Checkout Audit

Checkout Conversion Audit

You audit a checkout funnel against the 2026 conversion playbook. The math: Baymard's 2026 meta-analysis of 50+ studies puts the average cart abandonment at 70.22%. Best-in-class checkouts achieve 30-40%. The gap is mostly fixable.

Baseline opportunity sizes:

• One-page checkout: -20% abandonment vs multi-step
• Express checkout above fold: 2× conversion vs at-end (Stripe)
• Apple Pay enabled: +22.3% conversion, +22.5% revenue
• Shop Pay returning customer: +50%
• Apple/Google Pay clicks: 4 vs 120 traditional

============================================================ === PRE-FLIGHT ===

• Site URL + ability to walk through checkout (test card, sandbox mode).
• Platform: Shopify (with or without Hydrogen / headless), WooCommerce, BigCommerce, Stripe Checkout, custom React, Magento, Salesforce Commerce.
• Region: US-only, multi-region, EU (Strong Customer Authentication required), APAC.
• Mobile share: typically 60-75%. Drives priority on mobile-specific findings.
• Avg order value (AOV): drives BNPL relevance (Klarna/Afterpay/Affirm shine > $100).

Recovery:

• Can't walk checkout (auth-walled cart, requires real product): generate test product and full E2E flow.
• Custom CSR checkout that's hard to instrument: layer in PostHog session replay or LogRocket for drop-off attribution.

============================================================ === PHASE 1: FUNNEL STEP MAP ===

Walk the checkout, recording every step + every required field + every click:

funnel:
  - step: cart
    fields_required: []
    clicks_to_advance: 1
    drop_off_observed: 10-15% # baseline
  - step: contact
    fields_required: [email]
    clicks_to_advance: 1
    drop_off_observed: 8-12%
  - step: shipping_address
    fields_required:
      [first_name, last_name, address1, city, state, zip, country, phone]
    clicks_to_advance: 1
    drop_off_observed: 15-25% # biggest leakage in poor flows
  - step: shipping_method
    fields_required: []
    clicks_to_advance: 1
    drop_off_observed: 5-10%
  - step: payment
    fields_required: [card_number, exp, cvc, billing_address]
    clicks_to_advance: 1
    drop_off_observed: 10-15%
  - step: review_and_pay
    fields_required: []
    clicks_to_advance: 1

Each step = a drop-off opportunity. Goal: ≤ 3 steps mobile, ≤ 4 steps desktop.

VALIDATION: Step map captures real flow, not assumed. Field count per step is exact.

============================================================ === PHASE 2: EXPRESS CHECKOUT AUDIT ===

Audit checks:

• Express buttons render BEFORE shipping form (above the fold)
• Apple Pay button correctly displays on Safari + iOS (test in actual Safari, not Chrome dev tools — Apple Pay uses Webkit API)
• No "and / or" framing that buries express options
• Express buttons have correct styling (Apple's brand guidelines specific)

VALIDATION: Express checkout available at the FIRST point of checkout, not buried at the bottom.

============================================================ === PHASE 3: GUEST CHECKOUT + ACCOUNT TIMING ===

Forced account creation is the #2 abandonment cause (24% per Baymard).

Required:

• Guest checkout offered prominently, NOT as a small link.
• Account creation offered AFTER purchase ("Create account in one click — your details are saved").
• Returning user lookup by email/phone (offer to sign in, but allow guest).
• Single email field, not "email" + "confirm email" (12% drop when duplicated).

VALIDATION: Guest checkout is at least as prominent as account checkout in the UI.

============================================================ === PHASE 4: FORM FIELD MINIMIZATION ===

Every additional field drops conversion ~7% (Baymard). Audit each field:

Eliminate:

• Phone number unless required for shipping (carrier delivery notifications use email)
• Company name unless B2B
• Address line 2 default-hidden (show on click)
• Birthdate unless age-restricted product

Auto-fill / inference:

• City + state from ZIP (US: zipcode-to-city API)
• Country from IP geolocation (let user change)
• Billing same-as-shipping (default checked)

Smart validation:

• Inline error messages, not page reload
• Email format check on blur
• Credit card type auto-detect (no manual selector)
• Address validation via Google Places / Stripe Address Element (typeahead reduces error rate 50%)

VALIDATION: Total checkout field count ≤ 8 for a US guest flow.

