Example study

Cast, Validated by Cast

Cast is a synthetic-user research platform. You describe your idea, your markets and what you're trying to decide; a Casting Agent proposes Ideal Customer Profiles, builds a cast of AI users calibrated to each one (a real spread of enthusiasts, pragmatists and skeptics, never just the people who already agree with you), interviews them with the same unbiased question techniques a trained researcher would use, never steering the witness, never naming a number first, never asking a leading question, and reports rankings and themes, labeled directional and...

A finished Cast study, read only. 24 interviews with AI users across 4 Ideal Customer Profiles (ICPs), run on October 20, 2026. Everyone quoted here is an AI user, synthesized to match an ICP. They are not real people.

Run a study like this See how it works

What the study found

All four buyer types adopt once their specific blocker is named and answered; none of the four objected on price.

Every one of the four groups compared Cast's price favorably to what they already spend, from a founder's unpaid cold-DM hustle to a consultant's six-figure fieldwork budget. What varied was the specific thing each group needed proof of before trusting the findings: founders wanted evidence the answers aren't just agreeable; marketers wanted evidence regional and audience nuance survives; consultants wanted confidentiality guarantees and client-defensible labeling; policy makers wanted unambiguous distance from the formal public record. Five of 24 AI users started genuinely skeptical. All five ended the interview willing to try it, once shown the specific capability that answered their specific objection.

Who was asked

4 Ideal Customer Profiles, each with its own cast.

ICP A · 6 AI users

Early-stage startup founders

Pre-seed to Series A founders who have to decide which customer to build for with no research budget and no time to wait.

Markets: United States, Saudi Arabia, Germany

Founder or co-founder, pre-seed to Series A
Has personally made a call on which customer segment to prioritize in the last 12 months
Has no in-house research team

ICP B · 6 AI users

Marketing and growth leaders

Brand, growth and product marketers who need to test messaging, creative or positioning before committing media spend.

Markets: United States, Brazil, Japan, Nigeria, Sweden

Owns or directly influences messaging, creative or positioning decisions
Has tested or wanted to test messaging before a campaign in the last 12 months
Works at a company with a defined target audience or customer segment

ICP C · 6 AI users

Management and strategy consultants

Consultants at boutique to large firms who need defensible primary research on a client deadline, often under confidentiality constraints.

Markets: United Kingdom, Singapore, Mexico, India, Denmark, South Africa

Works at a management, strategy or specialist consulting firm
Has personally needed primary customer research on a client engagement in the last 12 months
Works under confidentiality constraints (NDA, client confidentiality) at least sometimes

ICP D · 6 AI users

Public-sector policy makers and analysts

Government and think-tank policy professionals who need an honest read on how a proposed policy, program or regulation would actually land before it is finalized.

Markets: United States, Ghana, Belgium (EU), United Arab Emirates, Poland

Works on policy development, analysis or advisory in a government, think tank or regional authority
Has needed a read on public or stakeholder reaction to a proposal before it was finalized in the last 12 months
Works on proposals where candor from affected groups is hard to get through formal channels

Top findings

What each group said, ranked by how many AI users said it.

ICP A · 6 interviews

Early-stage startup founders

Founders adopt fastest of all four groups: the real alternative is doing no research at all, and Cast's price beats that by a wide margin.

1. The real competitor is doing nothing, not another research tool6 of 6 AI usersEvery founder's honest current process was informal outreach, a low-response survey, or pure gut feel.
2. Trust, not price, is the real adoption barrier for this group4 of 6 AI usersFounders burned by a flattering-but-wrong report or a too-agreeable vendor demo want proof of honest disagreement before trusting any new research tool, independent of cost.
3. A visible, specific anti-sycophancy mechanism resolves that trust barrier directly3 of 6 AI usersWhen told specifically about a built-in check against agreeable, flattering answers, previously skeptical founders moved to willing to try it, more than once unprompted.

