HOW TO READ YOUR RESULTS
See what a study tells you. and its limits.
A completed study shows A/B preference counts, sentiment scores, per-profile responses and an age breakdown. Independent mode shows paired score comparisons instead of preference votes. Export the inputs and outputs to keep an audit trail.
Example messages.
a campground emails past guests about a 25% weekday discount on tent sites from October 1 to November 15.
Tent sites 25% off Sunday to Thursday, Oct 1 to Nov 15.
Quieter trails, cooler nights. Weekday tent sites 25% off, Oct 1 to Nov 15.
Run a clear comparison.
Use the same audience and product facts for both versions. Choose a question before running: clarity, credibility, interest or preference.
Head-to-head mode shows both texts together. Independent mode scores each separately against the same scale, then compares the scores for each profile.
What the results mean.
The messages shown here are example inputs. Run the demo to see actual model outputs; no result percentages are fabricated on this page.
The survey supplies observed profiles. AI generates the reactions to your text. The included records are unweighted, and the results are not responses collected from people.
Frequently asked questions
How do I start?
Try a free study or open Studio to choose your audience. Paste both texts, add relevant context and run the study.
What should I do with a close result?
Inspect individual responses and revise the contrast. Small score differences are not a significance test. Use real audience feedback or a live experiment to validate consequential decisions.