FOR PRODUCT TEAMS

Synthetic research for product discovery teams

Product discovery teams hit a research bottleneck before they hit a strategy bottleneck. More PMs, more roadmap surface, more discovery questions, but the same one or two researchers serving the whole org. Many PMs are capable researchers, but they’re also juggling ten other priorities. Researchers want to take on more, but the queue is already full. Most discovery questions never get respondent signal because there isn’t enough capacity on either side. Candor is the layer that adds capacity: evidence-grounded synthetic respondents that PMs, designers, and product ops can interview about problems, concepts, value props, and pricing without booking the UXR team. Researchers stay in deep work. PMs do more of the research they were going to do anyway, with more depth.

Why product discovery is the chronic research bottleneck

The math is consistent across most product organizations. A typical B2B SaaS company has eight to twelve PMs for every dedicated UX researcher, sometimes much worse. Each PM has a roadmap of features under consideration, each feature has discovery questions, each discovery question wants research signal. The research team can run one or two real studies at a time, with weeks of recruitment per study. Output capacity is fixed. Input demand is unbounded.

The result is one of three things, none of them good.

Most discovery questions go unanswered. PMs make roadmap decisions on intuition because real research is too slow to consult. Some of those decisions are right. Some aren’t. The org never knows which is which until the feature ships and underperforms.

The research that does happen is thinner than anyone wants. Skilled PMs can run good research, but with ten priorities and limited weekly time for it, even strong researchers run three calls instead of ten and reach the decision before the sample really warrants. The PM isn’t doing bad work; they’re doing the best work the time budget allows. Researchers feel the same constraint from the other side: they could go deeper if they had capacity, but the queue is full of other PMs’ questions waiting their turn.

Researchers become a bottleneck nobody loves. PMs resent the queue. Researchers resent being the gate. Leadership resents the slow pace. The function that’s supposed to make the product team faster makes it slower because there isn’t enough of it.

Synthetic research changes the math on which discovery questions can get answered. Not all of them. But enough of them to free researcher time for the studies that genuinely need real customers.

What synthetic research lets product teams do

The use cases where synthetic research is a clean fit for product discovery work.

Problem discovery. Before scoping a solution, surface unmet needs and pains in your target audience. Candor lets you interview a population of evidence-grounded personas about their current behaviors, workarounds, and constraints in hours. The output sharpens what to ask real customers about, and which segments are worth recruiting.

Problem validation. You have a hypothesis about a pain point. Is it real for specific segments, or is it your team’s projection? Pressure-test the hypothesis against synthetic personas grounded in published research about that audience. Get per-segment verdicts with reasoning.

Concept testing. You have a feature concept, a positioning idea, or a workflow direction. Test multiple versions across personas before committing engineering resources. Find out which version resonates, which falls flat, and what the reasoning is.

Value-prop testing. Five angles, eight personas, 40 reactions, in hours. Which message lands and which gets ignored or distorted.

Price testing. Where does price-sensitivity break across personas with different willingness-to-pay anchors. Useful even for pre-revenue products figuring out positioning, not just for live products refining tiers.

Assumption validation. Stress-test the assumptions baked into a roadmap, plan, or pitch. Per-assumption verdicts grounded in evidence.

Each of these is a question PMs ask weekly. Each of these used to require a queue, a recruitment cycle, or a guess. Synthetic research makes it a same-day question.

Which questions still need real users

Honesty matters here, because the wrong framing breaks trust with the UXR team that’s supposed to be Candor’s ally inside the org. Synthetic research doesn’t replace real user research for the following kinds of questions.

Prototype usability and design feedback. Candor doesn’t support multimodal interaction with mockups, wireframes, or prototype frames today. If your discovery question is “does this design make sense to a user clicking through it,” that’s a real-user research job. Multimodal prototype walkthrough is a candidate future direction, not a shipping capability.

In-product behavior research. How real customers actually use your shipping product, where they drop off, what they ignore. Real product analytics and real user-session data are the ground truth.

Statistically-bounded quantitative research. “27% of customers prefer Option A, plus or minus 3 percentage points at 95% confidence” requires real-respondent N at sample sizes synthetic research isn’t built to produce.

Substantiated claims and regulated work. Marketing claims, regulatory submissions, clinical documentation: anywhere the research is going on a label or in a filing, real respondents are required.

Anything where the cost of being wrong demands real validation. Bet-the-company strategy decisions, major architecture pivots, anything irreversible. Synthetic research is the right first pass. Real customers are the right final pass.

A useful test: if the discovery question is “what shape does this problem have, for whom, and how do they currently solve it,” synthetic research is in scope. If the question is “did this real human in this real situation actually do what I think they did,” it isn’t.

How a product team uses Candor in a typical sprint

The pattern that’s emerging in product orgs running Candor alongside their UXR function. Most of the pipeline runs in the background while the PM gets on with other work. Active user time is roughly 30 to 60 minutes across the full workflow.

