An online store sees a high abandonment rate and imagines a large pile of nearly completed orders waiting to be rescued. That interpretation is too simple. Some shoppers are browsing, comparing, saving an item, checking total cost, or deciding not to buy. Others encounter a preventable problem such as a payment error, confusing form, or surprise fee. A benchmark can tell you that abandonment is common; your own evidence must tell you why it happens.
Baymard’s checkout research reports an average cart abandonment rate around 70% across the studies it aggregates and examines usability issues in the checkout process. The research also distinguishes reasons such as ordinary browsing and not being ready to buy from friction that a merchant may be able to address (research). The headline average is not a target, not an estimate of your recoverable orders, and not proof that redesigning checkout will produce a particular lift.
Understand what the denominator includes
Abandonment calculations can vary depending on whether they count sessions, carts, or checkout starts, and on how the analytics system handles repeat visits and cross-device behavior. A person may add an item on a phone and purchase later on a laptop. A guest who leaves a cart open for weeks may count differently from someone who reaches payment and encounters an error.
Before comparing your number with an industry benchmark, document the exact event definitions, timeframe, and data source. Check for bot traffic, tracking gaps, and duplicate events. Use the same calculation consistently over time. Benchmark numbers are most useful as context, not as a scorecard.
Also distinguish cart abandonment from checkout abandonment. A shopper can add an item and never begin checkout; another can enter address and payment details before leaving. Those steps represent different levels of commitment and may point to different causes. A store that combines them into a single headline rate cannot tell whether people are still browsing products or are blocked while trying to pay. Record the funnel stages separately before choosing an intervention.
Separate intent from preventable friction
Build a funnel that distinguishes add-to-cart, checkout start, shipping details, payment attempt, and completed purchase. Break it down by device, product, traffic source, and customer type when sample size permits. A sharp drop at payment may indicate a technical failure or missing payment option. A drop before checkout may reflect price, delivery terms, or comparison shopping.
Ask buyers and nonbuyers what happened. A short post-purchase survey can ask what nearly stopped the customer. Abandoned-cart surveys have response bias: people willing to answer may differ from those who disappear. Support tickets, payment logs, and moderated tests provide other angles. No single source reveals every reason.
Payment-provider logs can reveal declines that analytics cannot interpret, while a customer interview can surface concerns that never produce a support ticket. A session replay may show where someone hesitated but cannot tell you what they were thinking. Use these sources as complementary evidence. Respect privacy requirements and avoid collecting more behavioral data than needed to answer the question.
Look for fixable problems without promising recovery
Common areas to inspect include shipping costs revealed late, unclear delivery timing, forced account creation, unavailable payment methods, confusing address forms, coupon-code distraction, and errors on mobile. Test the full checkout yourself. Ask a few users to complete a realistic task and observe where they hesitate.
An ecommerce forum thread describes shipping surprise as a concrete source of cart hesitation, while another owner reported ad clicks without sales and uncertainty about the cause (shipping discussion, clicks and no sales). These are anecdotes that can suggest questions; they do not establish prevalence or prove that shipping explains your abandonment.
Estimate business impact realistically
Prioritize issues by how many shoppers encounter them, the value of affected orders, and the cost and risk of a fix. If a payment integration fails on one browser, repair it. If shoppers abandon after seeing a necessary shipping charge, hiding that charge is not a solution; show it earlier so the buyer can make an informed choice. Some exits are healthy because the product or terms are not right for the shopper.
Track completed orders, payment failures, refunds, cancellations, average order value, and support contacts after making a change. A checkout that produces more orders but also more mistaken purchases may not improve the business. Note promotions, product availability, and traffic changes alongside the intervention.
Report the rate with the date range and cohort definition attached. “Our cart abandonment is 70%” without a method can sound precise while concealing what was counted. A more useful internal note might say that among measured checkout starts on mobile over a defined four-week period, a stated share completed orders, with payment errors tracked separately. This language makes comparison possible without implying that the number is an industry truth.
Finally, separate recovery tactics from checkout improvement. A reminder email may bring some shoppers back, but it does not establish that checkout itself was usable. A discount can recover an order while lowering margin or training buyers to wait. Any intervention should be evaluated against the economics and customer relationship it creates, not only the number of transactions recorded.