Ecommerce CRO Checklist: 42 Checks in Funnel Order, Each With the Number That Proves It
An ecommerce CRO checklist ordered by funnel stage. Every check names the metric that shows it is failing and the fix, plus what Shopify checkout lets you test in 2026.
The short answer: An ecommerce CRO checklist should run in funnel order: measurement, landing pages, speed, navigation and search, product pages, cart, checkout, then post-purchase. Do not work through it top to bottom. Measure first, find the stage with the biggest drop on your highest-revenue traffic, and fix that stage. The most common fixable reason shoppers leave is surprise cost: 40% of US online shoppers abandoned a purchase over extra costs like shipping in Baymard Institute's 2026 survey. On Shopify, the Information, Shipping and Payment steps of checkout can only be extended with app blocks on Shopify Plus, so most stores should test everything up to the checkout button. Facts checked October 6, 2026.
Most CRO checklists are lists of good ideas. The problem is that a store has 42 good ideas and time for three. This one is built to help you choose. Every check has three parts:
- The check: what to look at.
- The failing signal: the number that tells you it is broken on your store, not on stores in general.
- The fix: what to change, and whether it is worth an A/B test or just a fix.
If a check's failing signal does not show up in your data, skip it. That is the point.
How to use this checklist
- Do Stage 0 first. Without it, every later check is a guess.
- Build one funnel: landing page, product view, add to cart, checkout start, purchase. Split it by device (mobile and desktop) and by traffic source (paid social, paid search, organic, email).
- Find the biggest leak on the traffic that pays. A 70% drop on organic blog visitors matters less than a 40% drop on paid mobile traffic you spend money to buy.
- Work only that stage's checks. Fix the obvious bugs outright. A/B test the changes that are a judgment call.
- Score on revenue per visitor where you can, not conversion rate alone. More orders at a lower basket size can mean less money.
Some cart abandonment is not fixable. In the same Baymard survey, 42% of shoppers said they abandoned because they were just browsing. Your goal is the other reasons.
Stage 0: Measurement, before you change anything
| Check | Failing signal | Fix |
|---|---|---|
| Every funnel step is tracked | You cannot say how many sessions reached add to cart last week, by device | Track product view, add to cart and checkout start as events or click steps |
| Paid traffic is tagged | A large share of sessions show as "direct" while ads are running | Add UTM parameters to every ad, email and social link. Use a UTM builder and one naming convention |
| Revenue reaches your analytics | You know orders per day but not orders per landing page or source | Connect order revenue to sessions, or at least compare source-level orders in your store reports |
| Mobile and desktop are separated | You report one blended conversion rate | Report the funnel per device. Mobile and desktop usually fail at different steps |
| Test math is set before testing | You stop tests when they "look good" | Estimate sample size and duration before launch with a sample size calculator and a test duration calculator |
Without this stage you will fix the wrong thing with confidence. If you only do one stage this quarter, do this one.
Stage 1: Landing pages and traffic match
Paid traffic lands on specific pages for specific reasons. If the page does not continue the ad's message, visitors leave before they see a product.
| Check | Failing signal | Fix |
|---|---|---|
| The page matches the ad or query | Bounce on paid landing pages is much higher than on organic ones for the same product | Repeat the ad's offer, product and image in the first screen |
| Visitors land close to the product | Paid visitors land on the homepage and few reach a product page | Send ads to the product or collection page they promote |
| The offer is visible on mobile without scrolling | Mobile heatmaps show few clicks below the first screen and low scroll depth | Put product, price and primary button in the first mobile screen |
| One primary action per page | Clicks spread across many links; the main button gets a small share | Remove competing calls to action from paid landing pages |
| Promotions match | Discount in the ad is missing or different on the page | Show the same code or price on the page, applied automatically if your platform allows |
Stage 2: Speed and stability
Google's Core Web Vitals give fixed thresholds. A page passes when at least 75% of page views meet the "good" level for each metric.
