FOR CONSUMER INSIGHTS

Synthetic research for consumer insights teams

Consumer insights teams face a math problem most stakeholders never see. Concept volume is project-driven and stacks fast: a typical project generates 10 to 50 concepts across upfront screening, iteration, and finalist rounds, and most teams run multiple such projects a year. Traditional panel concept rounds cost $15K to $50K each (industry median around $23K) and take four to eight weeks per round. Budget realistically covers a handful of panel rounds per project, so most concepts get killed on internal judgment with no respondent signal behind the decision. Candor screens the unpromising concepts in hours so panel budget concentrates on the survivors. It’s a first-pass filter for your existing methodology, not a replacement for the panel work that substantiates launch decisions.

Why CPG concept testing is breaking under current economics

The traditional concept-testing cycle was built for a slower product world. Big brands launched a small number of large bets per year. Each bet got rigorous panel research. The economics worked because the bets were few and the budgets were larger.

The volume has changed. Brand teams now test 10 to 50 concepts per project across line extensions, packaging variants, campaign angles, claim variations, channel-specific positioning, and DTC sub-brands, and most teams run multiple such projects a year. Panel research economics haven’t shifted at the same rate. The result is that most concepts never get respondent-grounded testing at all. They get killed in the brand-team room, by a manager whose intuition might be right or might not be.

Three failure modes follow from that.

Promising concepts get killed early. A concept the brand manager finds unconvincing gets cut, but it might have tested well with a real audience. The team never finds out.

Weak concepts survive too long. A concept the team is emotionally attached to advances to panel research, costs $23K (industry median), and underperforms. The team learns it was weak, but the budget is already spent.

The whole portfolio runs on intuition. When most concepts never see respondent signal, the brand team is making allocation decisions across the launch funnel on gut feel. That’s not what insights teams want to support, but the budget math has forced it.

Synthetic research changes the budget math by adding a layer that runs before the panel layer. Each synthetic concept test runs in roughly 1 to 2 hours of elapsed time (most of it background pipeline work), so a project’s full concept pool fits comfortably in a single focused work week of insights-team time. The strongest survive to panel research with respondent-style reasoning behind the cull, not internal politics.

What synthetic research changes for consumer insights teams

The concrete capabilities that matter to insights work.

Concept screening at portfolio scale. Run the project’s concept pool (10 to 50 concepts) through evidence-grounded synthetic concept testing in roughly 1 to 2 hours of elapsed time per concept (most of it background pipeline work). Each concept gets reactions from 8 to 16+ personas with calibrated personality and bias profiles. The output is qualitative-style depth on which concepts resonate, which fall flat, and the reasoning behind both. See concept testing.

Price testing without panel contamination. Synthetic price testing across persona variance surfaces willingness-to-pay patterns and acceptable-tier signal without the panelist contamination that comes from respondents who’ve seen too many price tests in the past quarter. See price testing.

Value-prop and message testing at speed. Five claim angles, eight personas, 40 reactions, in hours. Which message lands, which gets ignored, which gets distorted into a meaning you didn’t intend. See value-prop testing.

Early-stage discovery for new categories. Before the brand team commits to a category-entry hypothesis, talk to a population of evidence-grounded personas about their current behaviors, their unmet needs, and the language they use to describe both. See problem discovery.

Assumption validation before launch gates. Stress-test the assumptions inside a launch plan or brand positioning brief against an evidence-grounded population. Per-assumption verdicts with reasoning. See assumption validation.

Each of these maps to a question consumer insights teams already work on. The shift is that the screening layer becomes affordable across the whole portfolio rather than reserved for the survivors.

Where Candor fits in the brand and innovation research cycle

The teams getting the most value from synthetic research don’t replace panel methodology. They use it before and between panel rounds. The pattern that’s emerging in consumer-insights operations:

  1. Discovery and category exploration in Candor. Run problem-discovery interviews across the target consumer audience. Identify the segments and unmet needs that warrant deeper concept work.
  2. Concept screening in Candor. Take the project’s concept pool (10 to 50 concepts) through synthetic testing. Kill the unpromising ones with respondent-style reasoning. Promote the survivors with documented synthetic findings the brand team can read.
  3. Value-prop and price-sensitivity testing in Candor for the surviving concepts. Sharpen the positioning angles and price anchors before sending to panel.
  4. Panel validation for the strongest survivors. Concentrate panel budget on the three to six concepts that have already survived a synthetic screen. Get statistical confidence, claim-substantiation data, and category-norm benchmarks where they’re required.
  5. In-market tracking post-launch. Real customer behavior, brand health tracking, and sales-pull data are the ground truth for what actually happened. Synthetic research isn’t a substitute here.

Steps 1, 2, and 3 used to compress into “the brand team picks the strongest six and prays.” Now they become structured research stages with respondent-grounded reasoning at each gate. Step 4 becomes more focused because the concepts arriving at panel research have already been pressure-tested. Step 5 remains a real-customer job.

The cost math compounds. Concentrating panel rounds on finalists that have already survived synthetic screening (vs. spreading panel budget across un-screened concepts) typically reduces total panel spend while raising total research throughput. Less wasted panel money. More research coverage. Better gate decisions.

Which research types stay on real-panel research

Honesty here matters because the wrong framing breaks trust with the insights director who has decades of methodology experience. Synthetic research doesn’t replace traditional panel research for the following kinds of questions.

Substantiated claims for label, regulatory, or advertising use. Anything going on a pack, in an ad, or in a regulatory filing needs real-respondent data with documented methodology. Panels are designed for that documentation. Synthetic research is not.

