How to Run an Ecommerce CRO Audit: A 10-Step Checklist for Your Store

Name the primary conversion (a completed purchase) and two supporting KPIs: conversion rate, which is orders divided by sessions, and revenue per visitor. Revenue per visitor stops you from celebrating a discount that lifts orders and shrinks profit.

Confirm the core ecommerce events are firing with the correct item and value data. Google's GA4 ecommerce documentation lists the recommended events for the purchase journey, including view_item_list, select_item, view_item, add_to_cart, view_cart, begin_checkout, add_shipping_info, add_payment_info and purchase.

Reconcile analytics purchases against the order count in your store admin for the same 30 days. If the two numbers do not line up closely, your funnel report is fiction and tracking comes before optimization.

Check that payment and app domains are excluded as referrals, so a checkout redirect does not start a new session and hand the sale to the wrong channel.

Check what your consent banner does to data collection. If a large share of visitors decline analytics cookies, your funnel is a sample rather than a census, and you should read it as a direction rather than a precise figure.

Freeze a baseline: at least 30 days of data, or 90 days if your traffic is low. Write those numbers down so you can measure what the audit actually changed.

Device: mobile, desktop and tablet, reported separately for every funnel step.

Channel: branded search, non-branded search, paid social, email and direct. Cold paid social traffic and branded search convert nothing alike, so judging them together tells you nothing.

New versus returning visitors. Some gap is normal: Contentsquare's 2026 Digital Experience Benchmark puts returning visitors at a 2.9% conversion rate against 1.7% for new visitors. If your new-visitor rate trails far further behind, you likely have a persuasion and trust problem rather than a checkout problem.

Your top ten landing pages by sessions. Check whether a handful of templates carry most of your entries, because those templates are where a fix pays back fastest.

Country and currency. Shipping cost, duties and a missing local payment method quietly kill conversion in specific markets.

Build a five-step funnel: sessions, product viewed, added to cart, checkout started, purchased. Record the completion rate between each pair of steps, not only the overall conversion rate.

Rank the leaks by the number of sessions lost, not by the worst percentage. For example, a step that loses 25% of 40,000 sessions matters far more than one losing 60% of 900.

Rebuild the same funnel for mobile only, and check whether the biggest leak sits at a different step on a phone than on desktop.

Compare this quarter against your own previous quarter before you compare yourself to any industry benchmark. Your own trend is the cleanest signal you have.

Write down the top three leaks with their session counts. Steps 4 to 9 are about finding out why they happen.

Run a five-second test on the homepage with someone who does not know your brand: what do you sell, who is it for, and what should they do next? If they cannot answer, the headline is the first fix.

Check that navigation labels match how customers describe your products, not your internal catalog structure. Your search query reports and support emails are the source of truth here.

Review internal site search. Shopify's behavior reports include Searches by search query, Searches with no results and Searches with no clicks. Zero-result searches are a free list of products, synonyms and spellings you are missing.

Time your popups and promo bars. A discount overlay that covers the page before a first-time visitor sees a product is a bounce generator, not a list-building tactic.

Put the reassurance a first-time buyer needs near the top: shipping threshold, returns window and a genuine review count.

Check that filters match how people actually shop: size, price, use case or occasion, not only color and brand. On mobile, confirm filters are reachable without scrolling past the whole grid.

Product cards should carry price, rating, the main variant and a second image on hover or swipe. Every missing detail forces an extra click that some shoppers will not make.

Decide what happens to out-of-stock products. Pushing them to the end of the grid keeps their URLs live, while unpublishing them can break links and lose search traffic.

Test pagination and load-more behavior. If tapping back from a product page dumps the shopper at the top of the grid again, you are losing people who were still browsing.

Keep collection copy short at the top and move long SEO text below the grid, so the products stay the first thing a visitor sees.

Images: a clean primary shot, a scale reference, the product in use, and a zoom that works with one thumb on a phone.

Price, shipping cost or threshold, and returns window visible near the buy box, not buried in a footer link or an accordion.

Add to cart visible without scrolling on a phone, with a sticky buy bar once the shopper scrolls into the description.

The first three lines answer what it does for me, in plain language. Specifications, materials and dimensions sit below that for the shoppers who want them.

Reviews with photos near the buy box, with the total count visible. Keep the critical reviews: a page of nothing but five stars can look filtered.

Variant selection that never dead-ends. Unavailable sizes should be visible and marked, with a back-in-stock option rather than a silent failure.

A delivery estimate expressed as a date range rather than a vague promise, because slow delivery is a documented checkout abandonment reason (see Step 7).

Extra costs revealed late. In Baymard Institute's research, 40% of US shoppers who abandoned a checkout did so because of extra costs such as shipping, tax and fees. Show shipping cost or the free-shipping threshold before the final step.

Slow delivery: 20% abandoned because delivery was too slow in the same Baymard research. If you cannot ship faster, state the date clearly and offer a paid faster option.

Security doubts: 19% did not trust the site with their card details. Check that recognizable payment marks, a visible returns policy and a polished, consistent checkout design are all in place.

Forced account creation: 18% abandoned because the site made them create an account. Guest checkout, with an optional account offered after the order, removes that block.

Length and complexity: 17% abandoned a checkout that was too long or complicated. Count your fields, delete the optional ones, and use address autocomplete.