| Check | Failing signal | Fix |
|---|---|---|
| Largest Contentful Paint | Above 2.5 seconds at the 75th percentile on mobile | Compress and correctly size the hero image. Do not lazy-load the main product image |
| Interaction to Next Paint | Above 200 milliseconds | Remove unused apps and third-party scripts; defer the rest |
| Cumulative Layout Shift | Above 0.1 | Reserve space for images, reviews widgets and banners that load late |
| App and script count | Speed got worse after installing apps | Audit installed apps every quarter. Uninstalling a Shopify app can leave code in the theme, so check the theme too |
| Field data, not lab data | You only check Lighthouse on your own laptop | Use field data (Chrome UX Report, PageSpeed Insights origin data) for real visitors |
Speed fixes are usually fixes, not tests. Ship them.
Stage 3: Navigation, search and collection pages
| Check | Failing signal | Fix |
|---|---|---|
| Visitors can find categories | High exit rate on the homepage and few clicks on navigation | Name categories the way customers do, not internal product line names |
| Site search returns results | Searches with zero results, or searches followed by exits | Add synonyms and common misspellings; show popular products on empty results |
| Filters match how people choose | Filter use is low on large collections | Filter by the attributes buyers compare (size, price, use case), not by internal tags |
| Collection cards show what decides the click | Few clicks from collection to product page | Show price, key variant (color, size) and rating on each card |
| Dead clicks | Heatmaps show clicks on images or text that do nothing | Make the clicked element a link, or remove the cue that makes it look clickable |
Stage 4: Product pages
This is where most stores lose the most visitors with intent. Watch the add-to-cart rate from product views.
| Check | Failing signal | Fix |
|---|---|---|
| Price, shipping cost and delivery date are visible | Low add-to-cart rate; cart visitors leave at the shipping step | Show shipping cost or free-shipping threshold and a delivery estimate next to the price |
| Images answer the questions copy cannot | Visitors click through all images and still leave | Add scale, in-use and detail images; add video for products people need to see moving |
| Variants are clear | Add-to-cart errors from "select a size" or many variant clicks before leaving | Preselect the most common variant where it is safe; show out-of-stock variants clearly |
| The add-to-cart button is obvious | Low click share on the button in click heatmaps, especially on mobile | One high-contrast button; on mobile, a sticky button once the visitor scrolls past it |
| Returns policy is near the button | Shoppers open the returns page and then leave | Summarize returns in one line next to the button and link the full policy |
| Reviews are real and readable | Low add-to-cart on products with few or no reviews | Show review count and rating near the title. Never write or buy reviews |
| Description answers buying questions | Support tickets ask about sizing, materials or compatibility | Put those answers on the page, in plain language |
Scroll heatmaps tell you where to put things. If 30% of mobile visitors reach your returns line, it is effectively invisible.
Stage 5: Cart
| Check | Failing signal | Fix |
|---|---|---|
| The total is clear before checkout | Visitors leave from the cart after the shipping estimate appears | Show estimated shipping and tax in the cart. In Baymard's survey, 12% abandoned because they could not see the total upfront |
| Free-shipping progress | Average order value sits just under your free-shipping threshold | Show how far the cart is from free shipping |
| Express payment is offered | Mobile checkout starts are low relative to carts | Show express wallets (Shop Pay, Apple Pay, Google Pay, PayPal) in the cart |
| Discount field does not send people away | Visitors leave the cart to search for codes | Apply promotions automatically, or make the code field less prominent |
| Edits are easy | Many cart updates, then exits | Allow quantity and variant edits in the cart without a page reload |
Stage 6: Checkout
Checkout has the clearest research. In Baymard Institute's 2026 survey, the fixable reasons for abandonment included extra costs (40%), slow delivery (20%), not trusting the site with a card (19%), forced account creation (18%), a long or complicated checkout (17%) and site errors (17%). Baymard's testing found an ideal checkout can have 12 to 14 form elements, while the average US checkout shows 23.48 by default.
