ecommerce-and-checkout-ux

name	ecommerce-and-checkout-ux
description	Use when designing or reviewing an e-commerce journey across PLP, PDP, cart, guest checkout, payment, returns, refunds, and order status with transparent cost and no dark patterns. Use form-ux-design for field anatomy and landing-page-and-conversion-design for campaign pages.
metadata	{"portable":true,"category":"06-sector-and-domain-ux","compatible_with":["claude-code","codex"]}

E-commerce & Checkout UX

Acknowledgement: Shared by Peter Bamuhigire, techguypeter.com.

A store has one job: let a specific shopper find the right thing, believe it's the right thing, and pay for it with zero friction and zero trickery — then trust you enough to come back. This skill gives the surface-by-surface architecture (PLP → PDP → cart → checkout → payment → post-purchase) and the conversion craft that converts because the path is clear and the price is honest, never because the shopper was rushed, confused, or opted-in without consent. The doctrine's Mission applies in full: a store that opens in Inter on a generic grid of identical cards tells the shopper no one authored it — and a hidden £4.99 "service fee" at step four tells them no one respects them. Both lose the sale and the repeat custom.

Use When

• You are building or rebuilding any part of a retail buying journey: a category/search PLP, a PDP, the cart/basket, the checkout flow, payment/address entry, or the post-purchase surface (order confirmation, status, returns/refunds).
• A funnel leaks and you need to diagnose where shoppers abandon — usually the cart→checkout seam (surprise costs, forced sign-up) or the payment step (too many fields, no trust cues).
• You are asked to "lift conversion" and the instinct (yours or a stakeholder's) is reaching for a countdown timer, a pre-ticked insurance add-on, an auto-renewing "membership," or a free-shipping threshold that quietly slips an item into the basket — STOP and apply the anti-dark-pattern rules here instead.
• You need to specify guest-first checkout, total-cost transparency, address/payment field minimisation, or a returns/trust surface that earns the next purchase.

Do Not Use When

• The task is a marketing/landing page ending at a single CTA, not a purchase flow — use 14-conversion-and-web-page-patterns/landing-page-and-conversion-design. That skill ends at the click; this one owns everything after it.
• The task is generic form field anatomy, validation, or wizard mechanics in isolation — use 04-web-and-ui-design/form-ux-design. This skill composes those patterns into the retail context (it does not re-teach label placement or the five field states — it cites them).
• The task is the ethics framework / dark-pattern taxonomy itself, independent of retail — use 00-cross-cutting-ops-qa-a11y/design-ethics-and-anti-dark-patterns (the paired skill). This skill applies that doctrine to the specific retail dark patterns; that one is the general catalog.
• The task is only the button/error wording — use 10-content-design-and-ux-writing/ux-writing-and-microcopy, then return here for placement.

Required Inputs

Input	Source	Evidence
Catalogue, price, inventory, delivery, tax, and returns rules	Commerce, finance, and operations owners	Representative products, totals, availability, and policy sources
Shopper segments and priority journeys	Research, analytics, and product	Search/browse intent, devices, locales, and conversion baseline
Payment, identity, fraud, and accessibility constraints	Engineering, risk, and compliance	Supported methods, failure codes, consent, and testing requirements

• The catalog reality. Roughly how many products, how they are categorised, what attributes matter for filtering (size, colour, price, brand), and whether variants exist (a shirt in S/M/L × 3 colours). You cannot design a PLP filter or a PDP variant picker for a catalog you haven't seen.
• The true total cost structure. Every component the shopper will pay: item price, tax/VAT (inclusive or added?), shipping tiers, any genuine fees. This is non-negotiable input — the honest-pricing rule below cannot be applied without it. If fees are "to be decided later," flag that the design will surface them early by construction.
• The fulfilment and returns truth. Real delivery options/times, the real returns window and process, who pays return postage. The trust surface must state what is actually true.
• Payment methods and any account/identity constraint. Which methods (card, wallet, BNPL, mobile-money), and whether anything genuinely requires an account (vs. wanting one for marketing — those are different, and only the first justifies gating).

Workflow

Map the buying journey from discovery through post-purchase, then remove friction without weakening price transparency, consent, cancellation, returns, or recovery.

