Hit rate per jury: count how often it was right.
The hit rate shows how often the 99takes jury had the version ahead that then won in the market. For that, you report the result after launch. 99takes only publishes it per jury from ten decidable results, always with their number and a range.
Hit rate
Four questions after launch.
Report a market result
no creditsWhich version did better in the market, with the raw numbers per version and their source, such as clicks and impressions from Google Ads or conversions from GA4. Also possible in your account under “Runs”. Only a comparison of two versions can be reported.
“B did better than A on demo requests.”See the hit rate
no creditsPer jury, with the number of results and a range. Published from ten decidable results; before that it says “not measured yet”.
“How often was the jury right?”Refine the audience
no creditsIf the jury is off, refine single personas; the others stay stable.
“Head of IT judges too harshly. Refine that persona.”Share the result
no creditsA report with its own link, for your team or client. Links can be withdrawn, and from the Pro plan a report can be limited to invited readers.
“Turn this run into a report.”Only what can be decided is counted.
A simulation is a prediction. Whether it holds is shown by the market, and that takes reported results. Until there are enough of them, 99takes says so instead of showing a number.
| Counted | Not counted |
|---|---|
| Runs with a clear winner and a reported market result | Undecided runs |
| Results of exactly the jury that judged | Markets without a clear winner |
| Number of results and range, always shown | Market gaps within the noise, such as 3.1 % against 3.0 % click rate on 200 impressions |
The next run starts from what you learned.
Check the next copy
Two versions or a whole page against the same jury.
Buyer simulation →Have a better version written
Challenger versions from the objections, without made-up proof.
Challenger versions →The hit rate belongs to your account and to exactly the jury that judged. Why a simulation comes before an A/B test, not instead of one: alternative to A/B tests.
Questions about the hit rate
When is a result decidable?
When the run had a clear winner and the market did too, with a gap larger than the noise. Only then can the jury have been right or wrong. Everything else is stored but not counted, and the reason is stated.
When is there a rate?
From ten decidable results per jury. Before that it says “not measured yet”, even if the first verdicts were right: four out of five would be 80 %, but with so few cases it means nothing.
Which numbers should I report?
A count and a base per version, with their source. For ads, clicks and impressions from Google Ads, Meta or LinkedIn; for landing pages, conversions and sessions from GA4 or your A/B tool; for emails, clicks or opens and recipients. Only then can it be checked whether the market gap was more than noise.
What does reporting cost?
Nothing. Reporting a result, looking up the rate and refining the audience cost no credits.