Everything people actually ask before they run something, answered short. The uncomfortable questions are in here on purpose — if the honest answer to one of them rules Reynard out for you, better that you find out on this page than after paying.
No. Every voice in a report is simulated by a language model. No respondent was recruited, nobody was paid to answer, and no profile in the panel corresponds to a living person.
Profiles built from aggregate public statistics: age band, education, household income, region and a value orientation, combined in roughly the proportions those attributes occur in the population. The United States panel uses Census Bureau data plus eight value segments; the Dutch panel uses CBS data plus eight value groups.
Whether an idea, price or line is understood and accepted by a broad cross-section, and which parts of that cross-section disagree and why. It is a way to find the misreading and the objection you had not thought of, before you commit.
A chatbot gives you one answer from an averaged perspective. A run puts the same scenario to a weighted set of segments and reports them separately, so you see the split rather than the middle, with each segment's contribution scaled to how much of the population it is.
A free check returns in roughly half a minute. The larger tiers take longer; the in-depth tier, which includes an alternative scenario and a second round, takes a few minutes.
An acceptance score, a per-segment breakdown rather than a single average, quotes giving the reasoning behind each segment's reaction, and region and segment breakdowns. Higher tiers add an emotion map, concerns and appeals per segment, a risk analysis, an alternative scenario and a downloadable PDF.
Each segment's contribution to the total is scaled to its share of the population, so a segment that is a fifth of the country moves the result more than one that is a twentieth. /census-weighting works this through with numbers.
Yes, on the paid tiers: by value segment, life stage, region, education, income band, age band and gender. The narrower the target, the thinner the public statistics behind it, so narrow runs should be read more cautiously than broad ones.
The Netherlands and the United States. Both panels are built the same way; the Dutch one came first and is used throughout the site as the proof market.
On the paid tiers, yes — you can put a follow-up to individual simulated respondents. The in-depth tier allows a full exchange rather than a single answer.
We do not know, and neither does anyone else selling this. No accuracy figure has been measured here in a way we would stand behind, so none is published.
Because a number nobody could have failed to produce is not information. A meaningful figure needs a prediction fixed before the answer was visible, a test that could have gone badly, and a commitment to publish the bad runs too. /accuracy sets out exactly that experiment, written down in advance, and it has not been run yet.
No. Significance comes from sampling real units from a real population. Nothing here is sampled, so there is no margin of error, no confidence interval, and nothing that should be described as representative in the statistical sense.
Not as fieldwork, and you should not present it as such. What it is legitimately good for in that setting is showing your reasoning: that you considered how different parts of the market would read the thing, that you found a specific objection, and that you changed something because of it. Present it as a simulation you used to narrow the options, never as a measurement of what people said.
Narrow and specialist audiences, anything that depends on very recent shifts in mood or price levels, and any question about your existing customers — who are a self-selected slice of a population and usually unlike the average in exactly the way that matters.
No. There is no purchase intent in a simulation. A favourable reading is a reading of a description, not a forecast of revenue.
Yes. The first check costs nothing, needs no account and no card. It exists so you can see the shape of the output before deciding whether it is worth paying for.
$12, $25 or $39 per run in dollar markets, and €9, €19 or €29 in euro markets, depending on the tier. The price shown is the price billed.
No. Every run is a one-time payment. There is nothing to cancel, and nothing renews.
Not for the free check. For paid runs you need an account so the report can be stored and reopened.
No. Scenarios are not used for model training, and they are never shared between customers. Your scenario is visible to your account and to nobody else's.
No. The profiles are generated from aggregate published statistics. No individual's records are used, and no profile corresponds to a real person.
Yes. Reports belong to your account and can be deleted from it. /trust describes what is stored and for how long.
A different product on the same panel: an agent opens the URLs you give it in a real browser, navigates, clicks, scrolls and fills things in, and reports the journey step by step with screenshots. Alongside that, eight weighted population groups read the page cold and report first impression, friction and clarity.
Not if the login needs two-factor authentication, an SMS code or a captcha. A failed sign-in is reported as a friction point and the run continues on whatever is reachable without an account.
No. It is not an accessibility audit, not a performance benchmark and not a security test. It reports where a first-time visitor's path breaks and how the page reads.
A simulated cross-section of a population: how it is built, how it is weighted, and the limit it cannot cross.
Read the comparisonThe category as a whole: where it came from, the three shapes it takes, and the criticisms of it stated properly.
Read the comparisonWhy a segment that is 21.9% of a population must count for more than one that is 6.9%, worked through with numbers.
Read the comparisonNo accuracy figure has been published. This page says why, and states the experiment intended to produce one.
Read the comparisonWhat is stored, where it runs, and what is never used for training.
Read the comparisonThe first check costs nothing and needs no account. Write the thing you are testing, pick the market, read the report.
Read the comparisonOne check costs nothing and needs no account. You see the format of the output before you decide whether it is worth paying for a full run.
Start the free check