0. Run the pre-flight, then commit the store's ONE distinctive idea

Run 04-web-and-ui-design/distinctive-by-design first — it gates the build. A storefront is a high-convergence surface: the reflexive output is Inter + a uniform grid of identical white cards + a generic "Add to cart" + an indigo accent, indistinguishable from every template. Commit ONE authored organising idea (a signature display face for product titles, an editorial PLP rhythm that breaks the uniform grid, an authored photography POV, a distinctive but legible price treatment) per doctrine/design-doctrine.md §0 and pairing-principles.md. State it before any markup. No banned fonts; design around real product data, never lorem.

1. PLP — let them find and compare honestly (see references/checkout-flow-patterns.md §PLP)

• Scent + scan. Each card carries the load-bearing decision data: real product image, name, price (the real price, with any "was" price only if genuinely the prior price), rating, and key variant/availability signal. Don't bury price; don't fake a crossed-out "RRP."
• Filters and sort that match the catalog, not a generic set. Filters must reflect the shopper's real decision attributes; show active filters as removable chips; never reset their selection silently. Keep results honest — sponsored items, if any, are labelled (disguised ads are a dark pattern, ref interaction-anti-patterns.md and the doctrine).
• No fake urgency on cards. "Only 2 left" appears only when it is true and live. Real low-stock is a legitimate, helpful signal; an invented one that resets on reload is banned.

2. PDP — earn the "Add to cart" with truth (see references/checkout-flow-patterns.md §PDP)

• The decision set above the fold: real imagery (zoomable, multiple angles), name, honest price with the all-in cost knowable (tax-inclusive where the market expects it; shipping at least estimable), variant picker with only in-stock combinations selectable (or clearly marked out-of-stock — never let them pick a phantom), and a single clear primary action.
• One primary action: Add to cart / Buy. Secondary actions (wishlist, share) are visibly subordinate (ref interaction-anti-patterns.md B3 — one primary action per view). The "Add to cart" must do exactly and only what it says — it never silently adds a warranty, an add-on, a subscription, or a second item (sneak-into-cart, banned).
• Believability content: real specs, real delivery estimate, real returns terms, real reviews (with the genuine distribution, not only 5-stars). Answer "is this the right thing and what happens if it's not?" on the page, not after payment.

3. Cart — show the true running total, add nothing unasked (see references/checkout-flow-patterns.md §Cart)

• Line items the shopper actually chose — never a pre-added accessory, insurance, or "popular add-on" ticked by default (pre-checked opt-in + sneak-into-cart, both banned).
• The running total is honest and complete-as-knowable: subtotal, then tax and shipping shown or clearly estimated here, not sprung at the final step. If shipping needs a postcode, ask for it in the cart so the total isn't a surprise later (drip pricing / hidden costs, banned — interaction-anti-patterns.md, doctrine §"Hidden costs").
• Free-shipping / threshold nudges are honest: "£6 from free shipping" is fine; auto-adding a filler item to cross the threshold is theft-by-friction. Edit-quantity and remove must be obvious and reversible (a removed item is undoable — ref interaction-anti-patterns.md C3).

4. Checkout — guest-first, minimal fields (see references/checkout-flow-patterns.md §Checkout)

This is the conversion-critical seam. Apply form-ux-design for field mechanics; apply these retail rules on top:

• Guest checkout is offered first and given equal or greater visual weight than "Sign in / Create account." A forced-registration wall before purchase is the single highest cart-abandon cause and a banned pattern (ref interaction-anti-patterns.md B1 "Forced Sign-Up", doctrine). Offer account creation after the order, as a one-tap "save these details" using data they already entered — never as a gate.
• Minimum field set. Ask only what fulfilment and payment truly need. Use autocomplete attributes, address lookup/autofill, correct input types and mobile keyboards, single-column, labels above (ref form-ux-design Rules 1–5; interaction-anti-patterns.md B2 Form Friction). Default billing = shipping with a single toggle to differ.
• A real progress signal on a multi-step checkout (or a single-page accordion), so the shopper knows how many steps remain. Inline, solution-bearing validation — never a late "card invalid" wipe.
• No surprises before the pay button: the final total (items + tax + shipping + any fee) is shown in full adjacent to the pay action, with the delivery date and the returns/cancellation terms one glance away.

5. Payment & address — trust at the moment of money (see references/checkout-flow-patterns.md §Payment)

• Recognised payment methods up front (wallets/express pay that skip address entry are a legitimate friction-cut — Apple/Google Pay, mobile-money where regional). Express pay must still show the full total before confirming.
• Trust cues that are real: the padlock/secure-checkout signal, recognised card/wallet marks, and a clear statement of what is charged and when. For any saved-card or wallet, the charge is exactly the displayed total — no padding.
• Address entry minimised via lookup; format and validate per locale; show errors inline next to the field. Card fields use the right input mode, auto-detect card type, and never block paste.

