--- title: "How Minds Builds Synthetic Research Panels | Minds" canonical_url: "https://getminds.ai/research/methodology" last_updated: "2026-06-22T02:02:45.523Z" meta: description: "Minds builds synthetic research panels from internal context, validated external data, and market research partner data, then benchmarks outputs against..." "og:description": "Minds builds synthetic research panels from internal context, validated external data, and market research partner data, then benchmarks outputs against..." "og:title": "How Minds Builds Synthetic Research Panels | Minds" "twitter:description": "Minds builds synthetic research panels from internal context, validated external data, and market research partner data, then benchmarks outputs against..." "twitter:title": "How Minds Builds Synthetic Research Panels | Minds" --- Minds June 21, 2026·Methodology·Minds Team # **How Minds Builds Synthetic Research Panels** Minds builds synthetic research panels from internal context, validated external data, and market research partner data, then benchmarks outputs against real surveys. [Run a synthetic research panel](https://getminds.ai/?register=true) Minds is a synthetic market research platform for testing concepts, messaging, positioning, pricing, and audience reactions before teams commit to full fieldwork. The product grew out of research work connected to HFBK Hamburg and MIT research contexts, then developed into a commercial platform for market researchers, agencies, product teams, and go-to-market teams. That research background informs the methodology, but it is not an external certification or endorsement by those institutions. This page explains the public methodology behind Minds without exposing proprietary implementation details. ## The Method in Brief Minds follows three visible steps: 1. Define a research group from grounded evidence. 2. Build individual Minds with many scoped knowledge bases. 3. Benchmark synthetic outputs against real survey results where comparison data exists. The goal is not to invent a fictional average customer. The goal is to create a research-useful simulation of a defined audience segment, with enough context to respond consistently to concepts, claims, messages, and tradeoffs. ## 1. Group Creation Starts With Evidence Minds does not start from a single generic prompt like "act as a customer." Each research group is built around a defined audience, use case, and decision context. Groups can be grounded in three kinds of source material: - **Internal data:** approved first-party context such as customer notes, prior research summaries, CRM segments, interview themes, support themes, usage patterns, brand knowledge, product documentation, or sales objections. - **Validated external data:** public, licensed, or otherwise validated market context such as category reports, reviews, search signals, demographic patterns, industry data, and product-category language. - **Market research partner data:** structured inputs from research partners, panel providers, segmentation studies, survey data, or research taxonomies where the client has the right to use the data. The source mix depends on the study. A B2B pricing study may need buying-committee context and procurement language. A consumer concept test may need category behavior, usage occasions, cultural signals, and prior survey findings. A message test may need competitor language and customer objections. ## 2. Each Mind Uses Many Small Knowledge Bases Minds are not static persona cards. Each individual Mind is assembled from thousands of small, scoped knowledge bases that describe relevant background, constraints, needs, vocabulary, preferences, objections, buying context, category familiarity, and decision criteria. Those knowledge bases are selected and weighted for the research question. A pricing study, a concept test, and a brand-positioning study may use the same audience label but require different context. This structure lets a panel behave less like one averaged persona and more like a group of individual respondents with overlapping but distinct beliefs, priorities, and objections. ## 3. Research Runs as Interviews, Surveys, and Group Discussions Researchers use Minds for: - Concept testing - Message testing - Positioning research - Pricing and packaging reactions - AI focus groups - Audience and segment exploration - Early qualitative probing before human fieldwork A Mind can answer individually, join a panel readout, compare perspectives with other Minds, or respond to moderator follow-ups. The system is designed to surface disagreement, uncertainty, objections, and the reasons behind a response rather than only producing a summary sentiment score. For the underlying academic category, see our guide to [silicon sampling](https://getminds.ai/blog/silicon-sampling). For the broader commercial workflow, see the [synthetic research guide](https://getminds.ai/blog/synthetic-research). ## Accuracy