# Synthetic Audiences for Research Institutes | Minds Source: https://getminds.ai/for/research-partners Turn proprietary research, segments and methodology into reusable synthetic Audiences, and publish them to clients as attributed Marketplace add-ons. Drag with a mouse or use the arrow keys to rotate the audiences. Press Space to pause or resume motion. On touch screens, swipe left or right to rotate and up or down to scroll the page. # Turn years of research expertise into reusable AI audiences. Build reusable synthetic audiences from your research, methodology, segments, and accumulated expertise. Give clients access to pre- and post-survey exploration with Minds while extending, rather than replacing, your core research offering. ## From questions to decisions. - ### Sharpen the questionnaire before fieldwork Explore hypotheses against a synthetic version of the target population first, so the instrument you field is the one worth paying for. The real panel then answers a better questionnaire. Questionnaire - ### Turn your segments into reusable Audiences Bring your segmentation, methodology and respondent data in once, then query those segments as often as a client asks. Proposed distributions stay reviewable before anything is created. Segments - ### Give clients a post-survey deep dive When the report lands and the client asks the follow-up the questionnaire never covered, answer it the same week instead of scoping another wave for a single question. Follow-up - ### Publish an Audience under your own name Eligible partners can publish an Audience as a Marketplace add-on with visible attribution, and manage the plan-specific access and pricing that applies to it. Publishing Q1Qualitative Before we field this, what would you want the survey to ask? Wording is ambiguous Missing the obvious topic Too long to finish ### Wording is ambiguous 42.2% Two of these questions could be read either way, so I answered what I thought was meant. If someone else reads them differently, you are comparing two different things. Target population for the next wave 1 01/38 ### Missing the obvious topic 34.4% Nothing here asks about what it costs to keep using it, which is the part I actually talk about. That is the reason I would change, and it is not on the list. Target population for the next wave 39 01/31 ### Too long to finish 23.3% By the middle I was picking answers to get through it rather than thinking about them. Anything after that point I would not trust from myself. Target population for the next wave 70 01/21 ### Sharpen the questionnaire before fieldwork Explore hypotheses against a synthetic version of the target population first, so the instrument you field is the one worth paying for. The real panel then answers a better questionnaire. Q2Categorical Which everyday change would you make first? Trade up in quality Cut a recurring cost Save time in the week Change nothing for now Segment C 41 Trade up in quality was the preferred response in this illustrative result. ### Turn your segments into reusable Audiences Bring your segmentation, methodology and respondent data in once, then query those segments as often as a client asks. Proposed distributions stay reviewable before anything is created. Q3Multiselect Which of these findings would you want explained further? Why the segments split The one unexpected result What a question really meant What to ask next time - Why the segments split52% - The one unexpected result44% - What a question really meant36% - What to ask next time29% **Leading response:** Why the segments split52% ### Give clients a post-survey deep dive When the report lands and the client asks the follow-up the questionnaire never covered, answer it the same week instead of scoping another wave for a single question. Q4Numeric15 How useful would a ready-made audience like this be to your team? 12345 Ø4.1 - 1 - 2 - 3 - 4 - 5 **Average:** 4.1 ### Publish an Audience under your own name Eligible partners can publish an Audience as a Marketplace add-on with visible attribution, and manage the plan-specific access and pricing that applies to it. The Sinus-Milieus embody decades of research into the ways people live in our society. Together with Minds, we have transformed this knowledge into Milieu-Minds, making it directly usable as synthetic target audiences. We can finally run qualitative research at scale, without a high-cost survey absorbing the team's capacity for weeks. We ask a question and hear back from the audience in minutes, and the research we already invest in becomes far more valuable. For us as an agency, this fundamentally changes how we work with our clients. We can test ideas with different target audiences in minutes and gather rapid feedback on implementation approaches. It feels like having a dedicated panel for every client. ## How we validate our Results We compare synthetic audience responses with real survey data to measure how closely they reflect real-world response patterns. Minds does not see the real survey results while generating its answers, and its