ICP B · 6 interviews

Marketing and growth leaders

Marketers with the clearest, costliest past failure adopt fastest; the rest want proof Cast reaches real regional and audience nuance, not a flattened average.

1. A specific, costly past failure from untested messaging is the strongest predictor of fast adoption2 of 6 AI usersMarketers who had personally traced a real, expensive campaign miss back to untested messaging wanted to use Cast as a standard pre-flight check on every future campaign.
2. The biggest resistance is a flattening worry, not a cost or speed worry4 of 6 AI usersMarketers working across regions, personas or hard-to-reach audiences worried a synthetic-audience tool would collapse real differences into one generic average, echoing a past disappointment with an online panel or...
3. Per-persona, per-segment separation in the report directly answers the flattening worry3 of 6 AI usersOnce shown findings are reported separately by ICP and persona rather than averaged together, marketers who had been burned by a blended survey or flattened panel moved toward willing to test it.

ICP C · 6 interviews

Management and strategy consultants

Consultants see the clearest time and cost case of any group, gated by two distinct concerns: client-defensibility of the findings, and strict confidentiality under NDA.

1. Compressed fieldwork timelines are the single biggest unmet need in this group3 of 6 AI usersConsultants on tight, client-driven engagement timelines described fieldwork regularly consuming a third to half of total project time, directly limiting evidence quality under deadline pressure.
2. Confidentiality under NDA is a hard, specific requirement, not a general privacy preference2 of 6 AI usersConsultants working on confidential deals or client engagements needed explicit, written confirmation that no information about the client or target persists outside their control, as a precondition, not a nice-to-have.
3. Explicit, clearly labeled uncertainty is reassuring to risk-averse firms, not a weakness to hide2 of 6 AI usersAt firms with strict internal quality-assurance review, the directional-and-uncalibrated labeling was the single most reassuring detail, since it matched exactly the language their own QA process already requires.

ICP D · 6 interviews

Public-sector policy makers and analysts

Policy makers want speed and candor the formal process cannot always deliver, but need explicit, permanent labeling that this is never a substitute for the formal public record.

1. The formal public process often arrives too late to inform the decisions it is meant to support2 of 6 AI usersPolicy staff repeatedly described internal recommendation deadlines falling before the formal comment period even opens, leaving them to rely on small, informal, self-selected outreach instead.
2. Social-desirability bias in formal consultation is a known, specific problem Cast is expected to solve, not just speed up2 of 6 AI usersPolicy analysts described real instances where formal consultation participants gave polished, expected answers that didn't match candid reaction surfaced later through informal channels.
3. Legitimacy protection is the central, non-negotiable concern for anyone close to the formal process1 of 6 AI usersA director responsible for the formal public comment record needed absolute, permanent, unremovable labeling distinguishing this from public input before seeing any value in it at all.

In their words

Verbatim, from the interviews.

Portrait of Priya Shenoy, an AI user — Priya ShenoyLedgerline, San Francisco, CaliforniaEarly adopter · ICP A

Portrait of Jasmine Carter, an AI user — Jasmine CarterLumenne Skincare, Los Angeles, CaliforniaEarly adopter · ICP B

Portrait of Nathaniel Price, an AI user — Nathaniel PriceBirchgate Partners, London, United KingdomEarly adopter · ICP C

Portrait of Elena Vasquez, an AI user — Elena VasquezCity of Austin Economic Development Office, Austin, TexasEarly adopter · ICP D

How this study ran

One brief, one hour, no recruiting.

24
AI users cast: 24
interviews held: 10
questions asked: 4
ICPs compared

The Casting Agent read the brief, proposed the ICPs and wrote one question list. Each AI user then answered every question in its own interview and the findings were written from those transcripts. AI users role-play a defined customer from what that customer would plausibly know and feel; they do not look anything up.

Run a study like this.

Describe your idea and your customers. Meet your cast in about an hour.

Run a study like this