ChatGPT can do more than draft a survey. Connected to User Intuition, it can plan a study, recruit real participants, have an AI moderator interview them, and hand you the findings with every quote traceable to the person who said it. This guide covers what is specific to ChatGPT: installing the plugin, how you confirm a launch in the chat, and how to prompt ChatGPT so the work runs in User Intuition rather than only in the chat.
How do you connect User Intuition to ChatGPT?
User Intuition is in the ChatGPT plugin directory, so setup takes one click and a sign-in:
- Open the User Intuition plugin in the ChatGPT plugin directory and install it.
- Sign in with your User Intuition account when ChatGPT asks. If you are new, create a free account first.
- Start a chat and ask in plain language: “List my User Intuition studies.”
No developer mode, connector setup or server URL is needed. On a ChatGPT Business, Enterprise or Edu workspace, a workspace admin may need to enable User Intuition first under Admin console → Plugins. Once it is enabled, members install it and sign in as usual. The ChatGPT integration page summarizes what the plugin can do, and the MCP quickstart covers setup in other clients. To connect Claude, Codex or Cursor instead, see how to connect AI agents to consumer research via MCP.
Fallback: a custom MCP server. On Business and Enterprise workspaces you can still add User Intuition by hand. Open Plugins, select +, choose Add custom MCP server, enter https://mcp.userintuition.ai/mcp and sign in with OAuth. A stakeholder who only reads findings can use https://mcp.userintuition.ai/read-only/mcp instead, which searches studies and opens reports without creating or changing anything. For most teams, the plugin directory listing is the simpler route.
What does a study run look like in ChatGPT?
Under the hood, ChatGPT calls the same research tools as any other client, and the MCP research API lists them by job. What differs in ChatGPT is how you ask and how you confirm. The run below is for a granola brand whose sales data shows buyers drifting away after a spring price change.
You: Use User Intuition to find out why some of our granola buyers stopped buying after the spring price change. US shoppers who bought it at least twice last year and not in the past three months.
ChatGPT asks how to recruit. Lapsed buyers on your own customer list fit a study that invites them by email; shoppers who left the brand without ever joining a list fit the panel, recruited in the United States with a screener on past granola purchases. You pick the panel. ChatGPT creates the study, passes your request to the User Intuition planner and relays its questions, such as whether a switch to a store brand counts as lapsing.
The saved plan in the chat. ChatGPT shows the complete saved plan as text: objectives, screener, and a guide that has each participant reconstruct the last time they bought the granola and the first trip they didn’t, then ladders from what changed to why to how it felt. Ask for edits in plain words and ChatGPT shows the new saved version.
Confirming in the conversation. For a panel study, ChatGPT requests an estimate and lists it as readable amounts: the total cost, your balance and the expected timeline. It asks for an explicit confirmation and launches after you reply, unless a workspace owner has set a monthly panel budget, in which case one-time launches within it go ahead without asking. If the plan or the country changes after that, ChatGPT comes back with a fresh estimate and asks again. For your own customers, it asks you to approve the saved plan before it creates the invitations.
Fielding. In this study the moderator interviews each participant by voice, following their story rather than a fixed script and laddering 5–7 levels deep. Your team can watch interviews as they happen in Live view, and you can ask for progress from this chat or from any other chat with the plugin available.
Results. Ask for the findings, and ChatGPT returns each one with a count of how many participants raised it and the verbatim quotes behind it, linked to the source interview and, for voice and video, to the moment in the recording. To see what a finished study looks like before you run one, explore an example study.
How do you prompt ChatGPT so the study runs in User Intuition?
ChatGPT will happily draft a discussion guide on its own, so a few habits keep the work inside User Intuition, where real participants answer it.
- Check the plugin is on. If ChatGPT answers from its own knowledge, confirm User Intuition is available in the chat before you describe the study.
- Name User Intuition in the request. “Use User Intuition to…” tells ChatGPT to call the research tools rather than answer from what it already knows.
- Say who and where. Name the recruiting method and the country up front, because a panel study needs both before it can be estimated.
- Ask for the saved plan first. The plan comes back from your User Intuition account, which confirms the study exists there and not only in the chat.
- Share concept images by link. For a pack or ad concept, give ChatGPT a public link to each image with the label participants should see, such as Pack A.
