E-commerce CRO Guide

Where e-commerce conversion is actually decided

Most stores measure conversion as one number: orders divided by sessions. The number hides where the loss happens. A typical funnel loses visitors in four places, and each one has its own fixes. The collection page loses shoppers who cannot find or judge a product. The product page loses shoppers who cannot decide. The cart loses shoppers who hesitate at the total. The checkout loses shoppers who hit friction or doubt. Baymard Institute's long-running checkout research puts average documented cart abandonment near 70%, and the reasons shoppers give have barely moved in a decade: extra costs too high, account creation required, checkout too long, total not visible up front, not enough trust, not enough payment options. Work backwards from that list and you have an e-commerce CRO program.

Product page: the decision screen

The product page has one job: make the shopper confident enough to press Add to cart. Everything on it either builds that confidence or competes for attention with the button. On mobile, the first screen usually shows a gallery and a title; price and the button are one or two scrolls down, and a large share of visitors never make it there. The fixes are structural rather than cosmetic: price and Add to cart in the first viewport, a sticky Add to cart when the button scrolls away, variant selection before the button rather than after a failed click, and delivery cost and time stated on the product page rather than discovered in checkout. Reviews belong next to the price with a count and an average, not in a tab. Returns belong in one sentence next to the button. Images should show scale, context and detail, because the shopper cannot touch the product and the photos are doing that job.

When you run a predictive attention heatmap on a product page, the most common finding is that the hero image or a promotional banner carries more visual weight than the button. Reducing the noise around the button usually does more than making the button bigger.

Collection and search: the finding screen

Collection pages convert indirectly: their job is to get the right shopper to the right product page quickly. Filters should match how shoppers think (size, price, colour, use), not how the catalogue is organised internally. Each card needs a single clear image, the price, and at most one badge; a wall of badges makes every product look the same. Sort defaults to relevance, with best sellers and newest one tap away. Empty search results should suggest categories instead of a blank page, because a shopper who searched has told you exactly what they want.

Cart: the hesitation screen

The cart is where the shopper sees the total for the first time. The cart should confirm what was added, show the total with a shipping estimate, and present one primary Checkout button. Free-shipping thresholds work when the progress is visible. Discount code fields should not dominate the layout; a large empty code box sends shoppers off to search for a coupon and a third of them do not come back. Cross-sells go below the checkout button, never above it, and are limited to a few relevant items.

Checkout: the friction screen

Guest checkout is the single most-cited fix in abandonment research; offer account creation after the order, when the shopper has already committed. Express payment options at the top of checkout remove the form entirely for a growing share of shoppers. Show shipping options and cost before asking for payment details. Keep the order summary visible on mobile without an extra tap. Put reassurance next to the payment fields, where the doubt is, rather than in the footer. Every form field should have a visible label, the correct mobile keyboard, autocomplete enabled and inline validation.

Speed, layout shift and the apps you installed

Performance is a conversion lever, not a technical nicety. A product page whose Largest Contentful Paint exceeds 2.5 seconds on mobile is losing shoppers before it has said anything. Layout shift is worse: a review widget or banner that loads late and moves the Add to cart button as the shopper reaches for it causes a mis-tap and a bounce. Most of this comes from third-party apps and scripts added over months. Audit them by the question "does this still earn its weight" and remove the ones that do not.

Trust and persuasion, in the right order

Trust is a placement problem. Reviews, guarantees, secure payment badges and contact details work when they sit at the moment of hesitation: next to the price, next to the button, next to the payment fields. The same elements in a footer are invisible. Persuasion levers help when they are true: real stock levels, a real delivery cutoff, a real number of reviews. Countdown timers that reset and "only 3 left" that never changes train shoppers to distrust everything else on the page.

How to run e-commerce CRO as a program

1. Instrument the funnel: sessions, product views, add to cart, checkout started, orders. Read the drop between each step, not the overall rate.
2. Audit the page at the biggest drop first. A predictive heatmap and a CRO audit tell you what is wrong before you have to test everything.
3. Ship the top three fixes. Structural changes (button placement, delivery info, guest checkout) come before copy and colour.
4. Measure the step rate, not the overall rate. Re-run the audit to confirm the heat moved and the friction findings cleared.
5. A/B test when a page has the traffic for it; otherwise ship and measure the trend. Most stores do not have the volume to test every change and should not pretend to.
6. Repeat monthly on the next page. Conversion is won a page at a time.

For the full list of checks, see the CRO checklist. For Shopify stores specifically, the Shopify CRO audit page walks through the theme and app problems that do not show up in analytics.