Traditional vs synthetic user research

Traditional user research recruits real people through panels, surveys and interviews, and measures what they say with statistical confidence, over weeks. Synthetic user research interviews synthetic participants (often called synthetic users) built from published evidence about a type of customer, and returns directional findings in hours. Panels prove a result. Synthetic research narrows what is worth proving.

Traditional research panels (Dynata-style, Forsta-style, Qualtrics-panel-style: any pre-recruited respondent database you pay per completed response) are the established methodology for quantitative research, substantiated claims, and statistical-confidence reporting. Candor is a synthetic research platform. The honest comparison is how the two fit together. Most teams can use both, in different parts of the same research cycle. Some teams won’t need full panel rounds at all, and that’s a legitimate choice depending on the kind of research the team does.

What are the differences between traditional and synthetic user research?

DimensionTraditional research (panels)Synthetic research (Candor)
Respondent typeReal humans (recruited panel)Evidence-grounded synthetic participants
Timeline per study4 to 8 weeks (with recruitment + analysis)Hours
Cost per study$15K to $50K per panel concept round ($23K median)Free plan, $10 per interview ($7 on the $75-a-month Standard plan)
Sample size typical200 to 1,000+ responses (B2C), 50 to 150 (B2B)Up to 24 synthetic participants per study
RecruitmentRequired, weeks of screeningNot needed
Quantitative point estimatesYes, with statistical confidenceNo (directional signal only)
Substantiated claimsYes (designed for it)No (not appropriate)
Brand and satisfaction trackingYes (consistent wave-over-wave methodology)No (synthetic is moment-in-time)
Concept screening (early-stage)Yes (slow + expensive)Yes (strong fit)
Discovery interviewsYes (slow + expensive)Yes (strong fit)
Value-prop and message testingYes (slow)Yes (strong fit)
Price testingYes (slow + panel-contaminated over time)Yes (strong fit, no panel contamination)
Problem validationPossible but rare due to costYes (strong fit)
Hard-to-recruit audiencesMajor constraintMinor constraint
Established category benchmarksYes (decades of comparability)Building

This table is a starting frame, not a final scorecard. Panel research is right for some questions. Synthetic is right for others. The teams that pick the right tool per question, rather than treating it as a category-vs-category choice, get the most out of both.

What traditional research panels are best at

Traditional panel research is the gold standard for several research jobs. Anything that needs real human respondents at statistical N, with documented methodology and reportable confidence intervals, belongs on a panel.

Quantitative point estimates. When the answer needs to be “27% of customers prefer Option A, plus or minus 3 percentage points at 95% confidence,” that’s a panel job. Synthetic research can produce directional signal. It can’t produce statistically-bounded point estimates from real-respondent N.

Substantiated claims. Anything going on a label, in a regulatory submission, in a clinical document, or in a publicly-defensible market claim needs real-respondent data with documented methodology. Panels are designed for that documentation. Synthetic research is not.

Brand and satisfaction tracking. Real customer perception over time, sampled consistently across waves, with year-over-year benchmarks. Synthetic research can hypothesize about brand-perception drift. Only real-respondent tracking can measure it.

Established benchmarks. Industries with decades of category-norm benchmarks (CPG concept-testing thresholds, ad-testing performance bands, NPS distributions by sector) have those benchmarks because they were built on panel data. New methodologies inherit category trust slowly. Established panel methodology already has it.

Anonymized aggregation at scale. When you need 500 or 5,000 responses to a structured survey for segmentation, market sizing, or share-of-wallet analysis, panel infrastructure handles that scale. Synthetic research is a different tool.

What Candor (and synthetic research generally) is best at

Synthetic research solves a different set of problems, the ones where speed, breadth, and cost are the binding constraints. How it works walks through the pipeline end to end.

Early-stage discovery. Before you scope a panel round, you want to know what shape the problem is. Synthetic research lets you talk to a population of evidence-grounded synthetic participants about their pains and current behaviors in hours. The output sharpens what to ask your real respondents about, and which segments are worth recruiting. See problem discovery.