Payment coverage: in the same list, 9% abandoned because there were not enough payment methods and 10% because the card was declined. Wallets and a retry path on a declined card are both worth checking.

Check the three Core Web Vitals against Google's published thresholds on web.dev: Largest Contentful Paint within 2.5 seconds, Interaction to Next Paint of 200 milliseconds or less, and Cumulative Layout Shift of 0.1 or less, all measured at the 75th percentile of page loads.

Read field data, not only lab scores. PageSpeed Insights reports Lighthouse lab results alongside Chrome User Experience Report field data from the previous 28-day collection period. The field data is what real shoppers experienced.

Test one URL of each type: homepage, a collection page, your best-selling product page and the cart. A fast homepage tells you nothing about a heavy product template.

Check tap targets and thumb reach. Primary actions belong in the lower half of the screen on a phone, with enough space that a thumb cannot hit two controls at once.

Check checkout inputs on mobile: correct keyboard types for email and number fields, working autofill, and no zoom-on-focus jump.

Finish on a real mid-range Android phone on cellular data, not on your own newest device over office wifi.

Scroll and click heatmaps on your five highest-traffic templates. Look for content below the point where most visitors stop scrolling, and for clicks on things that are not links.

Watch 10 to 15 session recordings of visitors who reached checkout and did not buy. That narrow filter is far more useful than watching random sessions.

Check rage clicks and dead clicks. In Microsoft Clarity's definitions, a rage click is repeated clicking in one small area in rapid succession, and a dead click is a click that gets no visible response. Both flag the moments where a shopper expected something to work and it did not.

Run one on-site survey on the product page: what almost stopped you from buying today? One open question collects more usable material than a ten-question form.

Run a post-purchase survey asking what nearly made them leave and where they first heard about you. Responses build up with every order, so the sample keeps growing even on a small store.

Mine reviews: your own three-star reviews and competitors' one-star reviews are a list of objections your product page should already be answering.

Tag 30 days of support tickets and chat transcripts by theme. Repeated pre-sale questions are missing product page content, and repeated post-sale questions are missing checkout clarity.

Write one row per finding: what you observed, the evidence, the funnel step it affects, the device it affects, and the proposed change.

Score each row with PIE, the prioritization framework created by Chris Goward and published by Conversion: Potential (how much room for improvement the page has), Importance (how much traffic and revenue flows through it), and Ease (how hard the change is to implement or test). A common approach is to score each 1 to 10 and average the three.

ICE works the same way with Impact, Confidence and Ease, and suits teams that already have a specific variant in mind. Pick one framework and apply it to every row, because the value is in the consistency.

Split the list in two. Broken things and obvious usability failures get fixed immediately. Genuine judgement calls, such as a new page layout or a pricing presentation, become test ideas.

Give every top-ten item an owner and a date. An unassigned audit finding has a habit of surviving to the next audit.

Book a re-measure 30 days out against the baseline you froze in Step 1, using the same segments.

Just fix the broken things. A size chart that is not clickable, a shipping cost hidden until the last step or a form that rejects valid addresses does not need a test. Testing a defect only delays the repair.

Test at the template level, not the page level. For example, one product page might see 200 sessions a month while the product template across your whole catalog sees 20,000.

Take bigger swings. The smaller the lift you want to detect, the more traffic a test needs, so small stores should test changes large enough to produce a large effect: a rewritten hero, a restructured product page, a different checkout flow.

Substitute qualitative evidence for statistical power. A handful of unmoderated user tests can expose a confusing checkout within days, which an underpowered experiment may never do.

If you must compare before and after, keep the periods equal, exclude promotions and seasonal spikes, and treat the result as a signal rather than proof.

Start in Shopify's own behavior reports: Conversion rate breakdown, Conversion rate over time, Sessions by device, Sessions by landing page and the three search reports. They need no tag setup and come from the same platform that records your orders, which makes them a useful cross-check on GA4.

The same behavior reports include web performance reports for Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift, broken down by page URL and by page type. Use them to see which templates miss the Step 8 thresholds.

Check your checkout is on the current foundation. Shopify's developer documentation confirms that checkout.liquid and additional scripts were sunset for the Thank you and Order status pages on August 28, 2025. Script tags on those pages were sunset for Plus stores on the same date and for non-Plus stores on August 26, 2026. Legacy customizations and tracking pixels attached that way may now be silently dead, so re-check your purchase tracking from Step 1.

Audit installed and removed apps. Shopify's help center notes that some apps add theme code that is not removed on uninstall. Leftover and extra third-party scripts add work to the page, which can hurt Interaction to Next Paint on exactly the templates that sell.

Check whether heavy homepage sections, autoplay video and third-party review widgets are the reason your Largest Contentful Paint misses 2.5 seconds on mobile.

WooCommerce Analytics is built around orders, revenue, products, coupons, customers and taxes. Its documented report list has no session or funnel report, so your funnel has to come from GA4. Get that tracking right before Step 3.

Measure cart and checkout separately from your homepage. WooCommerce's caching guidance says Cart, My Account and Checkout must stay uncached because they show customer-specific data, so a fast cached homepage says little about how quickly those pages respond on your hosting.

Plugin sprawl is the WooCommerce equivalent of app bloat. Each plugin can add scripts, styles and database queries to the pages where speed matters most, so remove the ones you no longer use.