This morning a father posted on LinkedIn asking the startup community for advice on behalf of his son. The young founder has completed over 150 pages of market research, distilled it into a focused 20-page investor document, and is preparing for tax-relief investment discussions. The question put to the community was which build route to choose: hire a development team, white-label an existing platform, or build with AI assistance.
It is a loving post and a serious effort. And in the whole description of the journey, from research pack to investor document to build route, there is no mention of a user. Nobody in the story has reacted to the idea except the people who wrote it.
Why does research volume feel like progress?
Because it is controllable. A founder posted something similar in a SaaS community this week: improving the product is easier than finding out if anyone wants it, because building is something you control and a customer conversation is not. Research has the same property. Page 151 is always available. The document never says no, never ghosts you, never asks what it costs. It answers to you.
A stranger answers to nobody. That is precisely what makes their reaction evidence and the document's contents description.
What can desk research never tell you?
It can size a market, map competitors, and estimate pricing bands. What it cannot do is produce a single observed behaviour. Whether a specific person with the problem will change what they do, spend money, or even take a twenty minute call is not in any report, because it has not happened yet. Worse, desk research often cites other desk research, so confidence compounds while evidence stays flat. We have called this the citation loop before: ten sources repeating one assumption look like ten confirmations.
What does one stranger change?
One stranger with the problem, reacting to a concrete offer, produces the first data point the 150 pages do not contain: a reaction that costs you something to hear. Maybe they say the problem is real but handled. Maybe they ask the price and go quiet. Maybe they say yes and then do not show up, which is its own answer. Three strangers saying the same unprompted thing is a pattern. This week a founder watched three agency owners independently tell him why they would not pay 799 dollars a month for his product, and his business model changed in a day. No report would have done that.
How do you turn a research pack into evidence?
The pack is not wasted. It tells you who should have the problem and what the alternatives cost, which is exactly what you need to find the right strangers and ask non-leading questions. The conversion is simple to describe: every claim in the document that matters gets a name attached to it, a real person whose behaviour confirmed or broke it. Claims with no name stay marked as assumptions, however many citations they carry. An investor reading 20 pages will eventually ask the same question the document avoided: who did you talk to, and what did they do?
Key takeaways
- Research volume measures effort. Demand is only measurable through people who owe you nothing.
- Documents cannot refuse you, so they cannot validate you either.
- Desk research compounds confidence, not evidence, when sources cite each other.
- Use the research to pick the right strangers and the right questions, then attach a real name and a real behaviour to every claim that matters.
Page 151 will not save the idea. The first stranger might.