880 | 1,652 |\n| 4229 | Other Electrical & Electronic Retailing (incl. light fittings) | 1,700 | 328 |\n| 3494 | Other Electrical & Electronic Wholesaling | 4,379 | 1,285 |\n| 3243 | Tiling and Carpeting Services | 20,560 | — |\n| 2029 | Other Ceramic Product Manufacturing | 445 | — |\n| 2432 | Electric Lighting Equipment Mfg | 378 | — |\n\n**IBISWorld corroboration:** Floor Covering Retailing AUD 3.7bn / 2,049 businesses; Furniture & Floor Covering Wholesaling 6.6bn / 1,832; Electrical & Lighting Stores 2.8bn / 1,802; Ceramic Product Manufacturing 791.5m / 452 — with **\"imports accounting for a high share of industry revenue\"**. The AU tile trade is an **import** trade.\n\n**Addressable: ~1,500–2,000 AU businesses at $2m+; ~1,800–2,450 including NZ; working number ~2,000.** Full universe including sub-$2m: ~6,000–9,000. **NSW leads wholesale and import, Victoria retail** — consistent with Sydney as the import gateway.\n\n**The organised supply side is tiny.** The **Australian Tile Council** (50+ years, \"Australia's only recognised tile industry representative\") claims members represent **~70% of all tiles sold**. Its live directory shows **140 members**: Tile Merchant 45, Contractor 31, Other 21, Consultant 14, Builder 11, **Wholesaler 6, Manufacturer 5**, Industry Supplier 4 — so the **entire organised supply side is ≈51 merchants and wholesalers plus 5 manufacturers**. Membership **AUD 150–330/yr**. By state: QLD 36, NSW 35, VIC 32, SA 28, WA 28. NZ: **New Zealand Tile Council 52 members**; TANZ ~197 listings (~100 genuine independents); **FloorNZ 314**; Lighting Council NZ 25.\n\n*(Correction: **ATTIA is the Australian Tea Tree Industry Association.** There is no \"Australian Tile and Terrazzo Industry Association\" — an earlier research brief invented it.)*\n\n### Per-retailer ecommerce platform audit\n\nMeasured by fetching each site and matching platform markers.\n\n- **Magento 2** (`data-mage-init`, `mage/cookies`): **Amber Tiles, National Tiles, Beaumont Tiles, Beacon Lighting**\n- **Shopify** (`cdn.shopify.com`): **TileCloud, Earp Bros, Cult Design**\n- **WooCommerce**: **Signorino, Tile Factory Outlet**\n- **No storefront detected** (catalogue/brochure only): **Artedomus, Academy Tiles, Di Lorenzo, Skheme, Surface Gallery, Living Edge**\n\nPattern: large AU chains run **Magento/Adobe Commerce**; newer DTC and mid-size importers run **Shopify**; WordPress/Woo covers those that grew out of a brochure site; several premium importers sell **entirely offline**.\n\n**Consequence: the Cin7-to-Shopify gap does not reach the biggest AU players — they are on Magento, where Cin7 also drops attributes. A Shopify-first product addresses a subset of a subset.**\n\n### What could NOT be established\n\n- **Not public:** GS1 NPC subscription; CodeMark and WaterMark certification; GBCA Responsible Products recognition; Global GreenTag; NATSPEC Product Partner fee; Archify, Masterspec, EBOSS and Productspec supplier programmes.\n- **Whether Australia has a BPIR-equivalent mandate in progress** — none found, but the web-search budget was exhausted before ABCB could be checked directly.\n- **Current state of ETIM Australia** — site unreachable; which of the four sectors it covers is not published.\n- **Whether Design Content (A2K) still operates** in any form.\n- **Lighting-specific data standards beyond ETIM.**\n\n## Verdict, and the one test that settles it\n\n### All three original moats are gone\n\n**Moat 1, the building-materials taxonomy — gone.** ETIM is an open international standard covering building materials, and ISO 13006 already standardises ceramic tile classification. Owning a taxonomy is not available; it is a free public good. (With the caveat established above: ETIM's tile class has real holes, and extending it is both unavoidable work and a possible asset.)\n\n**Moat 2, AI extraction from messy supplier formats — gone.** This is a commodity capability shipping in cheap Shopify apps and mid-market platforms. LLM improvement makes it cheaper every quarter, compressing margin rather than building advantage.\n\n**Moat 3, multi-platform publishing — gone.** Cin7 documents Shopify, Magento and WooCommerce integrations; generic PIMs treat channel syndication as table stakes; and an iPaaS tier moves custom fields with no purpose-built product at all.\n\n**The squeeze:** this sits in the valley between a cheap single-platform Shopify importer and a roughly USD 30k/year enterprise PIM. The buyer there — a showroom with a few thousand SKUs across twenty suppliers and two storefronts — is real but small, hard to reach, and low willingness to pay. Both sides are expanding into it: Shopify apps adding channels and PIM features, PIM vendors adding self-serve tiers.\n\n### Demand is now the binding constraint\n\nThis is the weakest part of the dossier, and it decides the idea.\n\nThe research searched specifically for demand evidence and found **no forum threads, no reviews, no vendor complaints** documenting anyone suffering from Cin7's inability to carry rich spec data to Shopify. The only signal is agency marketing copy. Nobody has publicly articulated this need.