How Longest-dwell ranking works
Ranks by attention — average time on the product page = total dwell ÷ the views that actually timed it — surfacing considered items people study before deciding.
Longest dwell ranks by average time on the product page — total dwell divided by the number of views that actually reported a dwell time. Long dwell means shoppers are studying it: a considered purchase. It surfaces the items that hold interest. Products with no timed views are dropped.
1. The collection, unranked
A normal collection page. The order is whatever the catalog lists — the best performers could be buried on the second row where few shoppers scroll.
2. The ranking engine scores it
The engine divides each product’s total dwell time by the number of views that reported one — views the pixel never timed don’t dilute the average, and a single view counts for at most 10 minutes. A product seen by thousands but skimmed in seconds ranks below one that fewer shoppers study for longer — attention, not traffic. Engagement signals measure in whole days, so an hours-based window counts as one day, and a Period or Snapshot window falls back to the last 30 days.
3. The front row gets discovered
The strongest products now fill the front row — the slots that get the most views and clicks. That’s the payoff: ranking turns your best products into the ones shoppers see first. If fewer products have a timed view than your Max products cap allows, the leftover slots are filled from the collection’s current order, below everything with a real dwell time.