Where stores lose the sale
The store's FunnelModel of the customer journey stages. has fewer stages than a lead-generation funnel, but each one has its own way of losing the visitor.
| Stage | What the visitor is doing | Where the sale is lost | Guide |
|---|---|---|---|
| Landing | Arriving from an ad, a search result or a link | Page does not match what was promised; slow load on mobile; pop-up before the product | Product page optimisation, Speed and mobile |
| Finding | Browsing categories, using search, filtering | Search returns nothing; filters hide stock; categories too broad | This page, Faceted navigation |
| Evaluating | Reading the product page, looking at images, checking size and delivery | Unanswered questions; poor images; price or delivery unclear; no proof | Product page optimisation, Reviews and social proof |
| Basket | Adding, reviewing, considering more | Delivery cost surprise; no reason to add more; basket hard to edit | Average order value levers |
| Checkout | Entering details and paying | Forced account; long forms; missing payment method; errors | Checkout optimisation |
| Confirmation | Reading the order summary | Doubt about delivery; no next step | Retention |
Baymard Institute's long-running usability research on e-commerce sites documents recurring problems at each of these stages, from product page layout to checkout form design, and is the reference this section draws on for what buyers struggle with; it is cited in sources rather than quoted as figures here. To rate every stage at once and rank the fixes, the online store audit scorecard scores the store from listing pages to repeat purchase.
Product pages, order value, proof: the core guides
The core guides in this section cover the stages where a store's own decisions make the most difference.
Product page optimisation treats the product page as a set of questions the buyer needs answered before the button: what the product is, what it looks like in use, what it costs in total, when it will arrive, whether it fits or is compatible, and whether people like them have been happy with it. It covers images and video, titles and descriptions, price and delivery clarity, variants and sizing, trust near the button, and how to test changes. The SEO side of the same page (ranking, structure, structured data) belongs to the category and product pages guide.
Average order value levers covers the ways a store raises the value of each order (AOVAverage revenue per order.): delivery thresholds, bundles and kits, cross-sells and upsells, volume pricing and financing, and the trade-off each makes with conversion rate. It is the quietest lever a store has, because it raises revenue without a single extra visitor.
Reviews and social proof covers the store's review programme: collecting reviews in a way that is both effective and within the rules, timing the request, displaying ratings, photos and questions where doubt arises, handling negative reviews, and review markup. The general theory of proof elements belongs to the trust and proof guide in the CRO section. Ratings, reviews and customer photos are the store's main forms of social proofEvidence that other people have chosen, used or approved something, used to reduce a buyer's doubt..
Further guides
Further guides cover the pictures, the promises, the occasions and the choices that shape what the shopper sees:
- Merchandising the store for a campaign — choosing the products, the landing collection, homepage and menu slots, one price everywhere, and taking the campaign down.
- Briefing product photography for an online store — a tutorial for the brief: a shot listThe list of pictures a shoot must produce, built from the questions buyers ask about a product: angles, variants, scale, details and in-use shots. built from buyers' questions, variant and scale shots, feed-safe main images, file names and review.
- Selling gifts online — gift guides by recipient and budget, gift sets, wrapping and messages, last order dates, gift returns and the giver.
- Setting and keeping delivery promises — promises the store can keep, stated the same way on pages, feeds and ads, peaks and delays, and measuring promises kept.
- Measuring whether delivery promises are kept — a tutorial that joins the promise shown with each order to its dispatch and delivery dates, counts the share kept, splits each miss between the store and the delivery service and turns the pattern into a change.
- Clearing excess stock without cheapening the brand — choosing the route, containing the sale, honest reference prices, ads and feeds, and buying less next time.
- Product finder quizzes for online stores — when a finder beats filters, questions from what decides the purchase, honest results, consent and measuring whether it sells.
- Selling made-to-order and personalised products — showing the finished item, capturing personal details without errors, lead times, updates and terms stated before purchase.
Checkout and the CRO section
Checkout is where the most motivated visitors are lost, and it is deliberately not covered in this section, because it is a funnel problem that every site with a form shares. The checkout optimisation guide in the CRO section owns it: guest checkout, form length and field design, address and payment entry, error handling, delivery choice, and the trust signals that belong beside the pay button. The funnel analysis guide covers how to find the step where visitors leave, and the CRO hub covers research, hypotheses and testing in general.
What this section adds is the store-specific context. A checkout cannot fix a delivery cost that was hidden until the last step; that is a product page and basket problem, covered here. A checkout cannot recover a visitor who could not find the product; that is a search and navigation problem, covered below. And a checkout's conversion rate means nothing without the segments it is read in. Treat the CRO guide as the manual for the form and this section as the manual for everything that leads to it.
Search, navigation and filters
Visitors who use on-site search usually know what they want, which makes them the most valuable visitors a store has and the ones most often failed. The failures are mundane: a search for a plural or a misspelling returns nothing; a synonym the catalogue does not use returns nothing; a search for a product code works but a search for the product's common name does not; results are ordered by relevance rules that put out-of-stock items first. Each is fixable with the search log, which lists what people typed and how many results they got. Zero-result searches are the first report to read, because each one is a visitor who said what they wanted and was told the store did not have it. The site search analysis tutorial shows how to build that report in GA4, with a results count, a search terms exploration and the exits and refinements that follow.
