Scope: what this audit covers and what it does not
The question this audit answers is whether an online store turns the visits it already gets into orders and repeat customers, and which weakness in the store costs the most. It covers the whole store from the shopper's side and from the order data: whether purchases are recorded correctly, how shoppers move from listing pages to a product they want, whether product pages, images and reviews answer the questions that stop a purchase, whether delivery, returns and prices are promised honestly and kept, how much friction the checkout adds, how order value is raised without eroding margin, and whether the first order leads to a second.
It does not re-teach the work. The e-commerce marketing hub maps the guides for running a store, and each row below links the guide that owns its method. A single page that receives campaign traffic is scored by the landing page audit. A heuristic walk through one journey, with severity and confidence for each issue, is the expert review tutorial; this scorecard rates the store as a whole and ranks the fixes.
The product feed and the catalogue campaigns built on it are scored by the product feed and Performance Max audit, and email in depth, from consent to flows and reporting, by the email programme audit. Every scorecard sits in the audit library.
Before you score: access and evidence
Ask for analytics access and an export from the order system. Without both, the tracking gate cannot be scored and every funnel row stays provisional. Fix an evidence windowThe fixed period of data an audit scores, agreed before scoring, long enough to cover a full cycle for the business and kept the same at every rescore. that covers a normal trading period rather than a single sale, long enough for some first-time customers to have had the chance to order again, and keep the same window for every rescore.
Collect:
- analytics access with the purchase funnel from product view to order for the window;
- a walk-through of the store on a phone and on a desktop, from a category page to a completed test order;
- the delivery, returns and pricing policies as published, and what operations actually does;
- order data with order value, items per order, returns and repeat orders by first-order cohort;
- the live post-purchase messages and the review collection set-up;
- the promotion calendar and the margin behind each offer.
Write down the margin the business works to before you score the order value rows. The break-even calculator helps check what a promotion or a free delivery thresholdThe order value above which an online store delivers for free, set to encourage larger baskets without giving away too much margin. has to achieve to pay for itself on Contribution MarginRevenue left after every cost that varies with each sale., and the conversion rate calculator helps compare the store's own steps and periods with each other.
Criteria and scoring
Rate each row missing (absent or broken), partial (present but incomplete, inconsistent or out of date) or solid (matches the "what good looks like" column). The scale is your judgement against that column, not a number. Rows marked gate decide the band on their own when they score missing.
Tracking
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Purchase tracking (gate) | Every order is recorded once in analytics with its value and items, and the totals agree with the order system | Missing: orders doubled, missing or without values / Partial: recorded, totals drift from the order system / Solid: once per order, totals agree |
| Funnel measured by step | Each step from listing to order can be read, so the leak can be found rather than guessed | Missing: only sessions and orders / Partial: some steps tracked / Solid: every step readable by device |
The GA4 e-commerce tracking tutorial sets out the events and how to verify them, and the funnel analysis tutorial shows how to read each step and turn the leak into a hypothesis.
Finding and choosing products
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Listing pages | Category and listing pages let shoppers filter, sort and compare on the facts that decide the purchase | Missing: no useful filters, cards without the deciding facts / Partial: filters exist, not the ones buyers choose by / Solid: filters and cards built around the purchase |
| Product page answers | Each product page answers what stops a purchase: what it is, size or fit, price, delivery and returns | Missing: specifications only / Partial: answers present, delivery or returns hidden / Solid: every deciding question answered near the button |
| Product images | Images show the product as it will arrive, in use and at scale, consistently across the range | Missing: a single image, scale unclear / Partial: good for best-sellers only / Solid: consistent across the range |
| Reviews near the decision | Ratings, photos and questions sit near the price and the button, with negative reviews left visible | Missing: no reviews or only selected ones / Partial: reviews below the fold / Solid: near the decision, negatives visible |
Listing pages are scored against the category and listing pages guide, product pages against the product page optimisation guide, images against the product photography brief tutorial and reviews against the reviews and social proof guide, which covers negative reviews and incentives within the rules regulators set.
Promises
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Delivery promise kept (gate) | Delivery dates and costs shown before checkout are ones the operation keeps | Missing: costs or dates first shown at the last step, or promises routinely broken / Partial: shown early, kept unevenly / Solid: stated the same way everywhere and kept |
| Returns policy | The returns policy is clear, findable before purchase and matches what the business actually does | Missing: hard to find, or different from practice / Partial: findable, unclear wording / Solid: clear, near the decision, matched by practice |
| Honest prices and urgency | Reference prices were genuinely charged, and countdowns and stock warnings are true | Missing: permanent sales or countdowns that reset / Partial: mostly true, some stale messages / Solid: every claim true when shown |
The delivery promises guide treats the promise as something set with operations and measured as the share kept. The tutorial on measuring whether delivery promises are kept builds that share from the order, dispatch and delivery records, which is the evidence for scoring the gate. The returns row follows the returns policy and experience guide, which also explains how returnsShare of orders or units sold that customers sent back. belong in marketing results. Prices and urgency are scored against the honest persuasion guide, which draws on the CMA's general direction and is educational, not legal advice.