\n\nHold that alongside two other absences found independently:\n\n- Five years of ETIM permission for flooring, a 1,020-member merchant federation leading the sector, a dedicated Tegels commission — and no tile brand publishing data.\n- ETIM AU/NZ launched in 2021 with the Australian Industry Group behind it; its website is still a parked domain.\n\n**Three independent absences pointing the same way.** The optimistic reading is greenfield. The pessimistic reading is that structured tile spec data is something everyone agrees is nice and nobody pays for, because tile is chosen on appearance, price and availability, with compliance arriving once as a PDF.\n\nThe market is also small. Roughly 1,500–2,000 commercially serious Australian targets, about 2,000 with New Zealand. The organised Australian tile supply side is ~51 merchants and wholesalers plus 5 manufacturers. The biggest players run Magento, not Shopify, so the Cin7-to-Shopify gap does not even reach them. A Shopify-first product is a subset of a subset of a small market. **This has to be a global product with an Australian beachhead**, or not at all.\n\nAnd it is bigger than a connector. Because Cin7 Core caps attributes at 10 per set, 256 characters, one set per product, and gates them behind a higher plan, this cannot be \"read Cin7, write Shopify metafields\". The enrichment layer must own its own datastore and use Cin7 only for SKU, stock and price identity — materially more product than a sync tool, which raises the bar for how strong the demand signal must be before building.\n\n### The first real willingness-to-pay evidence\n\nAgainst all that, one finding partly answers the demand risk. **API and PIM integration is the top-tier upsell at two independent vendors**: bimstore at GBP 4,800/yr Enterprise, and BIMstreamer at EUR 16,000/yr Advanced. Two unrelated companies both place programmatic product-data integration in their highest-priced tier. That is the strongest signal in the dossier that someone pays for exactly this capability — and it sits in the **specification channel, not ecommerce**. It does not prove tile importers will pay, but it refutes the flat reading that nobody monetises structured building-product data.\n\nConcora Spec is now **Archbase**, and it publishes real pricing: **from USD 1,000/month**. That gives the spec channel its only public per-manufacturer price point.\n\n### What to do next, in order\n\n**First, the cheapest disconfirming test**, which is a conversation rather than a search: ask a tile merchant and an architect which fields they query per-SKU versus accept as a datasheet. If the answer is \"appearance and availability, plus a datasheet on request\", the standards work is a cost, not a moat, and the compliance-grade positioning collapses.\n\n**Second, resolve the load-bearing technical unknown:** request BIMobject partner API access and attempt an authenticated publish. If no spec library accepts programmatic ingest, the specification pivot is a services business by construction.\n\n**Third, if both survive, sell the service before the software.** Given the origin was a digital marketing strategy conversation, the fastest honest monetisation is a productised data-onboarding service for building-product suppliers, delivered with off-the-shelf tools rather than a new platform: take the supplier's website, ERP export and PDFs, produce a clean normalised catalogue, publish to their storefronts, charge per catalogue or per SKU. Zero build risk, immediate revenue, and it directly answers the questions this research cannot — real cadence, real willingness to pay, how many storefronts, how clean the SKU join is. It also produces the supplier-format corpus that would be the only compounding asset if a product version ever justified itself.\n\nOne pattern worth stealing whichever way it goes: SKU Launch replaces spreadsheet collection with a portal where suppliers **confirm pre-filled data** rather than typing, claiming completion rises from about 40% to over 80%. That is the same confirm-not-type principle this research reached independently for the review queue — evidence the pattern works, and that a competitor already applies it supplier-side.\n\nTwo distribution notes for whoever builds. The Tile Council of North America has over 200 member companies representing roughly 90% of ceramic tile production in North America and publishes the 2026 TCNA Handbook — a body with that concentration is a plausible partner or a plausible source of a competing data initiative. And Shopify lacks native functionality for sales reps to take orders at trade shows or in the field, a documented gap versus purpose-built B2B wholesale platforms; for a supplier selling to architects and trade, the storefront does not cover how the sale actually happens.\n\n## Method, confidence, and what was never checked\n\nAll research was conducted on 29–30 July 2026 by web search, enumeration and direct fetching. **No primary interviews, no vendor demos, no trials, no pricing calls, no product testing.** One exception: the Matrixify correction came from the idea owner's direct product knowledge.\n\nThe first pass ran 23 queries, grouped: competitors (7), standards (4), inference (2), channels (2), integration (1), open source (4), form factor (3). Later passes added enumeration work — the full 2BA data-supplier list, Forbo's 96 child sitemaps, the ETIM viewer's 5,680-URL sitemap, the ATC member directory, and platform-marker fetches against named Australian suppliers.