Adding a product to cart proves interest, not purchase readiness. Shoppers can still encounter unexpected delivery costs, forced registration, limited payment methods, technical errors, weak trust signals or a reason to delay. Find the exact checkout step where people leave, then investigate that step on the devices and payment paths they use.
This is both an ecommerce web design and conversion optimisation problem. Recovery emails may regain a few orders, but they cannot compensate for a form that fails or a total that changes unexpectedly.
Google Analytics recommends building a checkout funnel to locate abandonment and notes that a high drop-off can signal a technical problem Source: Google Analytics. Start with observed behaviour before choosing a remedy.
Make sure the funnel is measured correctly
Track the important ecommerce events and validate them in a test purchase. At minimum, distinguish:
- view cart
- begin checkout
- add contact details
- add delivery information
- add payment information
- purchase confirmation
Record error events, voucher failures and payment outcomes as well. A missing purchase event can make successful orders look abandoned, while duplicate events can understate the problem.
Google's checkout journey report uses begin checkout, shipping, payment and purchase stages and displays abandonment between them Source: Google Analytics. Match those steps to the actual store rather than adding events that do not correspond to a user action.
Find the highest-value drop-off
Build a funnel by device, browser, traffic source, location, product type, new versus returning customer and payment method. Look at order value as well as shopper count.
A sharp mobile drop after delivery details suggests a different cause from a desktop drop after card authentication. A segment with fewer abandoned carts but much higher order value may deserve attention first.
Watch sessions or conduct usability tests only after defining the affected segment. Otherwise the team may observe ten different problems and fix none of them.
Check for technical failure
Complete real test orders across common mobile devices and browsers. Use production-like addresses, vouchers and payment outcomes. Test on a slower mobile connection, not only office Wi-Fi.
Look for:
- buttons covered by keyboards or consent banners
- form errors that do not identify the field
- address lookup that rejects valid South African formats
- inventory changing after the shopper proceeds
- payment redirects that lose the session
- totals that do not update correctly
- pages that time out or reload entered data
- inaccessible controls and poor focus order
Keep client and server error logs tied to a checkout session identifier without storing sensitive payment data. The site speed and conversions guide helps connect page delay to commercial journeys.
Show the real cost earlier
The cart should make price, delivery cost, taxes, discounts and likely delivery timing understandable before the final payment step. If a precise delivery price needs an address, explain that and provide an estimate where possible.
Do not advertise a headline price and reveal compulsory charges at the last moment. The shopper may interpret the change as a breach of trust even when the calculation is technically correct.
For South African stores, verify regional delivery rules, collection options, free-delivery thresholds and surcharges for outlying areas. Test whether the address logic communicates exceptions clearly.
Reduce account and form friction
Allow guest checkout where the operating model permits it. Offer account creation after purchase, when the shopper has already received value.
Ask only for information required to fulfil, pay for or lawfully administer the order. Use sensible input types, autofill, address assistance and persistent values. Explain why an unusual field is necessary.
Check payment fit and failure handling
Offer payment options that match the customer base and order values. That may include cards, instant EFT, digital wallets or other locally appropriate methods, depending on the merchant and provider.
For each method, test success, decline, timeout, cancellation and return from the provider. A failed payment message should say what happened, preserve the cart and offer a safe next step. Never expose gateway codes without a useful explanation.
Monitor authorisation and completion by payment method. Low performance can come from technical integration, fraud rules, issuer behaviour or a mismatch with shopper preference.
Strengthen trust at the decision point
Keep returns, delivery, support and business identity easy to reach from checkout without forcing the shopper to abandon the form. Use accurate security and payment marks, not decorative badges that make unverifiable promises.
Show stock and delivery claims consistently between product, cart and checkout pages. Provide a visible support option for customers who encounter a problem, especially for high-value purchases.
If your website receives product interest but not completed orders, examine the promise and the transaction as one experience.
Treat delay separately from failure
Some shoppers use the cart as a shortlist, compare prices, wait for salary day or seek approval. That behaviour cannot be eliminated through interface changes.
Use consented cart recovery with the exact products, current availability and a direct return link. Avoid creating false urgency or sending repeated messages after the item was purchased elsewhere.
Separate voluntary delay from checkout failure in analysis. A shopper who leaves before checkout saw different friction from one who receives three card errors.
My take: fix the first broken commitment
My take is that the most useful question is not "Why did the cart abandon?" but "What new commitment did the next step demand?" The next step may reveal cost, request an account, require an unfamiliar payment method or create uncertainty about delivery.
Rank issues by affected revenue, evidence strength, fix effort and risk. Repair the earliest high-impact break, then rerun the same funnel. This creates a learning system instead of a redesign based on opinions.
Run controlled improvements
Write a hypothesis that names the segment, observed barrier, proposed change and expected metric. Protect payment success, error rate, average order value and support contacts as guardrails.
Release one material checkout change at a time when traffic allows. For lower-volume stores, use usability testing, error logs and before-and-after cohorts rather than waiting indefinitely for statistical certainty.
Record the result in a checkout decision log, including the affected segment, screenshots, technical evidence, release date and guardrail outcome. This keeps later redesigns from reintroducing a previously confirmed barrier.
Frequently asked questions
Is cart abandonment always a website problem?
No. Comparison shopping, timing and changed intent are normal. The store should focus on measurable technical, cost, payment and trust barriers it can control.
Should delivery cost appear on the product page?
Show the clearest useful estimate available. If the exact cost depends on location, explain the rule and calculate it as early as practical.
Will a shorter checkout always convert better?
Not automatically. Remove unnecessary work, but retain fields and explanations needed for fulfilment, payment, compliance and confidence.
Should we start with abandoned-cart emails?
Start by checking the funnel and technical journey. Recovery messages are a secondary tool after the checkout itself works reliably.
Diagnose the checkout before redesigning it
If you need a measured checkout review, get in touch for ecommerce website design and conversion rate optimisation.
Related guides
For further background, read our site speed and conversions and woocommerce vs shopify. The conversion funnel glossary explains the terminology used in those guides.