- Keep the quotation labels. When you ask for a summary, ask ChatGPT to keep the labels so each point traces back to its interview.
Prompts that follow those habits:
- “Use User Intuition to plan a study on why shoppers switch laundry detergent brands. Ask me how to recruit, then show me the saved plan and the estimate.”
- “Use User Intuition to test whether parents believe ‘no added sugar’ on our kids’ yogurt, and what proof they’d want to see. US panel, voice interviews.”
- “Use User Intuition to set up a concept test of the two pack images at these links, labelled Pack A and Pack B, on which pack shoppers would reach for first, and why.”
- “List my User Intuition studies and tell me which ones have reports ready.”
- “Get the report for the granola lapsed-buyer study, keep the quotation labels, and turn it into a one-page brief for the merchandising team.”
- “Use User Intuition to create a chat study for our own beta users about the new invoicing screen and give me the invite link.”
How do you read the findings in ChatGPT?
Read each count with its denominator. A share of participants and a share of friction moments count different things, so when a number is ambiguous, ask ChatGPT what it is out of.
Then ask for the evidence. “Show me the quotes behind the store-brand finding” returns the verbatims, each with its quotation label and source interview. For a voice study, open the recording at that moment and listen; tone and hesitation often explain more than the transcript does.
Ask “who disagreed, and why?” The participants outside the main pattern often hold the objection that matters most in a launch decision. If an earlier study may already hold the answer, have ChatGPT ask questions across past research before you commission a new one.
Finally, ask for the deliverable your team will use: a brief, a slide outline or a note to the category lead, with the quotation labels kept so anyone can trace a point back to the participant who made it.
What can you test from ChatGPT?
Pick the study type that matches the decision:
- Packaging and concept tests. Share two pack designs and ask which one shoppers would pick up first and why. The moderator follows up on what they noticed first, what the pack promised, and what would make them put it back.
- Claims and messages. Test whether “cut your grocery bill by 20%” sounds credible, and learn what proof would move people.
- Shopper and purchase journeys. Ask participants to reconstruct a recent purchase: how they chose the store, which brands they compared, and what nearly stopped them.
- Brand switching. Interview people who moved from one brand to another in the last six months about the trigger, the alternatives they weighed, and what would bring them back.
- Prototype tests. Give ChatGPT a live URL, such as a product-finder quiz or checkout flow, and participants work through it by voice or video while the moderator asks why they click, hesitate or leave.
- Product and app experiences. Invite your own users to talk through an onboarding flow or a new screen.
For a directional first read before real fieldwork, synthetic respondents are also available in chat.
How is a research plugin different from a CRM or sales-data connector?
Most ChatGPT connectors read systems of record. A CRM, commerce or loyalty connector can tell you which products a shopper bought, how often, and when they stopped buying a brand. It cannot tell you why they switched, because that reason was never recorded. A research plugin closes that gap: ChatGPT recruits people, User Intuition interviews them about the purchase and brand choices behind the data, and the answers return as cited evidence. The two work well side by side. Purchase history shows that repeat orders of a cereal fell after a price change; interviews with lapsed buyers show whether they traded down to a store brand, switched to a competitor with a better promotion, or simply stopped eating cereal at breakfast. Each of those answers points to a different response, which is why the reason matters as much as the record. Ask ChatGPT to pull both into one view, and keep the quotes attached.
Where does ChatGPT fit in a research workflow?
ChatGPT is the interface; User Intuition provides the participants, the moderation and the evidence trail. The agent shows you the saved plan and a cost estimate before anything is spent; in ChatGPT you confirm in the conversation, and in Claude you approve with one click. Participants come from a vetted 4M+ consumer panel (B2B audiences recruited to order per study), or your own participants, and User Intuition’s AI moderator runs 20–30 minute interviews that ladder 5–7 levels deep. Every panelist call is reviewed by the User Intuition team, panelists pass a video identity check, and only interviews that pass quality scoring are billed; see how quality works.
For agents beyond ChatGPT, or to build research into your own product, see the research API and MCP server for AI agents. AI-moderated in-depth interviews shows how each conversation gets to the reasons behind a choice.
Connect User Intuition in ChatGPT
New to User Intuition? Create a free account; signup includes 3 free interviews with your own participants, and panel recruiting is billed separately. Then install User Intuition from the ChatGPT plugin directory, sign in, and start a chat with one of the prompts above.