How to Do Market Research With ChatGPT (Step by Step + Prompts)

You have an idea for a store, a service, an app or a new offer, and before you jump in, you want to know if it actually has a shot at finding customers. Market research used to mean weeks of work or a consulting firm’s budget. Today, doing market research with ChatGPT lets you move much faster when it comes to structuring your thinking, drafting surveys and analyzing your data.

One common misunderstanding, though: ChatGPT doesn’t “know” your market. It doesn’t replace official numbers, interviews with real customers or what you see on the ground. What it does extremely well is organize the work. In this guide, I’ll walk you through how to do market research with ChatGPT step by step, with copy-and-paste prompts and the precautions you really shouldn’t skip.

Contents

1. What ChatGPT can (and can't) do for market research
2. Step 1: Describe the project and lay out your assumptions
3. Step 2: Define your target customers
4. Step 3: Prepare your surveys and interviews
5. Step 4: Analyze the competition
6. Step 5: Estimate the market size
7. Step 6: Analyze your data
8. Step 7: Write up the summary
9. Deep research: useful, but double-check it
10. Mistakes to avoid and limitations
11. My checklist before drawing conclusions
12. Frequently asked questions
13. The bottom line

What ChatGPT can (and can’t) do for market research

Market research tries to answer a handful of big questions: who are my potential customers, what problems do they have, how many of them are there, what are competitors already doing, what price can I charge, and which channels will reach them.

On all of these, ChatGPT acts as a methodology assistant. Here’s how I split the roles:

Task	Is ChatGPT useful?	What else you need
Framing the assumptions to test	Yes, very	Your knowledge of the project
Drafting a survey or interview guide	Yes, very	Real people to talk to
Listing possible competitors	Partly	Checking their websites and in person
Estimating market size	Only the calculation method	Numbers from official or industry sources
Analyzing survey responses	Yes	Real, reviewed data
Summarizing interview notes	Yes	Rereading the key passages
Telling you if the project will work	No	Tests, sales, customer feedback

In other words, ChatGPT is your best ally to structure and analyze, but the raw information has to come from the real world. That’s exactly what OpenAI’s terms of use say: answers aren’t always accurate, you should evaluate them before relying on them, and they shouldn’t be the only basis for an important decision.

Step 1: Describe the project and lay out your assumptions

It all starts with a clear description. Take Marcus, who’s thinking about opening an e-bike repair shop in a mid-sized city. Before talking to anyone, he asks ChatGPT to help him pin down what he needs to verify.

I'm doing market research for this project: [description of the product or service, location, target customers, planned price]. Help me lay out the key assumptions I need to verify before launching: about customers, their needs, competitors, pricing and sales channels. Rank them from riskiest to least risky.

What you get back is a list of assumptions like “e-bike owners in town struggle to get their bikes repaired quickly” or “they’d be willing to pay for a yearly maintenance plan.” Each assumption then becomes a question to check.

To push further, ask it to play devil’s advocate:

Act as a skeptical investor. What are the 10 reasons this project could fail? For each one, tell me how I could check whether the risk is real.

Step 2: Define your target customers

ChatGPT is very good at suggesting customer segments and typical profiles (often called “personas”). Just keep in mind that these are assumptions to confirm.

Based on this project, suggest 4 potential customer segments. For each one: who they are, their main problem, what they use today instead, what could win them over and what could hold them back. Put it all in a table.

Then:

For each segment, what specific information should I look for to confirm it really exists and is big enough? Tell me what kinds of sources I could check.

That second request matters: it turns imagined profiles into a concrete checklist.

Step 3: Prepare your surveys and interviews

In my opinion, this is ChatGPT’s most valuable use in market research. A good survey is hard to write: you have to avoid leading questions, vague wording and questions everyone answers “yes” to.

The online survey

Write a survey of no more than 12 questions to test these assumptions: [list]. Focus on past behavior rather than future intentions. Mix closed questions with 2 open-ended ones. Avoid leading questions and flag any that could bias the answers.

The interview guide

One-on-one interviews often teach you more than surveys, especially early on.

Prepare a 20-minute interview guide for talking to [profile]. I want to understand how they currently handle [problem], what frustrates them, what they've already tried and what they've paid for it. Don't mention my project until the end of the interview. Add follow-up questions.

Review an existing survey

Here's my survey: [text]. Review it like a market research expert: spot leading, ambiguous or pointless questions, and suggest a corrected version.

One important point: if you collect responses, tell participants clearly how their answers will be used and get their consent. And don’t ask for personal information you don’t need.

Step 4: Analyze the competition

ChatGPT can help you organize a competitive analysis, but don’t take its list of competitors at face value: it can miss some, make some up, or name businesses that have closed. Use its web search and check every name, or better yet, give it the list yourself.

Here are the competitors I've identified and what I've seen on their websites, in their customer reviews and in their pricing: [notes]. Build a comparison table: offer, listed prices, target customers, strengths, weaknesses, recurring review themes. Then point out the gaps where no competitor seems to meet the need well.

For customer reviews, you can copy public reviews (without names) and ask for an analysis:

Here are customer reviews posted about competitors: [reviews]. Group the complaints and compliments by theme, from most to least frequent, and quote one example for each theme.

