Market research is one of the most important things a founder can do and one of the easiest to deprioritise.
It takes too long. The output is too generic. By the time the research is done the question has changed.
AI doesn’t fix all of that. But it compresses weeks of work into hours — if you know how to prompt it correctly.
Here’s the research process I use before any major product or market decision.
Start With the Job to Be Done
Before any research prompt, write down the specific decision the research is supposed to inform. Not “understand our market” — that’s too vague to produce useful output. Something like: “Should we prioritise enterprise or SMB for the next 12 months?” or “Is the compliance pain point strong enough to build a standalone product around?”
The more specific the decision, the more useful the research.
The Competitive Landscape Prompt
“I am building [describe your product and target customer]. I want to understand the competitive landscape. Give me: (1) The 5 most direct competitors and what each one does well and poorly, (2) The 3 most common reasons customers switch away from each competitor, (3) The gaps in the market that none of the current solutions adequately address, (4) What a new entrant would need to do to displace the current leaders. Be specific — I want analysis, not a list of features.”
This gives you a starting framework. Verify the specific claims against real sources — AI can hallucinate competitor details — but the structural analysis is usually sound.
The Customer Persona Prompt
“My product is [description]. My target customer is [describe them]. Based on this, give me a detailed customer persona that includes: their primary job responsibilities, the metrics they’re measured on at work, the problems that keep them up at night, what a bad day looks like for them, how they currently solve the problem my product addresses, what they’ve tried that hasn’t worked, and what would make them switch from their current solution. Be specific to [industry/sector]. Not generic — I can read generic personas anywhere.”
The Pricing Research Prompt
“I am pricing a [describe product] for [target customer]. Comparable products in this space charge [any pricing you know about]. Help me think through: (1) What pricing model is most common in this category and why, (2) What the key value metric is — i.e., what unit of value should pricing scale with, (3) What price points would be psychologically comfortable vs uncomfortable for this buyer, (4) What my pricing signals about my positioning — premium, value, enterprise — and whether that’s consistent with how I want to be perceived. Be specific.”
The Customer Interview Synthesis Prompt
After running customer interviews, use this to synthesise the patterns:
“I have completed [number] customer interviews. Here are the key quotes and observations from each: [paste your notes]. Analyse this data and give me: (1) The 3 strongest patterns across all interviews, (2) The most surprising finding — something that challenges my assumptions, (3) The pain point that customers described most viscerally — i.e., with the most emotion, (4) The jobs-to-be-done — what are they actually hiring a solution like mine to do, (5) Any red flags about assumptions I’m making that the interviews don’t support. Be direct — I want the uncomfortable findings, not just validation.”
The Market Size Sanity Check Prompt
“I am estimating the market size for [describe your product and target market]. My current estimate is [your number]. Walk me through a bottom-up market sizing calculation to verify or challenge this. Start with: how many potential customers exist, what percentage could realistically be reached, what ACV makes sense for this customer, and what market share is realistic in 5 years for a well-executed startup. Show your assumptions clearly so I can challenge them.”
The Research That AI Can’t Do
AI can synthesise, structure, and analyse. It cannot replace the 15 customer conversations you need to have before making a major product decision. The prompts above are a starting point and a framework — they’re not a substitute for real conversations with real customers.
Use AI to prepare better questions for those conversations, to synthesise what you hear faster, and to identify patterns across interviews. Use your own judgment for anything that requires understanding human motivation at the level that only direct conversation produces.
Get the Full Research Prompt Library
50 prompts for market research, customer discovery, competitive analysis, and more — free download.