A clearer size selector. More completed orders.

The commercial problem

Company had already paid to bring shoppers to its product pages. Yet mobile visitors were leaving before adding an item to the basket, even when popular sizes were available. The Director of Ecommerce wanted a defensible next experiment before approving another round of creative and promotion changes.

The evidence changed the brief

The team used FullSession funnels to isolate mobile product-page visits, then inspected sessions that reached the size selector without adding an item. Lift AI ranked that friction against the other conversion issues on the site, which is what put it at the top of the investigation list.

In 46 of 120 selected non-purchase replays, shoppers tapped an unavailable size, then repeatedly tapped Add to Bag. The selection looked active, while the explanation sat above the sticky purchase bar and outside the visible viewport. Heatmaps showed concentrated interaction in the same area. The selected replays revealed the failure mode. The experiment measured the effect of fixing it.

A smaller change than a redesign

The Senior UX Designer moved the availability message next to the size control, made unavailable sizes visually distinct, and kept the purchase button disabled until a valid size was selected. The team preserved the size guide and added a clear route to available alternatives. Engineering shipped the change behind an experiment flag in six working days.

What improved

Over 28 days, 79,843 eligible mobile visitors saw the original experience and 80,091 saw the revised version. Completed purchasers rose from 2,531 to 3,014. Purchase conversion increased from 3.17% to 3.76%: 0.59 percentage points, or 18.7% relative improvement.

Measured against the control rate, that is about 475 more purchasers than the treatment group would otherwise have produced. The team waited for the 30-day return window to mature: return rates were 18.2% in control (461 of 2,531) and 18.0% in treatment (543 of 3,014), and average order value was $86.40 and $86.10 respectively. Return rates and order values were similar in both groups.

How purchase conversion was measured

Visitors were assigned persistently on their first eligible mobile product-page visit. The primary outcome was one completed purchase within seven days, counted once per visitor. Both variants ran concurrently, with the same prices, promotions and inventory feed. The final cohort was allowed seven days to mature before analysis. Rates and relative improvement are rounded. Desktop visitors and products with no purchasable sizes were outside the claim, and the result is a treatment-cohort difference, not a business-wide or annualized figure.

What the team took forward

Company rolled out the change and retained the funnel as a weekly check. The Director of Ecommerce now asks for the affected journey, the observable failure and the validation plan before funding a broad redesign. FullSession supplied the behavioral evidence and the prioritization; the customer team designed, implemented and tested the fix.