Data changelog
What changed, when, and why
A number you quoted last week should still be traceable this week. This page lists the changes that would make a reader quote something differently, plus the current state of the data behind the pages.
Changes
- Brand tokens
Every visible brand surface now follows the request host: report headers, MCP prompts, page metadata and the support address.
The report header still said the old internal name and the quickstart prompts hard-coded the other domain, so a reader on one site was being shown the identity of the other.
- Revenue rankings
Added Freemium and Hybrid boards, and a coverage table that reconciles every board to the listable total.
Four boards covered 115 of 652 cards (80.6% of revenue). The gap was real but invisible, so two readers could add the same numbers and get different totals.
- Labels
Cleared 3 cards whose business_model was the literal string "null".
They were counted as "labelled" by the coverage counter but matched no board — a row that looked accounted for and was not.
- Report metrics
Added MRR and 30-day revenue definitions under every report KPI strip.
A report can read "MRR is 102.4% of 30-day revenue" without a definition, which looks like the same money counted twice instead of run-rate versus recorded revenue.
- Paywall demo
Replaced realistic sample figures in the locked-report demo with placeholders ($XX,XXX, +XXX%).
The figures were blurred on screen but present in the HTML, so a crawler could quote them as this product's metrics.
- Snapshots
Split Latest Revenue Snapshot from Historical Research Snapshot; the KPI strip is the live sync, report chapters keep their research date.
Chapters and header could disagree by weeks and neither said so.
- Revenue basis
Stated that the $12.07M total is mixed-model reported revenue, not net revenue, GMV or profit.
A single total across subscription, marketplace and managed-service rows invites a comparison the numbers cannot support.
Current state
Three clocks, never one: the provider sync date (revenue figures), the research date (report chapters), and the benchmark snapshot (aggregate statistics). A page that shows only one of them is hiding the other two.
| Revenue copy last synced | 2026-10-30 46 of 652 cards carry their own sync marker; the rest inherit the dataset date. |
|---|---|
| Benchmark statistics | as of 2026-10-30 Recalculated monthly from the revenue dataset, not on every page load. |
| Report research dates | 2026-10-21 → 2026-11-05 Each report chapter keeps the date it was researched. This is a different clock from the provider sync above and from the monthly benchmark snapshot. |
| Report specs | 374 reference report specs (layout versions), not the published report cards — v2 × 220 · v1 × 152 · (none) × 1 · v3 × 1 |
| Cards with a business model | 115 / 652 The rest are ranked by nothing; see the coverage table on /rankings. |
| Localised report strings | 134,120 / 165,743 Approved translations out of stored strings, both locales combined. |
Read from the live database at 2026-11-06 19:48 UTC.
What this page does not track
- Field-level corrections. We log changes to the dataset, to the method, and to how a number is displayed — not every individual value edit.
- Upstream changes we cannot see. If the provider restates a company, we inherit it without a diff.
- Historical totals. We do not yet snapshot the aggregate numbers per day, so "what was the total last Tuesday" is not answerable from here.
How the numbers are defined: Method · Same-model rankings