Engagement context
This report is anonymised from a real audit completed in early 2025. Details have been generalised; the structure, severity scores and roadmap conversion are intact.
The brief
- Client. Mid-market UK retailer. Online revenue approximately £180m. Three-year-old commerce stack on a major SaaS platform.
- Scope. Mobile checkout flow on iOS and Android web. Cart through order confirmation. Both guest and account paths.
- Depth. Heuristic review against Nielsen 10 plus WCAG 2.2 AA accessibility audit.
- Trigger. Mobile cart abandonment had risen 11 percentage points over six months. Analytics had not surfaced a cause.
- Timeline. Two weeks from kickoff to presentation.
Executive summary
The executive summary is the first page of the report. Three findings. Three sentences each. Everything else is appendix.
Three findings account for the majority of abandonment
The audit surfaced 31 findings, 6 of which were severity 4 (catastrophe) and 11 of which were severity 3 (major). Three of these findings explain the bulk of the mobile cart abandonment lift the team has seen over the last six months. Each of the three is shippable within the current sprint.
1. Saved card state is hidden during the payment step. Account holders re-enter card data on every order. This single fix is expected to recover three to five percentage points of mobile conversion.
2. Address autocomplete failures break the address step on Android. Customers entering UK postcodes see a near-empty dropdown and resort to manual entry, which fails validation on the next step. The fix is a configuration change to the autocomplete library.
3. Order confirmation page fails WCAG 2.4.13 focus appearance on the primary CTA. Sighted keyboard users and switch users cannot reliably activate the "Continue shopping" CTA. Customer support has logged 47 related contact-us reports in the last quarter.
The executive summary is read by the senior stakeholder, often on a phone, often during a meeting that started ten minutes ago. It has to land in two minutes. Lead with the top three.
Severity rubric
The rubric appears on page two of the report and is referenced by every finding. Made visible so that severity is a decision the team can interrogate, not a number the practitioner pulled from the air.
- 4 — Catastrophe. Blocks the user, loses data, breaches WCAG legal minimum, or causes financial harm. Fix immediately.
- 3 — Major. Slows or stops the user frequently. Fix before next release.
- 2 — Minor. Users work around it. Fix this quarter if possible.
- 1 — Cosmetic. Polish. Fix when convenient.
Worked findings
Six findings from the report, shown in full so the format is clear. Real audits would include all 31; the structure is identical for each.
Saved card state is hidden during the payment step
The problem. Authenticated users with saved cards arrive at the payment step and see an empty card form rather than their saved card. Users either re-enter the card (90 percent) or abandon to fetch the physical card (estimated 8 percent based on session recordings).
The recommendation. Surface the saved-card state by default on the payment step with the masked card details and the option to use a different card. Aligns the mobile flow with the existing desktop pattern. No new components required; the design system already includes the SavedCard pattern.
Address autocomplete returns no UK postcode results on Android Chrome
The problem. The Loqate integration is configured for "address" only, not "postcode". UK customers enter a postcode and see an empty dropdown, which they interpret as a system failure. Manual address entry then fails postcode validation at the next step because of a separate format mismatch.
The recommendation. Change the Loqate integration mode to "postcode-first" and align the postcode regex on the validation layer to match the dataset Loqate returns. Tested manually on five postcodes during the audit; all five now return valid suggestions.
Primary CTA on order confirmation fails WCAG 2.4.13 Focus Appearance
The problem. The primary "Continue shopping" CTA has outline: none applied in the global stylesheet, with no replacement focus indicator. Keyboard users cannot tell when the CTA is focused. Switch users cannot reliably activate it. 47 contact-us reports in the last quarter mention difficulty finding the way out of the confirmation page.
The recommendation. Apply a 2px outline-offset focus ring in the brand violet with a 4px outline-offset for visibility against the page background. Audit the rest of the design system for similar outline: none patterns; the working assumption should be that more exist.
Delivery date selector requires recall rather than recognition
The problem. Delivery dates are presented as a long scrolling list with no visual hierarchy between "next day", "weekend" and "later" options. Users scroll, lose context, and back out to the cart to check the urgency of the order.
The recommendation. Group delivery options into three labelled bands (Express, Standard, Later) with the cheapest option in each band pre-expanded. Existing pattern available in the design system's Stepper component.
Form labels disappear on focus across the checkout
The problem. Input labels float as placeholders and disappear when the user starts typing. Users with cognitive load, slower typing speed, or simply distraction lose the label and don't know what they're entering. Particular impact on the card-number step where the difference between card number and CVV becomes ambiguous.
The recommendation. Replace the floating-label pattern with a persistent label above each input. This is a design system change; estimated impact across 14 forms in the wider product. Recommended bundled into the next design system release rather than checkout-only.
Inconsistent CTA placement between cart and payment steps
The problem. Primary CTA is bottom-right in the cart and bottom-left in the payment step. Users hesitate or miss-tap. Not a high-impact finding individually but flagged because the fix is trivial and removes a small confusion.
The recommendation. Standardise primary CTA to bottom-right across all checkout steps to match the design system spec.
Roadmap output
Findings were scored against the impact-effort quadrant. The grid below is the one that went into the presentation deck.
Sprint 1 (this sprint)
- F-001 Surface saved-card state
- F-002 Loqate postcode configuration
- F-003 Replace outline: none with focus ring
- F-006 Standardise CTA placement
- Plus four other quick wins not shown here
Sprint 2-4 (this quarter)
- F-004 Delivery date grouping
- F-005 Form label redesign (design system)
- Two further design-system level fixes
Sweep (quiet sprint)
- Minor copy revisions in payment confirmation
- Spacing inconsistencies in cart summary
- Eight cosmetic findings bundled
Parked
- Two findings requiring a checkout replatforming to fix; revisit when next platform migration is scheduled.
Sprint 1 captures all the high-impact, low-effort findings — the ones that justify the audit budget on their own. The team committed to these nine items at the end of the working session and the tickets were written live.
What we learned
Three lessons from this engagement that generalise.
- Most "mystery" conversion drops aren't mysteries. The team had spent three months investigating the mobile abandonment lift with analytics and A/B tools. The audit found the cause in two days because the failure modes were qualitative, not statistical.
- WCAG 2.2 findings carry the most weight. The product team had assumed accessibility was "table stakes" and didn't expect substantive findings. Three severity-4 WCAG findings in a single flow shifted the conversation and unlocked engineering time the design team had not previously been able to access.
- Tickets written in the room get done. The audit's value compounded because the team didn't leave the working session without nine tickets written. Six weeks later, six of the nine had shipped. Audits that end on a "we'll get to it" note typically ship two or three.