Widgets that quietly get better every week.
A planned ranking layer for Idukki widgets. It would move the posts that drive add-to-cart into the top slots. It is not built yet. Today, you can A/B test widget layouts in Experiments.
Design partners get sandbox access during the private beta, a Slack channel with the product team, and a 24-month price lock on the public-launch tier.
Self-improving mediaNot live yet
Self-improving media
Four things the beta will ship with.
- 01
Per-widget ranker
A separate model per surface (PDP gallery, story rail, video wall, link-in-bio) because what wins on PDP rarely wins in email.
- 02
Significance gating
A clip would move into a top slot only once its lift clears statistical significance.
- 03
Auto-retirement
Clips that under-perform for 14 days drop to the long-tail. Rights-cleared but stale → cycled out without a manual review.
- 04
Honest dashboard
"Why is this clip on top?" answered with the same data the ranker saw. No black-box recommendations.
Who we’re building this with.
- Brands running > 8 widget surfaces who can't hand-curate each one weekly
- Growth teams who already love A/B testing and want the loop to close itself
- CMOs who need attribution they can defend in front of finance
What ships when.
Now · not built
The ranker is planned. No date is set. Widget layout A/B tests work today in Experiments.
Design partners
Let every widget learn which posts sell.
The ranker is in private beta. Design partners get sandbox access and a direct line to the product team.