Concept screening before panel rounds. Most concepts that go through traditional concept testing don’t make it to launch. Running every concept through a $23K-plus panel round is bad economics. Synthetic concept testing runs in roughly 1 to 2 hours of elapsed time per concept (most of it background pipeline work), so a project’s full concept pool fits comfortably in a single focused work week. Panel budget then concentrates on the survivors. See concept testing.

Hypothesis sharpening. Stress-test the hypotheses inside a roadmap, plan, or strategy against an evidence-grounded synthetic population. Get a verdict on each before deciding which to substantiate with panel research. See problem validation.

Value-prop and message testing at speed. Five angles, twenty synthetic participants, 100 reactions, in hours rather than four weeks. See value-prop testing.

Research in populations that are slow or impossible to recruit. Healthcare patients with narrow eligibility, regulated members, niche professional audiences, early-product segments. Synthetic research grounded in published evidence bypasses the recruitment friction.

Qualitative depth at speed. A traditional panel survey gives you 200 short responses to fixed-question structure. A Candor study gives you up to 24 long, conversational interviews with persistent synthetic participants that remember everything you’ve asked them. Different signal shape, different decision support.

The economics forcing the change

Most consumer-insights and B2B research operations face a structural math problem.

Concept volume is project-driven and stacks fast across every shape of team. A startup iterating on features, value propositions, and positioning typically tests 10 to 50 concepts per project. A consultancy running a strategy engagement is in the same range. A large company launching new products in-market is testing concepts across upfront screening, iteration, and finalist rounds at similar volume per launch. Most teams run multiple such projects a year, reaching hundreds of concepts tested in some form. Traditional panel concept testing for a full validation round runs roughly $15K to $50K (industry median around $23K) and takes 4 to 8 weeks per round including recruitment and analysis. Panel budget can’t cover the full project volume, so most concepts get cut on internal judgment with no respondent signal.

This is the structural waste. Concepts get killed without ever being tested. Some of the killed ones might have been winners. You’ll never know.

The math shifts when synthetic research handles the screening layer. Each synthetic concept test runs in roughly 1 to 2 hours of elapsed time (most of it background pipeline work, with the researcher’s active attention closer to 30 to 60 minutes), so a full concept pool is a single focused work week at most. The finalists advance to panel testing with respondent-grounded evidence already in hand. The cut concepts get killed with respondent-style reasoning, not internal judgment. Panel rounds get concentrated on the survivors with the highest signal-to-noise ratio.

B2B research has its own version of this problem. C-suite and senior-decision-maker research commands incentive premiums of $200 to $800 per completed response on top of base panel rates, and rounds take weeks because the audience is hard to recruit. Most B2B insights work runs at small N because the panel cost ceilings out. Synthetic research grounded in published B2B research can produce directional signal on dozens of hypotheses in hours, then concentrate the expensive real-respondent budget on the highest-value questions.

The frame isn’t “synthetic replaces panels.” It’s panels work better when synthetic research has filtered the question space first.

The hybrid research stack

The teams getting the most value from synthetic research don’t replace panel work. They use synthetic research before and between panel rounds. The pattern that’s emerging:

  1. Discovery in Candor. Run problem-discovery interviews across a broad synthetic audience. Identify the segments and pains that warrant respondent research.
  2. Hypothesis sharpening in Candor. Pressure-test the discovery findings with problem validation studies. Decide which hypotheses are worth substantiating.
  3. Concept screening in Candor. Run the project’s concept pool (10 to 50 concepts) through synthetic concept testing. Kill the unpromising ones with respondent-style reasoning, not internal judgment.
  4. Panel research for the survivors. Take the surviving concepts, value-props, and pricing hypotheses to a real-respondent panel for statistical-confidence validation, claim substantiation, and category benchmarking.
  5. In-market behavior tracking. Post-launch, real customer analytics and tracking studies are the ground truth for what actually happened.