Statistical point estimates with confidence intervals. “27% of category buyers prefer Concept A, plus or minus 3 percentage points at 95% confidence” requires real-respondent N. Synthetic research produces directional signal across persona variance, not statistically-bounded point estimates.

Brand health and tracking studies. Wave-over-wave consistency, year-over-year benchmarks, real perception drift in-market. These depend on real-respondent panels sampled consistently across time. Synthetic research is moment-in-time.

Anonymized aggregation at scale. Segmentation work, market sizing, share-of-wallet analysis with thousands of responses. Panel infrastructure handles that scale and synthetic research is a different tool.

Final launch-gate validation. Anywhere the cost of being wrong demands real-respondent ground truth before committing to manufacturing, distribution, or media spend. Panel research is still the right last step.

A useful test: if the question is “which of these concepts is stronger, for whom, and why,” synthetic research is in scope. If the question is “is this concept defensible to the regulator, the retailer, or the launch committee with real-respondent data behind it,” panel research is required.

How insights teams use Candor in a typical launch cycle

The pattern in consumer brands running Candor alongside their existing panel methodology, for a typical new-product launch cycle. The quarterly framing reflects the launch calendar, not the synthetic platform time. Each Candor study runs in roughly an hour. The rest of each quarter is concept development, stakeholder reviews, panel scheduling, and the other work that surrounds a launch.

  1. Quarter 1: Discovery sprint in Candor. Run problem-discovery interviews across the target consumer segments. Hours of platform time. Brief the brand team with synthesized signals before the concept-development workshop that typically takes the rest of the quarter.
  2. Quarter 2: Concept generation and synthetic screening. Brand team generates the concept pool (this is the slow step, not the testing). All concepts go through synthetic concept testing in Candor: a single focused work week of insights-team time. Brand team reviews the synthetic output and kills the unpromising ones, advancing six or so survivors to deeper work.
  3. Quarter 2: Synthetic value-prop and price testing for the survivors. Each survivor gets value-prop and price-sensitivity exploration in Candor in another work week of platform time. Concepts get sharpened, repositioned, or retired.
  4. Quarter 3: Panel concept testing for 3 to 4 finalists. This is the slow research step. Real panel research with recruitment, fieldwork, and analysis takes 4 to 8 weeks per round. Concentrate panel research budget on the strongest survivors.
  5. Quarter 4: Launch. Real-customer behavior, in-market tracking, and post-launch panel work take over from synthetic.

A consumer-insights team running this pattern can move multiple product launches per year through structured discovery-to-launch research, with synthetic research doing the heavy lifting in stages 1 to 3 (compressed into days of platform time, spread across quarters of calendar work) and panel research concentrated where it matters most.

Where to go next

To see how Candor compares to other research approaches, read Candor vs traditional research panels, Candor vs UserTesting, or the full comparison hub. For the category overview, see what is synthetic user research. For the platform walkthrough, see how Candor works. For role-adjacent landers, see Candor for product discovery teams and Candor for healthcare and regulated CX teams.

Common questions from consumer insights teams

No. Panel research is the gold standard for quantitative point estimates, substantiated claims, brand tracking, and any concept work that needs real-respondent data with statistical confidence. Candor is a different methodology layer: fast, evidence-grounded, qualitative-style synthetic concept testing. Most insights teams adopting Candor use it as the screening layer that runs before panel concept testing, so panel budget concentrates on the strongest concepts rather than spreading thinly across the whole portfolio. Some teams may decide they don't need full panel rounds for certain categories, and that's a legitimate call depending on what the launch decision requires.

Directional accuracy holds well for the questions synthetic research is designed for: which concept resonates more, which value-prop framing lands, where price-sensitivity breaks. Point-estimate accuracy is weaker than panel research because synthetic research isn't producing statistically-bounded numbers from real-respondent N. The pattern emerging in published synthetic-versus-real comparisons is convergence on directional signal and divergence on point estimates. Treat synthetic concept testing as a fast screening layer, not as a substitute for the panel work that produces the numbers a launch committee needs.

For directional claim screening (which of these five claim angles resonates more, which gets distorted into a different meaning, which feels overstated), yes. For claim substantiation (data that supports a regulated claim on a label or in advertising), no. That requires real-respondent data with documented methodology. The right use is upstream: screen claim angles synthetically to identify the strongest candidates, then substantiate the survivors in panel research with the rigor the regulatory submission requires.

Yes, and this is a known limitation of LLM-based synthetic research that consumer-insights teams should understand. Synthetic personas exhibit a documented agreement bias: they tend to be more cooperative with the framing of a question than real respondents would be. Candor addresses this with critic validation on every response (a separate agent checks each response against the persona's established profile for consistency) and with calibrated cognitive bias modeling (the bias library includes acquiescence bias as a modeled trait, so the bias can be partially controlled). The honest framing is that synthetic concept testing is best for relative-comparison questions ("which concept is stronger") rather than absolute-magnitude questions ("how much will this concept resonate"). Comparison handles the agreement bias well. Absolute magnitude doesn't.

The math depends on volume and category, but the pattern is consistent. Concept volume is project-driven: a typical project generates 10 to 50 concepts across upfront screening, iteration, and finalist rounds, and most teams run multiple such projects a year (reaching hundreds of concepts tested in some form). Panel-grade concept rounds run $15K to $50K each ($23K industry median). Routing all concepts through synthetic screening first and reserving panel testing for the three to four strongest survivors per project typically reduces total panel spend by 30 to 50 percent while raising research throughput, because synthetic screening costs are flat per study rather than scaling per respondent.

More FAQs →

Candor is in development.

Be the first to know when it launches.

No spam. Just a note when Candor is ready. Powered by Highline Beta.