\n\n### Confidence register\n\n**Verified from primary or vendor sources.** ETIM licence and API (etim-international.com); bSDD contents by direct enumeration; Cin7 integration behaviour (help.core.cin7.com, help.omni.cin7.com); Shopify metaobject limits (shopify.dev); Office.js capability (learn.microsoft.com); Anguleris acquisition of Concora; ISO 13006 existence and scope (iso.org); ABS business counts (8165.0); EPD Australasia and Eco Choice Aotearoa fee schedules; ETIM AU/NZ parked domain by direct fetch.\n\n**Vendor marketing claims, unverified.** SKU Launch's 40%-to-80%+ completion lift; Shopif-AI's 300-page capability; MindHarbor's normalisation claim; every Shopify app's field-mapping claim; Lucidworks' conversion-lift and revenue figures. All self-reported.\n\n**Third-party aggregator, medium confidence.** Pimberly USD 30,000/yr (Capterra); Akeneo CE feature freeze and end-of-life September 2026; Akeneo CE running cost EUR 1,500–3,000/month; PDF parser benchmarks; embedded-versus-standalone conversion rates of 70–85% and 45–60%; Material Bank's \"$25 per order\".\n\n**Refuted during research.** That vision-enrichment vendors do not serve building materials — this was recorded as the single most load-bearing and least verified claim, was checked adversarially, and was broken. It has been replaced with the narrower surviving claim about the unjoined intersection.\n\n**Flagged unverified.** A claimed list of tile-chemistry ETIM adopters (Mapei, Ardex, Weber, Sopro, PCI) conflicts with the 2BA enumeration. Do not rely on it. BIMobject's write API, as described above.\n\n### Not researched, therefore unknown\n\nLighting-specific data standards beyond ETIM. Magento and WooCommerce API detail — only Shopify's publish path was examined, so the multi-platform claim rests on one platform's mechanics. Competitor revenue, funding, headcount or customer counts. Customer reviews or satisfaction data for any named vendor. Legal and privacy implications of crawling supplier websites. Pricing for Sales Layer, SKU Launch, MindHarbor or any named Shopify app. Material Bank's rate card, the Archello / Archiproducts / Archify brand programmes, GS1 National Product Catalogue subscription, CodeMark and WaterMark certification, GBCA Responsible Products recognition, Global GreenTag, and BIMobject's tiers all remain unpublished — do not model revenue on them.\n\n### Link audit\n\nEvery URL in the body and the source index — 159 in total — was fetched and status-checked on 30 July 2026.\n\nOne genuinely broken link was found: `concora.com/2025/02/11/anguleris-acquires-concora/` returns 404. The acquisition is real but that URL is gone; it has been replaced with the PR Newswire release, and engineering.com and finsmes.com carry live coverage. A correction with it: the announcement date is **18 February 2025, not 11 February**.\n\nEvery other apparent breakage was a transcription fault in the source index rather than a dead page — a dropped closing parenthesis on the ETIM Wikipedia URL, missing `www.` on two hosts, and three URLs truncated with an ellipsis. All verified 200 once written correctly.","body_view":"full","sections":[{"id":"where-this-came-from-and-what-it-actually-is","title":"Where this came from, and what it actually is","words":583,"verdict":"thin"},{"id":"the-category-is-occupied-at-every-price-tier","title":"The category is occupied at every price tier","words":1129,"verdict":"ok"},{"id":"competitors","title":"Competitors and integration — the field-level evidence","words":2070,"verdict":"ok"},{"id":"the-standards-layer-measured-rather-than-assumed","title":"The standards layer, measured rather than assumed","words":1351,"verdict":"ok"},{"id":"standards-layer-the-full-etim-measurements","title":"Standards layer — the full ETIM measurements","words":1436,"verdict":"ok"},{"id":"filling-the-gaps-derivation-inference-and-the-line-that-must-not","title":"Filling the gaps: derivation, inference, and the line that must not be crossed","words":1310,"verdict":"ok"},{"id":"the-review-surface-is-the-product","title":"The review surface is the product","words":1224,"verdict":"ok"},{"id":"where-normalised-data-could-actually-go","title":"Where normalised data could actually go","words":1337,"verdict":"ok"},{"id":"specification-channel-the-14-platform-feasibility-table","title":"Specification channel — the 14-platform feasibility table","words":1663,"verdict":"ok"},{"id":"the-australian-market-sized-from-primary-data","title":"The Australian market, sized from primary data","words":950,"verdict":"ok"},{"id":"regulation","title":"AU/NZ — regulation, certification economics and bottom-up sizing","words":2208,"verdict":"ok"},{"id":"verdict-and-the-one-test-that-settles-it","title":"Verdict, and the one test that settles it","words":1002,"verdict":"ok"},{"id":"method-confidence-and-what-was-never-checked","title":"Method, confidence, and what was never checked","words":518,"verdict":"thin"}],"usage":{"chars":116185,"chars_remaining":883815,"chapters":13,"chapters_remaining":87,"below_floor":0,"above_ceiling":0},"url":"/ideas/gapfill/"}