Three years ago, a client walked into our office with a market feasibility study they had paid $68,000 for. It was 94 pages of charts, tables, forecasts, and competitor profiles. It looked professional. The branding was clean. The methodology section cited "proprietary datasets" and "machine learning-enhanced projections."
It was also mostly fabricated.
Not in the sense that someone sat down and invented numbers from scratch. It was subtler than that. The study combined real top-line figures from public databases with interpolated assumptions that had no basis in primary observation. The TAM calculation extrapolated from a single year of industry revenue and applied a growth rate borrowed from a different geographic region. The competitor profiles were compiled from company websites and press releases — no interviews, no calls, no validation.
The client had nearly committed $1.4 million in capital expenditure based on this study. When we ran our own ground-truthed assessment over the following 10 weeks, we found the real addressable market was about 40% of what the report claimed. Two of the five "key competitors" had exited the market entirely. And the growth rate was closer to 3% than the 11% the report projected.
This story is not unusual. It plays out in boardrooms every quarter. Here is how to avoid being the one holding the bad data.
Red Flag 1: No Named Interviews
The single fastest way to evaluate a market research deliverable: count the primary interviews. Not "we surveyed 500 respondents via online panel." Not "our team conducted qualitative research." I mean named, attributed conversations with identifiable market participants.
If your $80,000 study does not include a single conversation with someone who actually buys, specifies, or procures in your target market, you have paid for a literature review. Maybe a very polished one. But a literature review.
Look at the methodology section. If it says things like "secondary analysis of publicly available data" or "proprietary database compilation," that means nobody picked up a phone. The data is only as fresh as whatever database they downloaded it from.
Red Flag 2: Suspiciously Round Numbers
Real markets are messy. When a report tells you the market is "$5.0 billion" or growing at "exactly 8.5% CAGR," those are not measurements. Those are estimates that have been rounded, smoothed, and averaged into something that looks clean on a slide.
I had a client in the geophysical services space who received a report stating the California market for their services was "approximately $600 million." When we went out and mapped every active contract, every RFP in the pipeline, and every funded project across the state, the real figure was closer to $340 million in active spend. The $600 million number had included stalled projects, unfunded proposals, and a few contracts that had been completed years earlier.
Real numbers have decimals and caveats. When you see a number that is too clean, ask what is underneath it.
Red Flag 3: The "Global" TAM Trick
This one catches smart people all the time. A report says your total addressable market is $12 billion. Sounds enormous. But buried in footnote 14 on page 43, you discover that figure includes every country, every application, and every customer segment — including ones you will never serve.
Your actual serviceable market might be $400 million. Your realistic obtainable market in the first three years might be $20 million.
A credible study will give you the SAM (serviceable addressable market) and SOM (serviceable obtainable market) with clear definitions of how they got there. If the report only gives you the TAM and expects you to figure out the rest, they are selling you a big number to justify their fee.
Red Flag 4: Offshore Research Mills
This one makes people uncomfortable, but it is worth saying plainly. A significant portion of market research — including work commissioned by well-known firms — is produced by offshore research teams who have never visited the markets they are analyzing.
I have nothing against offshore talent. Some of the best data analysts I know are based outside the US. But if someone is writing a competitive intelligence report about the Pacific Northwest geotechnical services market from a desk in Hyderabad, they are missing context that you cannot get from a database. They do not know that two of those "competitors" are actually the same company operating under different DBAs. They do not know that the biggest contract in the pipeline is wired for an incumbent. They do not know that the regional trade association is dominated by three firms who share referrals.
That kind of ground-level intelligence only comes from being there. Or at minimum, from having spent years working in that market and maintaining current relationships.
Ask your research provider: Where is your team based? Have they worked in my target geography? Can they name three people in my target market they have spoken to in the last 90 days?
Red Flag 5: No Assumptions Table
Every market model rests on assumptions. Growth rate assumptions. Penetration rate assumptions. Pricing assumptions. Market share assumptions.
A trustworthy research firm will show you those assumptions explicitly. They will tell you: "We assumed 3% annual price increases based on the last five years of contract data." Or: "We assumed 12% market penetration based on comparable product launches in adjacent segments."
If the model is a black box — if you cannot see the inputs that produced the outputs — you cannot stress-test it. You cannot present it to a skeptical board member who asks, "What if the growth rate is half that?" You cannot do sensitivity analysis. You are trusting the number on faith.
Boardroom-defensible intelligence requires transparent assumptions. If your provider says their methodology is "proprietary," translate that as "we don't want you to see how the sausage is made."
What Good Data Looks Like
Good market data has fingerprints on it. You can trace the TAM calculation back to specific contracts, specific buyers, specific procurement cycles. You can see who was interviewed, when the conversations happened, and what those people actually said (even if anonymized for confidentiality).
Good data is uncomfortable sometimes. It tells you the market is smaller than you hoped. It identifies competitors you did not know about. It reveals that your pricing assumptions are 30% too high for the region.
Good data is specific. It says "The three largest buyers in this market are currently under contract with Company X through Q2 2026, creating a switching window in late 2025" — not "The market presents attractive opportunities for new entrants."
When we deliver a ground-truthed assessment to our clients, every number has a source. Every assumption is visible. Every recommendation traces back to an actual conversation with someone who matters in that market. That is the standard. Anything less is a guess dressed up in a nice font.