Steps 1 through 3 used to be either skipped (internal-judgment culling) or done as expensive small-N qualitative work. Step 4 was where the research budget had to live. Now steps 1 through 3 are fast and cheap, and step 4 gets concentrated on the questions that have already survived a synthetic screen.

This is the hybrid research stack. The teams ahead of the curve are running it now.

When to use which

Some research jobs belong on a panel. Some belong on synthetic. Most research operations need both.

Use traditional panels when

  • You need quantitative point estimates with statistical confidence
  • The output needs to substantiate a claim (regulatory, label, public-facing, advertising)
  • You’re tracking brand health, satisfaction, or perception over time with consistent methodology
  • You need anonymized aggregation at scale (segmentation, market sizing, share-of-wallet)
  • You’re operating in a category with established panel-based benchmarks
  • The research budget supports panel-grade rounds and the timeline supports 4 to 8 weeks per round

Use Candor when

  • You’re at the discovery or hypothesis stage and need fast, broad signal
  • You’re screening multiple concepts and want to concentrate panel budget on the survivors
  • You need qualitative-style depth in hours, not weeks
  • Your audience is slow or impractical to recruit (regulated, niche, early-product)
  • You’re stress-testing assumptions before committing to a panel round
  • You need to iterate on a value-prop, message, or price across many angles fast

For most teams, both answers are “yes.” The question is sequence: synthetic for the early and middle of the research cycle, panel for the validation and substantiation.

Where to go next

This comparison is one of several. For other angles, see Candor vs UserTesting, Candor vs Synthetic Users, or the full comparison hub. For the category overview, see what is synthetic user research. For how the hybrid model works in concept testing specifically, see concept testing with synthetic users.

Common questions about traditional vs synthetic user research

No. Panel research is the gold standard for quantitative point estimates, substantiated claims, brand tracking, and any research that needs real-respondent data with statistical confidence. Candor is a different methodology layer: fast, evidence-grounded, qualitative-style synthetic interviews. Most teams can use both, and some teams won't need full panel rounds at all depending on what kind of research they do. The right mental model when both are in play is that synthetic research runs before and between panel rounds, screening concepts and sharpening hypotheses so panel budget concentrates on the questions that warrant statistical-confidence validation.

Not in the panel-research sense. Statistical confidence intervals on point estimates require real-respondent samples at sufficient N, with documented sampling methodology. Synthetic research produces directional signal across a population of evidence-grounded synthetic participants: which concept resonates more, which value-prop lands, where price sensitivity breaks. If your decision requires "X% of customers, plus or minus Y at 95% confidence," that's a panel job. If your decision requires "this concept is clearly stronger than that concept and here's the reasoning," synthetic research can give you that faster and cheaper.

The math depends on volume. 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. Panel-grade concept rounds run $15K to $50K each ($23K industry median); most teams reserve panel money for 4 to 6 finalists per project and kill the rest on internal judgment. With synthetic screening first, every concept gets respondent-style reasoning rather than gut feel, and panel rounds concentrate on the survivors. Total research spend can drop materially while total research capacity rises.

Yes, with one important note: Candor models B2B and B2C as separate research domains, with different attribute models, bias profiles, and personality weightings. B2B research is a buying-committee problem with formal decision criteria and organizational context. B2C is an individual-consumer choice problem. Treating them with the same model loses signal in both directions. Candor's B2B model is grounded in organizational buying research. B2C is grounded in consumer behavior research. The user picks the audience type when setting up a Candor study.

When the entire research portfolio is brand tracking, regulated claim validation, segmentation at quantitative N, or anything else that needs real-respondent data with statistical confidence. If your team's research operation lives entirely in those zones, synthetic research probably isn't adding much to your stack. The teams getting the most value from Candor are the ones with a discovery, concept-testing, hypothesis-validation, or fast-iteration component to their work, which is most consumer-insights and B2B research operations.

More FAQs →

Bring evidence to your next decision.

Start with a free project, or walk through Candor with us first.

Book a DemoTry for Free