Foundation · 2 min
What you need to know
How do I measure ecommerce merchandising performance?
Measure merchandising from exposure through product selection, cart, purchase, contribution and returns. Compare placements and audiences while controlling position, popularity, inventory and promotion.
How do I know whether featured products are working?
Track who saw the featured area, which product and position appeared, and what those shoppers later bought. Use a holdout or rotation when possible to separate placement lift from existing popularity.
How do I measure recommendation performance?
Measure recommendation impressions, clicks, carts, purchases, net sales, contribution and incremental lift. Suppress already purchased or incompatible items so relevance failures are not hidden by click rate.
How do I measure product-card performance?
Track card exposure, product selection, quick-add success and downstream purchase by card design, position and device. A high click rate with low purchase may indicate missing information or misleading presentation.
How do I measure category-page performance?
Measure category entry, product exposure, filters, sorting, product selection, exits, purchases and contribution. Segment by category depth and device because browsing tasks differ.
How do I know whether sorting changes improved sales?
Compare equivalent visitors before and after the sorting change or run a controlled experiment. Examine product exposure, selection, conversion, contribution and guardrails rather than top-line sales alone.
How do I measure product-position bias?
Position bias is the tendency for highly placed products to receive more attention regardless of quality. Measure it through randomized rotation, controlled positions or comparisons of the same item in different positions.
A quick store check
Record both exposure and outcome for one featured position, then compare with a holdout or rotated position.
Foundation · 2 min
Frequently asked questions
Do clicks prove a recommendation worked?
No. Measure downstream cart, order, margin and incremental lift while accounting for position and shopper intent.
Should merchandising tests protect customer experience?
Yes. Track guardrails such as exits, search use, returns and support issues, not only revenue.