How can I identify checkout friction?

Interpret abandonment correctly.

Abandonment is initially an observation, not an explanation. People compare, check delivery options or buy later on another device. First define the steps your funnel actually counts: basket view, checkout start, payment attempt and confirmed order are different events.

Check measurement errors before making design decisions. Duplicate events, missing consent or payment-provider redirects can distort figures. Compare compatible periods and populations. One percentage without a definition is insufficient for prioritisation.

Is essential information visible in time?

Examine product variant, total price, available delivery and payment methods along the actual journey. Information belongs where the decision needs it. A general FAQ cannot replace a clear answer at the relevant choice. Check for conflicting details between basket and product pages.

When input fails, people should understand which field is affected and how to correct it. W3C describes understandable error notifications and success feedback. Preserving entered values helps people make corrections without restarting the whole process.

Actually complete the mobile journey.

A narrow screenshot does not test a mobile order. Test keyboard input, autofill, address changes, vouchers and return from the payment provider. Check whether errors are visible and the on-screen keyboard obscures the next action. Include zoom and operation without a mouse.

Use an appropriate test environment and test payment methods. A check must not accidentally create real orders, charges or customer notifications. Record device, browser, step and observed behaviour so the issue can be reproduced.

From finding to testable hypothesis.

Illustrative example: after an invalid postcode, the page jumps upwards while the error stays out of view. The hypothesis is not that the button needs another colour, but that people cannot identify the required correction. This suggests a targeted improvement to error feedback.

After implementation, first verify behaviour in the same sequence. Then review completion trends. Traffic, promotions and catalogue may have changed in parallel; a before-and-after comparison alone does not isolate the effect.

A short checklist for the next step.

Record four things per issue: affected journey, reproducible evidence, recommended change and acceptance check. Prioritise failed purchases and unclear errors over decorative details. Assign an owner and plan a verifiable completion.

Then observe confirmed orders alongside costs and returns. A higher completion rate may help but does not alone prove greater profitability. A controlled test may suit larger changes when the question and available data support it.

Sources and further documentation

W3C WAI: User Notification

Explore the topic

Ecommerce growth

Conversion optimisation

When is an A/B test worthwhile?