8. Consolidated priority actions — all three audits
Tier 1 — conclusions that are wrong, not just loosely worded.
| # | Action | Source |
|---|---|---|
| 1 | Delete or source the Mobeus/Bidco/Topco PE claim in 00 §10 — no file supports it; 06 records it as NOT FOUND |
S-2 |
| 2 | Re-derive 00 §2's "a van needs 6–8 riders" — it is evidence against one van serving 18–24, and restore 02's dropped "three times that many" warning |
S-3 |
| 3 | Restore the search-budget caveat to 00 and requalify every universal negative as "not advertised in this sample" |
S-1, §5.6 |
| 4 | Add Escapegoat (A$175 self-guided) and TrailHopper (A$90) to the competitor table — they sit inside the claimed gap | S-5 |
| 5 | Pull the Vine Hopper "one van, three products" model from 00 §13 and stop using its "75-min headway" as the loop sanity check |
D-4, M-h |
| 6 | Resolve 00 §3 vs §12 on response-time SLAs — pick one; the recommendation depends on it |
S-6 |
| 7 | Rewrite the FareHarbor conclusion — source says a 2026 signup would not pay 6%; "Australia is in scope" unsourced | C-1 |
| 8 | Withdraw the Camino load-factor proof — volume discount vs a "from" price; corpus contains its own counter-example | C-2 |
| 9 | Drop the Wine Australia 29% from the winery pitch — $1.2m floor, capped, rationed, closed 2026, paused 2028–30, and it is 29% of wholesale | C-4, S-7 |
| 10 | Fix the TrailHopper comparison — like-for-like A$189.95 vs A$189, the same price, not half | X-2 |
| 11 | Re-cut the Escapegoat A$75–275 "value of the vehicle" — it prices a guide + Adelaide transfer | C-3 |
Tier 2 — evidence hygiene.
| # | Action |
|---|---|
| 12 | Narrow "nobody prices rescue" — the corpus lists four priced support services (D-2) |
| 13 | Reconcile the 35 number collisions in §7.2; start with delivery-price counts, wine-collection counts and the three "only platform" claims |
| 14 | Apply the tagging convention to 00 (currently one tag) and normalise 03's private vocabulary (M-g) |
| 15 | Correct seasonality: 6–7 months = 26–30 weekends, below the README's 30–35 (M-f) |
| 16 | Merge the duplicate Explore Marlborough entries; mark Sold (D-3) |
| 17 | Correct "four NZ wine-pickup operators" → three (D-1); promote Calistoga from [NOT FOUND] to verified (§7.2) |
| 18 | Reframe "own bike $80" as a transport fare (D-12); add Takaro's written over-indulgence pickup promise (D-13) |
| 19 | Re-tag Bayleys, advcollective, silence-as-verified, opinion-as-verified (D-6→D-11) |
| 20 | Strip numbers from the Viator and GetYourGuide rows; the downside case rests on a 2019 competitor blog (§3.4) |
| 21 | Delete the two self-falsified Gaps in 05 (C-5); add the e-bike denominator (M-e) |
| 22 | Sync the two top-level files with 05's corrections, or mark them superseded (M-i/j) |
| 23 | Add missed operators: Sonoma Valley Bike Tours, Franschhoek Wine Tram; check the majors' self-guided tiers (§5.5) |
| 24 | Verify SA s.47 demerit points at primary source (§2); fix the grossetwines.com https link (§1) |
The audit findings above were applied to the research in place. Disputed text was
struck through and annotated rather than deleted, so a reader can see which way each
error pointed — a correction is worth more than a clean page, because the direction of
the error is itself evidence.
The substance of every correction is stated in the sections above, each with the
evidence that overturned it. This section records only the one change that is not
visible anywhere else.
The one outright deletion
An earlier version of the synthesis asserted a named private-equity ownership chain
for Cycling for Softies — a specific investment firm, a holding structure, and a dated
statutory filing event.
No source in this research supports any of it. The research that was cited for it
records that company's ownership as [NOT FOUND], behind an HTTP 403.
It was deleted outright rather than struck through, because an unsourced named
allegation about a real firm should not remain legible on the page in any form. This
note is the permanent record that it existed and was removed.
Cycling for Softies' ownership and backing remain [NOT FOUND].
What the corrections did not reach
Two classes of problem were identified and not fixed:
- The synthesis was never re-tagged. It carries almost no evidence tags, while the
regional research carries 24–65 each. Where the synthesis states something flat,
check the tag on the underlying regional finding before relying on it. - Several number collisions remain open — among them Macs Adventure's founding year
(2003 vs a company-register date of 2009), the UTracks and Discover France e-bike
supplements, and the UK dissolution count. These are listed in the cross-file audit
above and were not reconciled.
The regional detail files are good and are safe to spend against on specific,
quoted numbers. All 288 URLs resolve, every price re-fetched today confirmed, the
arithmetic checks out, and the files argue honestly against their own business case.
00-synthesis.md is not safe to spend against. It drops the caveat, drops the
tags, contradicts itself on both load-bearing negatives, inverts its own headline
number, asserts an unsourced PE ownership chain, and omits the Australian
competitors that most directly threaten the pricing and the claimed gap.
Treat 00 as an index to the detail files, not as a conclusion. Its §1, §2, §4,
§5 and §12 should each be re-derived from source before anyone books a van, quotes
the 29% grant to a winery, or prints "never more than twenty minutes away."