ChatGPT Prompts for Marketing Research: 12 From Your Data
12 ChatGPT prompts for marketing research: tie questions to a decision, mine your data, write unbiased surveys and interviews and set decision rules first.

ChatGPT prompts for marketing research are most useful when they start from a decision you have to make and the data you already have, then plan the cheapest research that can settle it. Give the model the business question, the decision, your existing data, how you reach customers, the alternatives and your budget. Below are 12 prompts for marketing research, from research questions to decision rules, plus a worked example for a bike shop deciding on e-bike rentals.
The marketing research prompt for a plan, survey and personas from your data returns research questions, what your data already answers, a method plan, an interview guide, a survey draft with bias notes, decision rules, persona and alternatives tables, a timeline and a budget.
Frame the research
1. Decision first
My business question is [question]. The decision it drives is [decision] by [date]. Write three to five research questions, each tied to that decision.
2. Mine the data you have
Here is data I already have [sources and counts]. Tell me which research questions it answers, what gaps remain, and how to sort it to learn more.
3. Method plan on a budget
With [budget] and [weeks], choose between desk research, interviews, a survey or a small test for these questions [questions]. Say who to recruit and how many.
Talk to customers
4. Interview guide
Write eight open interview questions about past behavior for [customer type] on [topic], plus two probes. No questions that pitch my idea.
5. Recruiting message
Write a short message inviting [group] to a [minutes] interview about [topic], with the thank-you [incentive], how answers will be used, and how to book.
6. Interview notes to themes
Here are notes from [number] interviews [paste]. Group what people did into themes, quote each theme, and keep what they said they would do separate.
Survey
7. Survey draft with bias notes
Draft a ten-question survey for [audience] on [topic] with a screener first and answer options for each, and flag any question that could bias answers.
8. Bias check
Review this survey [questions] for leading wording, double-barreled questions and missing answer options. Rewrite each problem question.
Analyze and decide
9. Decision rules up front
Before I collect data, propose go, change and stop rules for [decision] based on these questions [questions]. Mark every threshold as a proposal for me to approve.
10. Persona from real data
Build persona tables from this data only [data]. Leave cells blank and marked as gaps where the data says nothing.
11. Alternatives table
Build a table of the alternatives my customers use [list], with what each offers, its published price (or "check their page"), and why a customer might choose it over us.
12. Flagship: plan, survey and personas from data
BusinessQuestion [question], Decision [decision and date], ExistingData [sources with counts], CustomerAccess [how you reach customers], Alternatives [competitors and options], ResearchBudget [budget, weeks, people]. Return research questions, data gaps, a method plan, interview guide, survey, decision rules, tables, timeline and budget.
Worked example: should a bike shop rent e-bikes?
In the flagship example, a bike shop has 600 dollars, four weeks and a February decision. Before spending anything, the plan uses what the shop already has: 38 rental request emails, 9 reviews that mention the river trail and 140 e-bike repairs last year, and lists the gaps in each. It recruits 10 interviewees from people who asked about rentals and group riders without an e-bike, and runs a survey through the newsletter and in-store QR cards with at least 100 completes as the target. The decision rules (go, change to a visitor-only partnership, or stop) are written as proposals before any data comes in, and the budget spends 290 dollars, holding 310 for a follow-up test.
Variations
- Price change: set the decision to a new price and ask for questions about past purchases at different prices.
- New location: ask for a desk research plan on foot traffic sources you can check yourself.
- B2B: replace the survey with 12 buyer interviews and a short follow-up email question.
Tips
- Use the data you already have first. Support emails and reviews are research you have already paid for.
- Do not ask the model for market sizes or response rates; ask for ways to measure them.
- Price questions in surveys tend to run optimistic. Read them alongside what people actually paid before.
- Tell people how their answers will be used, and do not collect data you do not need.
- The bike shop is invented. PromptDig is not affiliated with OpenAI, Perplexity or Google.
Related on PromptDig
Need live web sources with citations? Use the Perplexity prompts for market research. For the open-ended survey answers, try the survey open-end codebook prompt and the ChatGPT prompts for survey analysis. More in the Marketing hub, browse all prompts, or share a research prompt that worked for you.