Confirm that the decline is real
Daily conversion can move because of a small order count, a weekend pattern, or a short promotion. Compare similar days and use enough sessions to distinguish a persistent decline from normal noise.
Mark the first date the change appears. That date gives you a practical boundary for checking price, inventory, advertising, reviews, content edits, and competitor activity.
Separate traffic quality from page performance
- Check whether sessions rose because campaigns began sending broader or less relevant traffic.
- Compare conversion by traffic source, marketplace, and child variation where the data allows.
- Review whether the product lost availability, delivery speed, or offer competitiveness.
- Look for new review themes or return reasons that create purchase hesitation.
Build a change timeline
List every meaningful change around the decline: price, coupon, main image, title, variation structure, stock status, fulfillment method, campaign targeting, and competitor offer. A timeline prevents unrelated edits from being blamed for the result.
Change one decision layer at a time
Fix availability and offer problems before rewriting content. If the offer is stable, test the highest-impact content element and record the result. Multiple simultaneous edits may improve performance, but they make the lesson impossible to reuse.