6. Subscriptions, warranties & add-ons — opt-IN, symmetric exit (see references/checkout-flow-patterns.md §Continuity)

• Every recurring charge, warranty, insurance, or add-on is unchecked by default and the shopper opts in deliberately (pre-checked = banned). State the renewal price, the cadence, and the cancel path at the point of opt-in.
• Forced continuity is banned. A "free trial" that auto-charges must say so plainly up front, remind before charging, and offer a cancel that is as easy as the sign-up (no roach motel — ref interaction-anti-patterns.md C4 Trap, doctrine "Forced continuity" / "Roach motel").

7. Post-purchase — returns, refunds, status, the trust that earns the next sale (see references/checkout-flow-patterns.md §Post-purchase)

• Order confirmation states what was bought, the all-in amount charged, the delivery estimate, and the returns window — no dead end (ref interaction-anti-patterns.md C3).
• Returns/refunds are findable and symmetric: initiating a return is no harder than buying was; the policy is stated in plain language before purchase, not hidden. A genuine, clearly-stated returns policy is one of the strongest honest conversion levers in retail.
• Order status is visible without a support ticket; status changes are communicated; refunds confirm timing in writing.

8. Specify, then build

Produce a surface-by-surface spec (see examples/checkout-flow-spec-worked.md for the shape): per surface — purpose, the real content/fields, the type/space treatment, the one primary action, the total-cost treatment, and the trust cue. Run the checklist below before any markup.

Anti-Dark-Pattern Checklist (run before building — every item must pass)

1. Guest checkout offered first, equal-or-greater weight than account creation; no forced-registration wall before purchase (B1).
2. Total cost shown in full before the pay button — items + tax + shipping + every fee; nothing sprung at the final step (no drip pricing / hidden costs).
3. Cart contains only what the shopper explicitly chose — no sneak-into-cart, no auto-added filler to hit a free-shipping threshold.
4. All add-ons, warranties, insurance, and subscriptions are unchecked by default — opt-in only, with price/cadence/cancel stated at opt-in (no pre-checked opt-ins).
5. Any recurring charge states terms up front, reminds before charging, and cancels as easily as it signs up (no forced continuity, no roach motel).
6. Scarcity/urgency is real and live — true stock counts and genuine deadlines only; nothing resets on reload (no fake scarcity/urgency).
7. Decline/opt-out wording is neutral — "No thanks," not "No, I don't want to save money" (no confirmshaming).
8. Sponsored/ad placements in PLP are labelled — never disguised as organic results.
9. Minimum field set, autocomplete/address-lookup, single-column, labels above, inline validation; billing=shipping default (B2 form friction).
10. One primary action per surface; "Add to cart"/"Pay" does exactly and only what it says (B3).
11. Returns/refund policy stated in plain language before purchase, and initiating a return is no harder than buying (no dead end, symmetric exit).
12. Checkout passes WCAG 2.2 — focusable controls, visible focus, ≥24×24px targets (SC 2.5.8), sufficient contrast, never colour-only for stock/error/required state.
13. No convergent default — type/colour/layout pass distinctive-by-design (no Inter + uniform white-card grid + indigo accent as a reflex).

If any item can't be satisfied (a stakeholder insists on a pre-checked add-on, fees can't be shown early, "make cancel hard"), refuse and offer the honest equivalent — never quietly ship the coercive version.

Decision Rules

Condition	Choice	Wrong-choice failure
Shopper can buy without a durable account need	Guest checkout first, account after purchase	Forced registration increases abandonment and data collection
Total cost varies by address/method	Show known subtotal early and explain remaining variables	Late fees destroy trust and create dark-pattern risk
Product choice requires comparison	Stable attributes, filters, and comparison support	Decorative cards hide meaningful differences
Payment outcome is uncertain	Idempotent pending state with safe status recovery	Blind retry can duplicate charges or orders

Capability Contract

• Must inspect commerce and payment contracts and representative journeys; review is read-only unless implementation is requested.
• May edit/test in-scope storefront UI, but must not place real orders, change prices/policies, process payment data, or publish without authority.

Degraded Mode

• If price, fulfilment, returns, or payment-failure rules are missing, stop checkout sign-off and issue a blocker list.
• Without a runnable store, produce a state/total specification marked unverified. Recover failures by preserving the basket, preventing duplicate submission, exposing order/payment status, and testing the safe next action.