Checkout
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Checkout friction | Costs are clear before the last step, guest checkout is offered and fields and steps are only those the order needs | Missing: forced account creation or late costs / Partial: guest checkout, extra fields / Solid: costs early, guest option, short form |
| Confirmation page | The confirmation page confirms the order, sets expectations and offers a sensible next step | Missing: a bare thank-you / Partial: confirms, sets no expectations / Solid: confirms, says what happens next, one next step |
The checkout optimisation guide works through the leaks usability research has documented, which drive basket abandonmentShare of shopping carts created that did not become an order.; the confirmation pages guide covers the page every new customer sees.
Order value
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Order value levers | Bundles, thresholds and cross-sells are set from margin and checked against what customers buy | Missing: no levers, or levers that lose margin / Partial: levers set by habit / Solid: set from margin and reviewed |
| Promotion discipline | Each promotion has a job and a break-even written before launch, and the calendar avoids training customers to wait | Missing: promotions as the calendar / Partial: jobs written, no break-even / Solid: a job, a break-even and full-price weeks protected |
Levers are scored against the average order value levers guide, which judges each on profit per visitor rather than on AOVAverage revenue per order. alone, and promotions against the guide to planning promotions.
After the first order
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Post-purchase messages | Order, dispatch, delivery, review request and replenishment messages exist and are kept current | Missing: an order receipt only / Partial: some messages, out of date / Solid: each message live and reviewed |
| Repeat purchase read by cohort | Repeat purchase is measured by first-order cohort and encouraged with reasons to return, not discounts alone | Missing: not measured / Partial: a store-wide figure, discounts as the only reason / Solid: read by cohort, reasons to return beyond price |
| Packaging and inserts | The parcel and anything in it carry a reason to return or review that suits the brand | Missing: nothing in the box / Partial: a generic flyer / Solid: one clear ask that suits the brand |
Messages are scored against the e-commerce email and SMS flows guide, repeat purchase against the repeat purchase and loyalty guide, which explains why the second order is the hardest and how Repeat Purchase RateShare of customers who placed more than one order in the period. is read, and the parcel against the packaging and order inserts guide.
How to score and prioritise
Score the two gates first. If analytics does not agree with the order system, the funnel, conversion and order value rows are provisional: what looks like a leaking step may be a tag that fires twice or not at all. If the delivery promise is broken, every order marketing adds also adds complaints, refunds and customers who do not come back. Fix both before anything else.
Then rank the remaining weaknesses in this order:
- The leak the funnel shows: the step where most shoppers drop is fixed before pages that already work.
- Promises and prices that risk trust or the rules: returns wording, reference prices and countdowns.
- Order value and repeat purchase, because they compound once orders flow.
- Image and packaging polish last.
Within each group, rank by the orders the weakness touches. A fix on the product template that carries most of the range outranks the same fix on a single page.
Reading the result: what to do with each band
Red means a gate scored missing. Hold new traffic campaigns, repair purchase tracking until analytics agrees with the order system, and agree a delivery promise with operations that is shown early and kept. Rescore after a full month on the repaired set-up.
Amber means the gates hold at least partially but several other rows are partial or missing. This is the usual result. Work a short ordered list, starting at the step where the funnel leaks most.
Green means the gates are solid and most rows are too. Move effort into testing on product pages and checkout and into the repeat purchase programme, and rescore each season to catch drift.
From findings to a prioritised fix list
Write each partial or missing row as a finding in plain words: what you saw, the evidence and the orders it touches. Turn the findings into one fix list with a line per fix: the finding, the fix, the owner, the guide that explains it and the rescore date. Order the list by the priorities above, not by how easy each fix is.
Group the lines by who does the work. Tracking sits with whoever runs analytics and the tag manager, promises with operations and customer service, prices and promotions with the person who sets them, pages and checkout with whoever runs the store platform, and messages and inserts with the retention or email owner. Present the band, the gate findings and the first few fixes to the owner of the store, and keep the full scorecard as the working copy for the rescore.
Illustrative example
A kitchen knife shop audits its store after a strong season
Pizvarre Knives, an invented online kitchen knife shop, audits the store after a gift season that brought 38,000 visits and 1,100 orders but few repeat customers. Purchase tracking scores missing: the confirmation page fires the purchase event on every reload, so analytics shows 1,290 orders. The delivery promise scores missing too: the dispatch date and the £4.95 delivery charge appear only at the last checkout step, and about one order in eight arrived after the date given. Product page answers score partial, because blade length and steel type sit in a collapsed tab. Post-purchase messages score missing, since nothing follows the receipt, and repeat purchase has never been read by cohort. The band is red. The fix list puts a purchase event that fires once first, a delivery promise agreed with the warehouse and shown on product pages second, dispatch and care messages third and a cohort view of repeat purchase fourth. After a full season, analytics agrees with the order system, the gates score solid and the band moves to amber.
Common mistakes
- Reading the funnel before checking tracking. A step that seems to leak may be a tag that fires twice. Score the tracking gate first.
- Promising faster delivery to win the order. A promise the warehouse cannot keep costs more later than it wins now. Score the gate on what is kept, not on what is shown.
- Hiding negative reviews. Selective reviews mislead shoppers and undermine trust. Score the reviews row on what is visible.
- Treating a higher average order value as success. A threshold or bundle that lifts the average can still lower profit per visitor. Score levers on margin.
- Measuring repeat purchase as one store-wide figure. It mixes new and old customers. Score the row on a cohort view.
- Fixing everything at once. Simultaneous changes to pages, checkout and promotions hide which one moved orders. Work the list in order and rescore on the same evidence window.