| Check | Failing signal | Fix |
|---|---|---|
| Guest checkout is available | Drop at the login or account step | Allow guest checkout; offer an account after purchase |
| Form fields are minimal | Long time between checkout start and payment | Remove optional fields; use address autocomplete |
| No new costs appear at checkout | Drop at the shipping step | Make the cart total match checkout. No new fees at the last step |
| Enough payment methods | Drop at the payment step, higher on mobile | Add the wallets and local methods your markets use |
| Trust at the payment step | Drop at payment on first-time visitors | Show a recognizable checkout, security cues and a contact option |
| Errors are tracked | Spikes in abandonment after a release | Test checkout on real phones after every theme or app change |
What Shopify lets you change in checkout in 2026
Shopify no longer supports checkout.liquid for the Information, Shipping and Payment steps. Customization runs through Checkout Extensibility, and app blocks on those three steps require Shopify Plus. Shopify also retired additional scripts and script tags on the Thank you and Order status pages: for Plus stores on August 28, 2025, and for non-Plus stores on August 26, 2026.
What this means for testing:
- On Shopify Plus: use checkout apps and branding for checkout changes, and use a testing tool that supports checkout extensibility if you need to test them.
- On other plans: treat checkout settings (guest checkout, payment methods, express wallets) as fixes, and run your A/B tests on everything up to the checkout button.
- Script-based tools do not run inside Shopify checkout. For any analytics or testing tool that installs through your theme, the funnel ends at "checkout started". Read completed orders from Shopify's reports or from a supported revenue source.
Stage 7: Post-purchase and recovery
| Check | Failing signal | Fix |
|---|---|---|
| Abandoned checkout messages work | Recovery emails have opens but few recovered orders | Measure recovered orders, not opens. Send the first reminder within hours, not days |
| Order confirmation sets expectations | "Where is my order" tickets | Show delivery estimate and tracking link on confirmation |
| Repeat purchase path exists | Low share of returning customers on consumables | Remind at the replenishment interval; offer one-click reorder |
| Post-purchase offers fit the order | Offer accepted rarely or causes refund requests | Offer accessories for what was bought, not random products |
Use a cart abandonment calculator to put a revenue number on your abandonment rate before deciding how much time this stage deserves.
How to prioritize and test what you find
Rank each failing check by three things:
- Revenue at stake: visitors reaching that step, times the drop, times average order value.
- Confidence: do you have the failing signal in your data, or only a hunch?
- Effort: a setting change, a theme change, or a rebuild.
Then decide fix or test:
- Fix without testing when it is a bug, a speed problem, a missing payment method or a broken element. There is no "variant B" for a broken button.
- A/B test when reasonable people would disagree: layout, copy, offer framing, free-shipping threshold, image order.
- Check the math first. Low-traffic stores often cannot detect small changes. If the test would need months, test a bolder change or skip the test and ship the fix. The minimum detectable effect guide explains how to choose.
Score winners on revenue per visitor. A variant that sells more of the cheapest product can win on conversion rate and lose money.
Copy this checklist
Paste this into a sheet. Fill the last two columns from your own data.