Is Benchmarked, Not Assumed Synthetic research should be evaluated against real research, not treated as magic. In benchmarked use cases, Minds has shown 80-100% directional agreement against real survey results. "Directional agreement" means the synthetic panel identifies the same winning concept, strongest objection, preference direction, or message pattern as the reference human study. Accuracy varies by audience, data quality, sample design, category familiarity, and question type. Minds performs best when the audience can be well-grounded and the task asks for stated preferences, concept reactions, message resonance, tradeoff reasoning, or objection discovery. Minds should not be treated as a universal replacement for human research. Use human fieldwork when: - The decision is high-stakes, regulated, legal, medical, political, or safety-critical. - You need statistical population estimates with confidence intervals. - You need to observe real behavior rather than stated preference. - The audience is poorly represented in available data. - The stimulus depends on physical, sensory, or in-store experience. - Final validation must come from recruited human respondents. The strongest workflow is usually hybrid: use Minds to explore the space, test more variations, refine the instrument, and narrow the options; then use human research for the final questions that need external validation. ## Key Terms **Synthetic research** uses AI-generated respondents or panels to simulate how a defined audience may react to questions, concepts, messages, and tradeoffs. **Silicon sampling** is the academic method of conditioning large language models on respondent profiles, asking survey-style questions, and comparing the resulting distributions against human survey data. **Synthetic panel** means a structured group of AI respondents built to represent a segment, audience, buying committee, user group, or market category. **AI persona** usually means a single simulated respondent profile. In Minds, a Mind is more than a static persona: it is an interactive respondent grounded in many scoped knowledge bases. **AI focus group** is a moderated session where multiple synthetic respondents react to the same prompt, concept, message, image, or product idea and surface agreement and disagreement. ## What This Makes Possible Traditional research is often rationed because recruitment is slow and budgets are finite. Minds changes the workflow by letting teams ask more questions earlier: - Screen early concepts before paying for full fieldwork. - Compare message options before media spend. - Find objections before sales enablement or launch. - Test positioning before rebuilding a deck or website. - Rehearse qualitative interviews before recruiting humans. - Turn prior research into an interactive panel that teams can query repeatedly. This is why we describe Minds as a decision-support layer for research teams, not a replacement for rigor. The method is most valuable when it helps teams make better use of human research by focusing fieldwork on the questions that matter. ## Further Reading - [Silicon Sampling: How LLMs Simulate Survey Responses](https://getminds.ai/blog/silicon-sampling) - [Synthetic Research: The Complete 2026 Guide](https://getminds.ai/blog/synthetic-research) - [AI Focus Groups: How They Work](https://getminds.ai/blog/ai-focus-group) - [The Spark Effect: Creative Diversity in Multi-Agent AI](https://getminds.ai/research/spark-effect-creative-diversity-multi-agent-ai) ## **Frequently asked questions** ### **What is synthetic research?** Synthetic research uses AI-generated respondents or panels to simulate how defined audience segments may react to questions, concepts, messages, and tradeoffs. It is best used for fast exploration, pre-testing, and decision support. ### **How are Minds groups created?** Minds groups can be grounded in approved internal context, validated external market data, and market research partner data. The available sources depend on the research question, client permissions, and the evidence needed for the audience. ### **How is a Mind different from a simple AI persona?** A simple AI persona is often a static profile. A Mind is an interactive respondent built from many scoped knowledge bases, so it can answer surveys, react to concepts, join group discussions, and explain its reasoning. ### **How accurate is Minds?** In benchmarked use cases, Minds has shown 80-100% directional agreement against real survey results. Accuracy varies by audience, data quality, research design, and question type, so results should be used as directional decision support rather than universal statistical proof. ### **Does synthetic research replace human respondents?** No. Synthetic research is strongest for early concept screening, message testing, objection mining, and research design. Human research remains important for high-stakes, regulated, behavioral, sensory, or final validation work.