Navigation and filters do the same job for people who browse. Categories should be shaped around how buyers think (by use, by recipient, by room, by size) rather than how the warehouse is organised, and filters should expose the attributes that decide the purchase, with counts, with the ability to combine and remove them, and without hiding items that are low in stock. The crawl and indexing side of filters is a separate problem with its own faceted navigation guide; the conversion side is simply whether a buyer can get from a category of hundreds of products to the few that suit them in a handful of taps on a phone.
Measuring conversion by segment
A store's conversion rate as a single number hides more than it shows. The conversion rate calculator gives the arithmetic, but the useful work is in the splits: by device, because mobile and desktop behave differently and most traffic is mobile; by traffic source, because a shopping ad click and a newsletter click arrive with different intent; by new against returning visitors, because returning customers convert far better and a shift in the mixA change in a blended metric caused by a change in the mix of segments, such as devices or sources, rather than by any segment performing differently. moves the blended rate without anything on the site changing; by product category and price band, because a store selling both consumables and furniture has two conversion rates, not one; and by landing page type, because category, product and editorial pages do different jobs. The new-against-returning split here describes visitors; whether a buyer is a new customer is a fact from the order system, which the guide to new and returning customers covers.
Three practices keep the measurement honest. Report each key performance indicator (KPIA metric tied directly to an objective.) with its segment, never alone. Pair conversion rate with order value and margin, because a change that raises one and lowers another may be neutral or worse. And treat movements as hypotheses: a drop in mobile conversion after a release is a bug until proved otherwise, and a rise after a campaign may be a mix shift. The conversion analytics material covers the reporting, and the A/B testing guide covers proving that a change caused a difference rather than assuming it. The wider picture, where conversion sits between acquisition and retention, is in the e-commerce marketing hub.
One conversion rate, three stories
A kitchenware store reports a site conversion rate of 2.1%, down from 2.6% the previous quarter, and the team suspects the new product page template. Split by segment, the picture changes. Desktop conversion is 3.4%, unchanged; mobile is 1.6%, down from 2.2%. Returning visitors convert at 5.8%, also unchanged, but their share of sessions fell from 40% to 28% because a paid social prospecting campaign doubled new-visitor traffic. Among new mobile visitors from that campaign, conversion is 0.7%, and the search log shows 1,900 zero-result searches in the month, mostly for a brand the store does not stock but the ads implied it did. The template is cleared; the fixes are a landing page for the campaign, a synonym rule in search, and a decision to judge the campaign on new-customer margin rather than on the blended site rate.
Common conversion problems
| Symptom | Usual cause | Where to look |
|---|---|---|
| High product page views, low add-to-basket | Unanswered questions; poor images; unclear price or delivery; no proof | Product page optimisation, Reviews and social proof |
| Baskets abandoned at the delivery step | Delivery cost or time revealed late; no threshold communicated | Average order value levers, Checkout optimisation, Delivery promises |
| Checkout started, not completed | Forced account; long or confusing forms; missing payment method; errors on mobile | Checkout optimisation |
| Conversion fine on desktop, poor on mobile | Slow load; layout that pushes price and button below the fold; filters unusable by thumb | Speed and mobile |
| Site search sessions converting worse than browse | Zero-result searches; poor ranking of results; out-of-stock items first | This page |
| Conversion rate moves with no site change | Traffic mix shift; new campaign; returning share changed | This page, Funnel analysis |
| Order value falling as conversion rises | Discounting; a promotion attracting single-item buyers | Average order value levers; diagnosis: average order value dropping |
| Orders arrive later than the store promised | Promise set by marketing or on an average day; different promises on page, feed and ads | Delivery promises, Measuring whether delivery promises are kept |
| Campaign visitors land and leave | Traffic sent to the full category, not a landing collection; page and feed prices differ | Merchandising for a campaign |
| Returns for colour or size; questions about what the product looks like | One image for every variant; no consistent scale reference | Briefing product photography |
| Gift orders miss the occasion | Last order dates hopeful rather than true | Selling gifts online |
| Customers wait for the sale before buying at full price | Clearance on the same dates every year; a sitewide code to clear a few lines | Clearing excess stock |
| Shoppers compare many similar products and leave without choosing | A range too wide or technical for filters alone; no help for shoppers who do not know the attribute names | Product finder quizzes, Category and listing pages |
| Personalised orders remade, returned or complained about | Details typed without a preview or a check; one delivery promise for made-to-order items | Made-to-order and personalised products |
The guides in this section carry the store-specific reasoning; the CRO section carries the general method for research, hypotheses and testing, and the acquisition hub covers whether the visitors arriving were the right ones in the first place.
For a symptom-first diagnosis of store problems such as falling order values or buyers who wait for the sale, the e-commerce troubleshooting section works through each cause with a check, a fix, a verification step and a prevention habit.
