How we vet market-research sources

The problem

Search for almost any market-size figure and you’ll find “syndicated” industry reports, usually priced around $3,000–$5,000, covering a narrow, oddly specific niche. Open one up and there’s often no disclosed methodology, no sample size, no interview count — a number with nothing underneath it, sold as if it were primary research.

A large share of these trace back to a small cluster of publishers who, by their own published output, put out hundreds to thousands of reports a year across every industry imaginable — not consistent with real primary research at that volume. Many operate under differently branded legal entities, some marketing themselves as US- or Europe-headquartered firms, so the connection between them isn’t obvious from the outside.

The part that actually causes damage: the same unverifiable number gets recycled across several of these “different” publishers. Three sources agreeing on a figure looks like corroboration — until you notice they all trace back to one original, unverifiable estimate. A fabricated number repeated three times isn’t three data points. It’s one, dressed up.

A large share trace back to a small cluster of publishers who, by their own published output, put out hundreds to thousands of reports a year.

What we do about it — in the product

Every AI call carries a source-quality guardrail. The reference search screens publishers and sets aside the ones it won’t stand behind, by name, never with their figures. Trust labels say what was checked (“methodology disclosed”), never that a figure is right.

See it live → run a reference search in the Market Size Calculator

What we actually do about it

Every real AI answer in our tools — not a canned response, an actual model call with web search — runs with a standing instruction to watch for this exact pattern before it cites or relies on a source:

Prefer sources with a name behind them — a named analyst or research team, a stated methodology, a publication date — over an unnamed “market research report” publisher.
Flag thin sources instead of dressing them up. A report abstract with a purchase price and nothing else is flagged as a “thin source,” not presented with the same weight as a properly sourced figure.
Check corroboration, not just repetition. If a figure shows up in multiple places, the AI is instructed to note whether those sources are independently derived or all trace back to one origin — repeated citation of one unsourced number is never presented as agreement.

This doesn’t mean a number gets hidden from you. It means a thin source no longer looks identical, at a glance, to a properly sourced one — same principle as the rest of this tool: the AI assembles and flags, you still decide.

What “methodology disclosed” does not mean. It doesn’t mean the figure is right, or that we checked it. Nobody has. It means the source is willing to show its working — a named analyst or team, a stated method, a date — which is the difference between a number you can argue with and a number you can only accept. A well-documented forecast can still be wrong, and a figure measured on a basis that doesn’t match your market is wrong for you however carefully it was produced. Follow the link and read it before you lean on it.

How to spot one yourself

Worth knowing even outside this tool — these are the same checks we’d run by hand.

Red flagNo named analyst or research team — only a publisher's brand name.

Red flagNo disclosed methodology: no sample size, no interview count, no stated primary source.

Red flagThe “full report” is paywalled at roughly $3,000–$5,000, and all you can see for free is a one-paragraph abstract.

Red flagThe publisher's own catalog lists an implausible volume of reports — hundreds or even thousands a year, across wildly unrelated industries.

Red flagThe exact same figure shows up, worded almost identically, across several “independent” publishers.

Red flagThe company markets itself as US- or Europe-headquartered, but its filings, registrations, or reviews trace the actual operation somewhere else entirely — a mismatch between the storefront and where the work is really done.

Good signA named analyst or research team you could look up.

Good signA stated methodology — survey size, interview count, or a named primary data source.

Good signA publication date, and enough detail to tell whether it's still current.

Good signPrimary or government data underneath it: census bureaus, labor statistics, SEC filings, World Bank / OECD / Eurostat.

Good signIf it's paid analyst research, a firm with a public track record — Gartner, Forrester, IDC, McKinsey, Bain, BCG, Euromonitor, Statista, IBISWorld and similar.

Sources

This page’s own screening rules, applied live in the reference search — the specific publisher lists stay internal, same line this piece draws throughout.

This is one piece of the same trust story behind every AI feature in this tool — you always see exactly what would be sent, and the math is never something the AI invents.

Try the Market Size Calculator →