Propensity One lite research preview
Paste 2-10 headline versions for one story, one per line. The model ranks them by how likely each one is to get the higher click-through rate in an A/B test. It runs in your browser; nothing you type is sent anywhere.
| # | Headline | Score | Chance the top pick beats it |
|---|
How good is it?
Loading the evaluation...
The chances stay close to 50% because most headline differences are small. Use it as a tie-breaker and to shortlist versions for a real test, not to replace one.
Limits
- Trained on news and social-issue headlines from upworthy.com, 2013-2015, in English. It hasn't been validated on product ads, other audiences or other languages.
- It reads the text only: no images, brand or audience.
- This is the lite, text-only model. The full Propensity One isn't public.