Step 5: Estimate the market size

This is the step where you need to be most careful. If you ask “how big is the bike repair market in the US,” ChatGPT may give you a number that sounds precise but could be wrong or outdated. It isn’t a source.

On the other hand, it’s very useful for building a calculation method that you’ll feed with verified numbers yourself:

Help me build an estimate of my local market size using a bottom-up approach. List the data I need (number of potential customers, purchase frequency, average order value…), tell me where I could find it, and set up the formula. Don't give any numbers yourself: leave blanks for me to fill in.

That last instruction keeps it from inventing numbers. You can then build the table in a spreadsheet.

Step 6: Analyze your data

Once your surveys are filled out and your interviews are done, ChatGPT becomes an excellent analyst, as long as you give it the data.

Survey responses

Export the responses as a CSV or Excel file and upload it to the chat. I explain the method in detail in my guide to analyzing a CSV file with ChatGPT.

Here are the responses to my survey. First describe the file: number of responses, columns, missing values. Then calculate the breakdown of answers for each closed question, compare the results by segment, and summarize the open-ended answers by theme with their frequency. Flag any limitations: sample too small, inconsistent answers.

Then double-check a few calculations by hand or in a spreadsheet: misreading a column happens.

Interview notes

Here are my notes from 8 interviews: [notes]. For each assumption on my list, tell me whether the interviews confirm it, contradict it or don't allow a conclusion, with the quotes that support it.

Asking for quotes is essential: it lets you check that the summary is faithful and doesn’t over-interpret.

Step 7: Write up the summary

Finally, ChatGPT can help you put your conclusions into shape, whether for yourself, a business partner or a banker.

Using everything above, write a one-page summary of my market research: the need identified, priority targets, competition, the market estimate (with the sources I gave you), risks and next steps. Clearly separate what has been verified from what is still an assumption.

That last sentence makes all the difference: a good market research summary is honest about what you don’t know yet. To turn these results into a sales action plan, read my guide on using ChatGPT for marketing, and to prepare your first content, my article on finding content ideas with AI.

Deep research: useful, but double-check it

ChatGPT Plus includes deep research, when it’s available. This feature goes through many web pages and produces a report with sources. In my experience, it’s a good starting point for desk research: finding public studies, trade press articles and key players in the industry.

But the report is still an automated summary: open the cited sources, check the numbers and their dates, and be wary of any figure without a source. Usage limits can also vary.

Mistakes to avoid and limitations

Treating ChatGPT’s numbers as data

This is the most serious mistake. A made-up or outdated market figure can throw off your entire project. Every number in your research should have a source you’ve checked yourself.

Replacing real customers with simulated ones

Asking ChatGPT to “play a customer” can help you prepare an interview, but its answers aren’t what real customers think. Nothing beats ten conversations with actual people.

Looking for confirmation

If you describe your project enthusiastically, the AI will tend to agree with you. Explicitly ask it for objections, risks and reasons it could fail.

Sharing personal or confidential data

Before uploading survey responses, remove names, emails, phone numbers and anything that could identify someone. Also make sure participants agreed to how their answers would be used, and follow your company’s rules for internal documents.

Deciding based on AI alone

OpenAI is clear about it: its answers shouldn’t be the only basis for an important decision. Starting a business, investing or taking out a loan all count. Have your research reviewed by someone qualified: a small business advisor, an accountant or someone who works in the industry.

My checklist before drawing conclusions

• My assumptions are written down and ranked by risk.
• I’ve talked to real people in my target market.
• Every competitor in my table has been verified.
• Every number has a source I’ve checked, with a date.
• I’ve checked a few of the analysis calculations myself.
• My summary separates what’s verified from what’s still an assumption.

Frequently asked questions

Can you do market research with ChatGPT for free?

Yes, for the most part: framing assumptions, drafting a survey, summarizing notes. The free version also lets you upload files, with limits. ChatGPT Plus ($20 a month) gives you more access, including file analysis and deep research when it’s available. See my article on ChatGPT Free.

Can ChatGPT tell you the size of a market?

It can suggest a number, but that number may be wrong or outdated. Use it instead to build the calculation method, and feed it with numbers from verifiable sources.

Is market research done with ChatGPT good enough for a bank?

Not as is. A lender expects sourced data and real-world evidence: interviews, surveys, first customers. ChatGPT helps you organize and present them clearly, not replace them.

Can you analyze survey responses with ChatGPT?

Yes, it’s one of its best uses. Upload a CSV or Excel file with no personal data, ask for a description of the file and then the analysis, and check a few results yourself.

Can ChatGPT find my competitors?

It can suggest leads, especially with web search, but it can miss some or make some up. Check every name and add your own research and local observations.

How long does market research with AI take?

It depends on the project, but AI mainly speeds up preparation and analysis. The part you can’t compress is the fieldwork: finding people to interview and collecting responses.

The bottom line

To do market research with ChatGPT: use it to frame your assumptions, draft your surveys, organize your competitive analysis, build your calculations and summarize your data, but feed it real information and verify every number. ChatGPT structures and analyzes; the fieldwork and the sources decide.

To go further, check out my guide to writing a good prompt.