responses are finalized before the comparison begins. This lets us measure where the results match—and where they differ. ### Approximation to real survey distributions (e.g. GSS, ANES, CES and PISA) ### 93.99% similarity to real survey results 301 Minds answered a UK Food Standards Agency youth survey, and their combined response patterns closely matched the published data. ### Closer to how real people answered Compared with standard AI prompting, Minds was preferred 59% more often and matched held-back interview answers 68% more closely. Grounded in data from ## Public Audiences Ready-to-use synthetic Audiences, built with personas, demographics, behavioral context, and visible grounding so you can start a Study without rebuilding the sample. ## [Minds, where you work.](/guide/integrations) Bring files and designs into your research, use Minds from your AI assistant, and connect results to your reporting tools. Availability depends on your plan and setup. ## How an Audience gets built. - 01 ### You describe your audience Enter a prompt and optionally add files or other data for context. Include the market, interests, or habits that matter to your research.Enter a prompt and optionally add files or other data for context. - 02 ### Sources are researched and verified The web is searched for relevant, credible sources to ground the audience. These findings inform the audience characteristics and proposed distributions.The web is searched for relevant, credible sources to ground the audience. - 03 ### Distributions are extracted The evidence is translated into demographic, behavioral, and attitudinal distributions. They show how characteristics such as age, interests, and buying habits vary within the audience.The evidence is translated into demographic, behavioral, and attitudinal distributions. - 04 ### Individual Minds are created Each Mind is built using the confirmed distributions, collected sources, and relevant public social-media signals. Together, these profiles represent different perspectives within your audience.Each Mind is built using the confirmed distributions, collected sources, and relevant public social-media signals. - 05 ### Your audience is ready You can now run studies, evaluate ideas, and compare concepts. Reuse the same audience for follow-up research and explore how different segments respond.You can now run studies, evaluate ideas, and compare concepts. ## Want to see a demo tailored to your institute? Prof. Friedrich von Borries Research Lead Good research changes what you do next. Our job is to make that loop short enough that teams actually run it. ## Questions teams like yours ask The practical questions research institutes and data partners usually ask before connecting their data or distribution model. **How can we bring our data into Minds?** Use research files, questionnaires, respondent data, links and written Audience definitions, then review the proposed distributions before creation. **Can we publish an Audience for other teams?** Yes. Eligible teams can publish an Audience as a Marketplace add-on and manage its plan-specific access and pricing. **Is a synthetic audience an alternative to our panel?** It extends the panel rather than replacing it. Institutes typically use it before fieldwork to sharpen the instrument and after delivery to answer follow-up questions, while the panel keeps carrying the representative measurement. **What can we bring in to build a partner Audience?** Research files, questionnaires, respondent data, links and written Audience definitions. You review the proposed distributions before the Audience is created, so nothing is generated from sources you have not approved. **How do we check a partner Audience before clients use it?** Compare its answers with results you already hold. Because the Audience is reusable, you can re-run that comparison after each update rather than validating once at launch. **Can clients query our Audience without us running the study?** Yes. A published Audience is used directly by the teams you grant access to, so routine questions no longer queue behind your own capacity. **How is our attribution shown?** A published Audience carries visible attribution to the partner behind it, so the teams using it see whose research and methodology it is built on. **Who decides access and pricing for a published Audience?** You do. Eligible partners manage the plan-specific access and pricing that applies to an Audience published as a Marketplace add-on. **Does this cannibalise our survey business?** It tends to move work the other way: the exploration that used to be unbillable becomes a product, and better-targeted questionnaires make the surveys you do field more defensible. **How do we get started as a research partner?** Book a demo with your intended use case. Most partners start with one segment they already know well, publish it internally first, and open it up once the answers match what they expect. © 2026 Minds. Your target audience. AI-driven and grounded in transparent evidence. Build within minutes.