Insight

Checkout Had No Cross-Sell Opportunity

Fame’s checkout had no product recommendation module, so shoppers with the strongest purchase intent saw no complementary, recently viewed or AI-recommended products before completing their order. This left a commercial opportunity unused: a relevant addition could increase basket value, but any module at this stage also risked distracting customers from payment. Our recommendation therefore focused on a compact checkout placement that would preserve the main purchase flow and keep the primary CTA clear. Fame was using Shopify Plus, which made checkout extensibility available, and Rebuy was already part of the store’s setup. Our Rebuy integration approach treats recommendations as a behavioural aid rather than extra merchandising, with relevance, placement and QA carrying as much weight as the number of products shown.

Jump to results

A/B Test

Rebuy Product Recommendations in Checkout

We ran an A/B test directly through Rebuy across mobile and desktop checkout. The control had no recommendation module. Variant 1 added a Rebuy checkout widget showing relevant or complementary products generated from the shopper’s cart and browsing behaviour. We hypothesised that exposing useful additions at the point of highest purchase intent would increase conversion and revenue without disrupting checkout. Conversion rate was the primary KPI, with revenue and revenue per visitor used to judge the wider commercial effect.

The Results

Variant 1 increased conversion rate by 2% and revenue by 5%, while revenue per visitor fell by 0.16%. Rebuy does not report session metrics for this experiment, so revenue per session could not be calculated. Two products were purchased through the widget during the test. This pattern indicates a positive overall response without evidence that the added module disrupted checkout, although the small product count and unreported confidence level mean the result should be treated as directional. Checkout tests collect eligible traffic more slowly than tests on higher-traffic pages, which makes significance harder to reach within the same period. Rather than extend the experiment, Blend recommended rolling out Variant 1 and moving to the next planned widget experiments. Follow-up tests should compare recommendation sources and product rules, then assess placement or presentation while keeping the core checkout path unchanged. Post-launch monitoring should track conversion, revenue per visitor and direct widget purchases together, because Rebuy’s reporting limits make no single metric sufficient on its own.

  • 2%

    increase in conversion rate

  • 5%

    increase in revenue
Book a CRO discovery call

See More A/B Tests