  1. Define the audience (5 to 10 minutes of user time). Describe who you want to understand (B2C or B2B context, learning goals, industry, region), and upload any company documents you want Candor to ground its personas in (PDF, DOC, DOCX, CSV, TXT, or MD).
  2. Audience generation runs (25 to 35 minutes in background). Candor retrieves real evidence about the target audience from published sources and uploaded documents, extracts behavioral signals with provenance, and segments the audience.
  3. Review segments and approve (~5 minutes of user time). Toggle which segments to include before persona generation.
  4. Persona generation runs (7 to 12 minutes in background). Candor generates 8 to 16 evidence-grounded personas across the selected segments, with calibrated OCEAN personality and cognitive bias profiles, and persona-specific memory.
  5. Generate the interview guide (~2 minutes). The PM picks the interview type (problem discovery, problem validation, concept testing, or price testing). Candor produces a question structure based on that selection and the learning goals. The PM can refine before running.
  6. Run the interviews (25 to 35 minutes in real-time for live mode; longer elapsed for auto-interview against multiple personas in parallel).
  7. Read the synthesis report (ready ~3 to 5 minutes after the last interview completes). Extracted signals, themes, archetype breakdown, tensions, opportunities. The PM brings findings into the next sprint review with structured evidence behind the recommendations rather than gut feel.
  8. If findings warrant real-customer validation, brief the UXR team. Surviving hypotheses, sharpened audience criteria, and a focused set of questions worth real-respondent investment.

Total elapsed wall-clock for a single discovery study is roughly 1 to 2 hours if the PM works through it continuously, but most PMs let the background stages (audience generation, persona generation, synthesis) run while they do other work, so the active attention required is closer to 30 to 60 minutes spread across half a day. A team running this pattern can move several discovery questions through synthetic research per PM per sprint, and feed only the survivors to the real-research queue. The UXR team works on fewer but higher-value real studies. The product team has evidence-backed reasoning behind more roadmap decisions.

How synthetic research fits with your existing UX research function

This is the question UXR leaders ask first, and they’re right to ask it. The honest answer.

Candor expands research throughput. It does not replace researchers. A team that adopts Candor without a UXR function is going to miss everything synthetic research can’t do: design usability, real-user behavior, regulated work, statistically-bounded findings. A team that adopts Candor with a UXR function gets more research done overall, with researchers focused on the studies that need their depth.

The right framing for UXR leaders is capacity, not substitution. Synthetic research lets PMs answer the questions UXR teams are too thinly-staffed to take on. The PM questions that were going unanswered or being self-served badly become researchable in a way researchers can actually inspect. The questions that need real users still go to real users. The UXR queue shortens. The org learns more.

Candor’s outputs are auditable. Every persona attribute is tagged with provenance. Every interview transcript is reviewable. Every synthesis report is structured. UXR leaders inspecting PM-led synthetic research can see the methodology, push back where it’s weak, and coach PMs on better question design. This is research the UXR team can stay in the loop on without being the bottleneck for executing it.

UXR teams that adopt Candor proactively tend to position themselves as the quality layeron synthetic-supported discovery: reviewing study designs, vetoing studies that need real customers, focusing their own efforts on the high-value real research the synthetic layer can’t handle. That positioning protects the UXR function’s strategic role. Instead of being the bottleneck for every research question in the org, the UXR team becomes the methodology partner across a much larger volume of research and the deep practitioner on the studies that genuinely need real customers. It’s a stronger seat at the table, not a smaller one.

Where to go next

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

Common questions from product discovery teams

No. Candor expands research throughput. The UXR function remains the right place for everything synthetic research can't do well: design usability, in-product behavior, statistically-bounded findings, regulated work, anything depending on lived human experience. Teams adopting Candor without a UXR function tend to miss those gaps. Teams adopting Candor with a UXR function tend to get more research done overall, with researchers focused on higher-value studies and PMs no longer queuing for every discovery question.

Pain-point exploration, current-behavior and workaround discovery, concept reactions, value-prop preference, price-sensitivity exploration, assumption stress-testing. Anything where the question is "what shape does this problem have, for whom, and how do they reason about it" is in scope. Questions about real human reaction to real artifacts (prototype usability, in-product behavior, video recordings of real customers) are out of scope and stay on the UXR team.

A PM running their own interviews is constrained by time, not skill. Many PMs are capable researchers, but they're also juggling ten other priorities, so the round that happens is typically thin: three calls, smaller samples, faster turnaround. Candor doesn't replace what skilled PM research can do. It adds capacity: 8 to 16 evidence-grounded personas across multiple questions in an afternoon, with calibrated personality and bias profiles drawn from peer-reviewed distributions, every attribute traceable to a source, and a critic agent validating each response for consistency. The output is auditable in a way fast PM-led interviews aren't, which means researchers can inspect and coach against synthetic studies without sitting through every call themselves.

Not today. Candor is for text-based interviews about problems, concepts, value propositions, prices, and assumption validation. If your concept lives in a Figma file, a packaging render, a video storyboard, or an interactive prototype, real-user testing platforms remain the right tool. Multimodal prototype walkthrough is a candidate future direction, not a shipping capability.

Order of hours, not days. The PM defines an audience, Candor's audience generation runs in the background (25 to 35 minutes), the PM reviews segments, persona generation runs (7 to 12 minutes), the PM picks the interview type (problem discovery, problem validation, concept testing, or price testing) and Candor generates an interview guide (~2 minutes), the PM runs the interviews (real-time for live mode, typically 25 to 35 minutes), and the synthesis report is ready a few minutes after the last interview completes. Total elapsed wall-clock is roughly 1 to 2 hours if the PM works through it continuously. Active attention is closer to 30 to 60 minutes spread across the workflow.

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.