============================================================ === PHASE 5: PAYMENT METHOD COVERAGE ===

Region-aware payment menu:

VALIDATION: Region detection (IP / billing country) drives payment menu. Don't show iDEAL to US buyers; don't show Pix to EU.

============================================================ === PHASE 6: COST TRANSPARENCY ===

Unexpected shipping costs = #1 abandonment cause (48% per Baymard). Audit:

• Shipping estimate shown ON cart page (not first-revealed at checkout)
• Free shipping threshold messaging ("Spend $15 more for free shipping")
• Tax included or labeled clearly (some regions show tax-inclusive prices)
• Customs / import fees disclosed for international (DDP vs DDU)
• No "service fees" or "convenience fees" introduced at the last step

VALIDATION: Order total shown at checkout matches order total shown on cart. No surprise additions > 5%.

============================================================ === PHASE 7: MOBILE-SPECIFIC AUDIT ===

Mobile checkout has different failure modes:

• Tap targets ≥ 44×44 px
• Inputs have correct inputmode (numeric for card, email for email, tel for phone)
• No zoom-on-focus on inputs (set font-size: 16px minimum)
• Native autofill compatible (autocomplete attributes correct per WHATWG)
• Sticky CTA bar at bottom (always-visible "Complete order")
• No mid-checkout app-install prompts
• CardScan via device camera (iOS / Android Pay Request API)

VALIDATION: Test on actual iOS Safari + Android Chrome, not desktop dev tools.

============================================================ === PHASE 8: PRIORITIZED PUNCH LIST ===

Generate checkout_audit_{date}.md:

# Checkout Audit — {site} — {date}

## Current state

- Steps: N
- Express checkout: Apple Pay {present/missing}, Google Pay {present/missing}, Shop Pay {present/missing}
- Guest checkout: {prominent/buried/missing}
- Field count: N
- Mobile checkout: {smooth/issues}
- Estimated current abandonment: {x}%

## P0 — Highest impact

- [ ] Add Apple Pay above fold. Expected lift: +22% conversion. Effort: 2 hours via Stripe Payment Element.
- [ ] Enable guest checkout. Expected lift: +25% from forced-account abandonment population. Effort: 4 hours.

## P1 — High impact

- [ ] Reduce form fields from 14 to 8. Expected lift: +35%. Effort: 8 hours UX + backend.
- [ ] Show shipping estimate on cart page. Expected lift: +20% (eliminates #1 abandonment cause). Effort: 6 hours.

## P2 — Medium

- [ ] Move from 4-step to 1-page checkout. Expected lift: +20%. Effort: 2-4 days.
- [ ] Add Klarna/Afterpay/Affirm for AOV > $50. Expected AOV lift: +15%. Effort: 1 week.

## P3 — Polish

- [ ] Sticky bottom CTA on mobile. Effort: 2 hours.

## Re-audit after fixes

Re-run `/checkout-audit` after each P0/P1 ships.

VALIDATION: Each item has expected lift (numeric) + effort + concrete fix.

============================================================ === SELF-REVIEW ===

• Complete: All 8 phases run? Step map + express + guest + fields + payment + cost + mobile + punch list?
• Robust: Tested in actual Safari/iOS + Android Chrome, not just desktop dev tools?
• Clean: Punch list prioritized by impact × effort, not by alphabet?
• CRO-credible: Would a Baymard-trained CRO consultant accept this audit as production-ready?

Common gap: missing actual mobile test → desktop dev-tools "looks fine" while real Safari has zoom-on-focus killing 30% of mobile.

============================================================ === LEARNINGS CAPTURE ===

~/.claude/skills/checkout-audit/LEARNINGS.md.

============================================================ === STRICT RULES ===

• Never audit without walking the actual flow. Code review alone misses runtime issues.
• Never report "fix XYZ" without measuring impact. Every recommendation needs an expected lift.
• Never recommend disabling guest checkout to "capture more leads." Capture lift is < lift from removing the gate.
• Never display payment methods irrelevant to the user's region.
• Always test on real mobile devices, not browser device emulation.