Twelve months, three Netorigo storefronts, three checkout configurations, one parallel A/B test. The result surprised us more than we expected, and the headline conversion number was not the most interesting part. The abandonment distribution per step and the demographic shifts mattered more.
The setup
Three partner storefronts (a content-aggregator in the mediaorigo mould, a service-platform in the Lunda mould, and a classic B2C retailer) put more than 71,000 visitor sessions through the test. In each one we split traffic evenly between three checkout configurations:
- Card-first. Bank card is the first and visually dominant payment option, bank transfer hidden under a 'more options' link.
- Transfer-first. Bank transfer is the first option (nodding at the partner-program patterns where users actively look for transfer), card placed below.
- Mixed. Card, Apple Pay, Google Pay, bank transfer and PayPal as equal-weight tiles, no preselection (the user decides).
On the Netorigo platform, the checkout storefront's useCheckoutOptions hook sets the display per session using an experiment_arm cookie, and the payment_method_selected event lands in a PostgreSQL event_log table via the outbox pattern.
Raw conversion numbers
- Card-first: 45.2% conversion (cart view to completed payment)
- Transfer-first: 38.7% conversion
- Mixed: 51.1% conversion
The mixed configuration won clearly, but not by a runaway margin. Transfer-first was the laggard, and card-first sat in the middle. So far, no surprise.
Cart abandonment by step
The most interesting signal is in the cart_abandonment_by_step event distribution. In the card-first configuration, 18.3% of shoppers abandoned at the payment step because they were looking for a 'more options' link they could not find (cursor heatmap analytics shows the cursor hunting the bottom third of the page for four to seven seconds). In the transfer-first configuration, the final step (the card-entry form) produced 14.2% abandonment because shoppers thought they had to use the transfer option and only realised they wanted card at the last moment.
The mixed configuration's abandonment rate stayed at a flat 9.8% across all steps, which tells us shoppers picked exactly what they wanted on the first try.
The demographic surprise
Our prior assumption was that the 35-and-up cohort would prefer transfer (familiarity plus bank-card scepticism), and the 18-to-34 cohort would prefer card. The test flipped that. The 35-to-44 group picked card 53.4% of the time in the mixed configuration, because they have already adopted Apple Pay and Google Pay at home. The 18-to-24 group, surprisingly, picked transfer 32.1% of the time, because many of them avoid using their own bank card online or are using a parent's card.
The biggest surprise: the 65+ group picked PayPal 28.7% of the time in the mixed configuration, higher than any other age group. The explanation is simple: they are eBay veterans, where PayPal is the default.
The one conclusion that surprised us
The largest causes of cart abandonment are NOT related to the payment-method choice. They are related to shipping costs appearing only at checkout. In the mixed configuration we saw 6.4% abandonment in the second when the +1490 HUF shipping cost first appeared, rising to 12.1% when shipping was 2900 HUF.
The forward-looking recommendation: do not keep refining the payment-method order, make the full landed cost visible on the product page. And the mixed checkout is the default now. Do not waste another quarter A/B-testing transfer-first.