Research method
Market Thesis Stress Test
- Explore
- Focus
Reading a whole category's public material with an AI model: competitor sites and pricing, what incumbents tell investors on earnings calls, and what customers say in reviews and forums. The model is asked questions that force thinking rather than summary, then one go-to-market angle is attacked over several rounds, each bringing new evidence, until what is left holds up. The output is a hypothesis to test with buyers, not a finding.
When to reach for it
Get up to speed on an unfamiliar category before a research engagement or a pitch.
Find the beliefs a whole category takes for granted, and spot the ones that may already be false.
Turn a hunch about positioning into a sharp hypothesis worth a message test.
Check whether "nobody does this" is true before building a strategy on it.
Often paired with Competitor Review.
Pros
Days of desk research shrink to a working session, using material that is already public.
Every claim traces back to a logged source, so the reasoning can be checked.
Forcing each round to bring new, contrary evidence catches weak angles that a single reading lets through.
No recruiting, scheduling or incentives.
Cons
It is only as good as the material fed in. Review sites and forums often block automated reading, so exports are usually needed.
AI models tend to agree with the angle they are handed unless the process forces them to attack it.
Public material shows what vendors claim and what some customers post, not how buyers actually decide.
No buyer is studied, so it cannot validate a decision.
Practical considerations: sample sizes, tools, logistics
Practical considerations
Set a floor
Agree a minimum before the analysis starts, for example eight competitor sites and 30 full reviews. Below it, label the output a desk scan and treat every pattern as a signal, not a count.
Count real voices only
A mention is one reviewer's own words. Review-site summaries, search snippets and vendor comparison pages are not customer voice.
Look at home first
Read what the company already publishes (ROI studies, case studies, analyst reports) before claiming a gap in the market.
Attack with new evidence
Each round must bring a source not already used, starting with "does anyone already do this?". A round that only reshuffles the same evidence does not count.
Pair it
Always follow with Copy Testing or Customer Interviews aimed at confirming or killing the surviving angle.
Never in isolation
Recipes that use Market Thesis Stress Test
A method earns its value when its signal corroborates others. Here is where this one does its work.
Last updated
.svg)
