Cart abandonment is one of the most repeated statistics in ecommerce and one of the most misunderstood.

Baymard Institute’s average across 50 studies is 70.22% (updated 22 September 2025). That is not a single-study result, and not every abandoned cart represents a broken checkout. People compare prices, save products, browse casually and change their mind.

The useful question is therefore not how to eliminate cart abandonment. It is:

Which abandonment is caused by friction we can actually remove?

At a glance

  • Separate natural abandonment from fixable friction before optimising anything.
  • Unexpected costs are a leading cause — surface them earlier.
  • Forced accounts and form friction are self-inflicted losses.
  • Measure the funnel by step, then diagnose with error data.

Separate intent from friction

A shopper who added a product to compare totals behaves differently from one who genuinely tried to buy but could not finish a form. Treating both as the same metric leads to poor decisions.

Natural abandonment Fixable friction
Browsing and comparing Unexpected cost
Not ready to buy Forced account creation
Changed mind Form problems
Saving for later Payment failure
— Confusing delivery or trust issue

Optimisation should focus on the second column.

Unexpected costs are especially destructive

The checkout is too late to reveal fundamental economics. Baymard’s cited survey lists extra costs such as delivery and tax among reported reasons for abandonment (cart abandonment rate and survey responses).

Whenever possible, expose cost expectations earlier. A product page that says shipping starts from a known amount, a cart that shows a clear estimated total and a checkout that confirms it feel very different from a checkout that silently adds delivery and a fee. The implementation depends on address and delivery logic, but the principle is simple: avoid surprise.

Do not force an account without a strong reason

A customer came to buy a product; requiring an account introduces another task. Forced account creation is another meaningful cause of abandonment. For normal D2C commerce, offer guest checkout unless account creation is genuinely necessary, and invite the user to create an account after purchase. The order is the primary task.

Reduce the number of decisions

A checkout can have few fields and still feel difficult. Users may need to decide on shipping address, delivery method, billing address, payment method, account and marketing preferences. Present those in a logical order and avoid unnecessary choices: default billing to shipping when they match, and do not design a fake selection experience when only one delivery method exists.

Form design matters

Checkout forms fail in small ways: ambiguous labels, unnecessary fields, poor validation, errors only after submit, lost input, the wrong mobile keyboard, strict formatting and unclear required fields. A user should always know what is required, what format is expected, what went wrong and how to fix it.

Errors should be specific. “Invalid information” is bad; “Enter a valid postal code” is better.

Payment is a high-anxiety step

The closer a user gets to transferring money, the more clarity matters. The payment step should clearly show the final total, currency, payment method, delivery choice, billing information and the purchase action. Do not introduce new promotional content here. Checkout is completion infrastructure, not another homepage.

Payment methods should reflect the market

More options are not automatically better; they add complexity. Select methods based on country, device, customer type, average order value and existing payment data. A B2B checkout may need invoice, net terms, purchase order and bank transfer; a D2C checkout may prioritise card and local accelerated methods.

Preserve user input

One of the fastest ways to turn a recoverable error into abandonment is deleting data. If payment fails, do not make the user re-enter their address. If one field is invalid, preserve everything else. If they go back to edit shipping, preserve payment state where security allows. Design checkout recovery explicitly.

Mobile checkout requires more than responsive CSS

On mobile, keyboards cover content, autofill matters more, errors are easier to miss, users have less context and slow interactions feel worse. Test with real devices and real payment methods. A desktop browser resized to 390 pixels is useful but not a complete mobile test.

Performance matters more at the end of the funnel

By checkout the user has already invested effort. A delayed address selector or an unresponsive payment button creates uncertainty at the worst moment. Watch Interaction to Next Paint, third-party payment scripts, validation logic, address lookup, analytics overhead and layout shifts — see Core Web Vitals for ecommerce. A technically valid checkout can still feel broken.

Measure the funnel by step

Do not stop at conversion rate. Instrument each step:

cart viewed → checkout started → address completed → delivery selected → payment started → payment failed → purchase completed

Then segment by device, country, payment method, traffic source and new versus returning. The goal is to locate friction.

Analyse error data

A payment failure rate is not a UX metric by itself. Break failures into causes — declined card, technical provider error, 3DS failure, validation error, expired session, inventory issue — and fix each differently.

Optimise with evidence

A useful sequence: quantitative funnel data, error logs, session or usability observation, a hypothesis, a change and measurement. Do not begin with “competitor X has a green checkout button.” Checkout optimisation is diagnostic work.

What to audit first

  • Are costs clear?
  • Is guest checkout available where appropriate?
  • Are unnecessary fields removed?
  • Are errors specific?
  • Is input preserved?
  • Are payment methods appropriate?
  • Does mobile work properly?
  • Is the final total unambiguous?
  • Can failures recover?

There is no perfect universal checkout — only a checkout with less unnecessary friction for a particular customer and commercial model.

Continue reading

Checkout and conversion work sits in Commerce at FRN.Q.