Quality Standards

• Totals, availability, delivery, consent, and return terms remain truthful and visible before commitment.
• Release evidence covers search-to-order, guest checkout, payment failure/pending, duplicate prevention, accessibility, and post-purchase recovery.

Anti-Patterns

• Forced account creation before checkout — the highest-impact abandon driver; guest-first is non-negotiable (ref interaction-anti-patterns.md B1).
• Drip pricing / hidden costs — fees, tax, or shipping revealed only at the final step. Bait- and-switch; invites abandonment, chargebacks, and (in the EU/UK/US) regulatory action.
• Sneak-into-cart — items, warranties, or "samples" added without an explicit choice.
• Pre-checked add-ons / insurance / opt-ins — consent by inertia; banned regardless of revenue.
• Forced continuity — a "free trial" that auto-charges with a buried cancel (roach motel).
• Fake scarcity / fake urgency — invented "only 2 left" or countdowns that reset on reload.
• Confirmshaming the decline ("No thanks, I hate good deals").
• A wall of equal-weight CTAs on the PDP/cart — the buy action drowns (ref interaction-anti-patterns.md D6 / B3).
• Too many fields, no autofill, late validation at the payment step (ref B2).
• The convergent storefront template — Inter, a uniform grid of identical cards, generic "Add to cart," indigo accent. Looks unauthored; the Mission targets it.
• A returns policy hidden until after purchase, or a return process harder than the purchase was — a trust killer and a dead-end (C3).

Outputs

Output	Consumer	Evidence and acceptance
Commerce journey and state specification	Product, design, engineering	PLP/PDP/cart/checkout/payment/post-purchase rules and totals are explicit
Checkout quality gate	Commerce, risk, accessibility QA	Representative devices/methods/states pass without hidden cost or blocked recovery

• A stated store/journey goal and the committed distinctive idea (what + why + human authority), written before markup.
• A surface-by-surface spec (PLP → PDP → cart → checkout → payment → post-purchase): per surface the purpose, real content/fields, type/space treatment, single primary action, total-cost treatment, and trust cue — in the shape of examples/checkout-flow-spec-worked.md.
• A completed Anti-Dark-Pattern Checklist confirming guest-first, total-cost transparency, zero pre-checked/sneak/forced-continuity patterns, and a symmetric returns path.

Examples

• examples/checkout-flow-spec-worked.md — a complete, build-ready surface-by-surface spec for a real store (a small specialty-coffee retailer): the PLP card and filters, the PDP variant picker and all-in price, the honest cart with early shipping, the guest-first single-page checkout with the minimum field set, the payment/trust step, the opt-in-only subscription, and the post-purchase returns surface — with the WCAG 2.2 and anti-dark-pattern constraints applied throughout.

References

• references/checkout-flow-patterns.md — the surface-by-surface retail spine (PLP, PDP, cart, checkout, payment/address, continuity, post-purchase), the guest-first checkout pattern, the total-cost-transparency rule, the full retail dark-pattern catalog with honest swaps, and the field-minimisation rules. Cites interaction-anti-patterns.md and the doctrine throughout.
• doctrine/design-doctrine.md — the Mission (§0, "the moat is looking human-made and trustworthy"): conversion is downstream of an authored, honest store. The Anti-Slop Charter bans the convergent storefront template.
• doctrine/references/interaction-anti-patterns.md — the behavioural detection checklist this skill enforces: B1 Forced Sign-Up, B2 Form Friction, B3 Vague/competing CTAs, C3 Dead End / no recovery, C4 Trap / Forced Action, D6 Oceans of Buttons.
• doctrine/references/wcag-2.2-criteria.md — checkout control focusability, target size (SC 2.5.8), contrast, non-colour state for stock/error/required.
• doctrine/references/web-performance-budgets-2026.md — PLP/PDP imagery as LCP, CLS reservation for product media; a slow storefront is an abandoned one.
• Sibling skills: 04-web-and-ui-design/distinctive-by-design (run first; gates the build), 04-web-and-ui-design/form-ux-design (field anatomy, validation, the five field states — applied inside checkout), 14-conversion-and-web-page-patterns/landing-page-and-conversion-design (the page before the click; this skill owns after), 00-cross-cutting-ops-qa-a11y/design-ethics-and-anti-dark-patterns (paired — the general ethics/dark-pattern doctrine this skill applies to retail), 10-content-design-and-ux-writing/ux-writing-and-microcopy (button/error/empty wording).