| Stage | Check | Failing signal | Your number | Fix or test |
|---|---|---|---|---|
| 0 Measurement | Funnel steps tracked | Cannot count add to cart by device | ||
| 0 Measurement | Paid traffic tagged | High "direct" during campaigns | ||
| 0 Measurement | Revenue reaches analytics | No orders per landing page | ||
| 0 Measurement | Device split | One blended rate | ||
| 0 Measurement | Test math set | Tests stopped early | ||
| 1 Landing | Matches ad | Paid bounce far above organic | ||
| 1 Landing | Lands near product | Paid lands on homepage | ||
| 1 Landing | Offer above the fold on mobile | Low scroll depth | ||
| 1 Landing | One primary action | Clicks spread thin | ||
| 1 Landing | Promotion matches | Ad discount missing on page | ||
| 2 Speed | LCP | Over 2.5 s at p75 mobile | ||
| 2 Speed | INP | Over 200 ms | ||
| 2 Speed | CLS | Over 0.1 | ||
| 2 Speed | App count | Slower after app installs | ||
| 2 Speed | Field data | Lab data only | ||
| 3 Navigation | Category names | High homepage exits | ||
| 3 Navigation | Search results | Zero-result searches | ||
| 3 Navigation | Filters | Low filter use | ||
| 3 Navigation | Collection cards | Low collection to product clicks | ||
| 3 Navigation | Dead clicks | Clicks on non-links | ||
| 4 Product | Price, shipping, delivery | Low add to cart | ||
| 4 Product | Images | All images viewed, then exit | ||
| 4 Product | Variants | Variant errors | ||
| 4 Product | Add-to-cart button | Low click share on mobile | ||
| 4 Product | Returns near button | Returns page, then exit | ||
| 4 Product | Reviews | Low add to cart, few reviews | ||
| 4 Product | Description | Repeated pre-sale tickets | ||
| 5 Cart | Total visible | Exit after shipping estimate | ||
| 5 Cart | Free-shipping progress | AOV just under threshold | ||
| 5 Cart | Express payment | Low mobile checkout starts | ||
| 5 Cart | Discount field | Exits to find codes | ||
| 5 Cart | Easy edits | Updates, then exits | ||
| 6 Checkout | Guest checkout | Drop at login | ||
| 6 Checkout | Minimal fields | Slow to payment | ||
| 6 Checkout | No new costs | Drop at shipping step | ||
| 6 Checkout | Payment methods | Drop at payment | ||
| 6 Checkout | Trust | First-time drop at payment | ||
| 6 Checkout | Errors tracked | Spikes after releases | ||
| 7 Post-purchase | Recovery messages | Opens without orders | ||
| 7 Post-purchase | Confirmation | Order status tickets | ||
| 7 Post-purchase | Repeat purchase path | Few returning customers | ||
| 7 Post-purchase | Post-purchase offers | Rare acceptance or refunds |
Running the checklist with Humblytics
Humblytics covers Stage 0 through Stage 5 on a Shopify or custom store: funnels with page view, click and form submit steps, click and scroll heatmaps with a mobile and desktop toggle, dead clicks, traffic by source and campaign, and A/B tests built in a visual editor without code. Install it by adding one script to your theme. See the Shopify integration for setup.
Know the limits before you start:
- Revenue on Shopify Payments and Shop Pay is not supported. Revenue attribution works with Stripe or Foxy, or with custom purchase events. If your store uses Shopify Payments, score tests on add-to-cart or checkout-start clicks and check order value in Shopify's own reports.
- No tracking inside Shopify checkout. Like any theme-installed script, the funnel ends at checkout start.
- No session replay. If you need recordings, Microsoft Clarity or Hotjar do that.
If your store sells through Stripe, tests can be scored on revenue per visitor directly. See revenue attribution and A/B testing. For a comparison of Shopify-native testing apps, read the best A/B testing tools for Shopify, and for multi-touch attribution across ad platforms, the best revenue attribution tools for ecommerce.
Start a free trial and add the script, or connect your agent through Claude or Codex to pull your funnel and find the biggest leak first.
Sources and freshness
- Baymard Institute, 50 Cart Abandonment Rate Statistics 2026, list last updated September 22, 2025; retrieved October 6, 2026. Supports the 2026 survey of 1,083 US adults (reasons for abandonment, including 42% just browsing, 40% extra costs, 12% could not see the total), the 12 to 14 ideal form elements and the 23.48 average in US checkouts.
- web.dev, Web Vitals and How the Core Web Vitals metrics thresholds were defined, retrieved October 6, 2026. Support the LCP 2.5 seconds, INP 200 milliseconds and CLS 0.1 thresholds at the 75th percentile.
- Shopify Dev Docs, Layouts and Best practices for editing checkout.liquid, retrieved October 6, 2026. Support that checkout.liquid is unsupported on the Information, Shipping and Payment steps, and the August 28, 2025 and August 26, 2026 sunset dates.
- Shopify, How to Customize Shopify Checkout, published July 8, 2026; retrieved October 6, 2026. Supports that app blocks on the Information, Shipping and Payment steps require Shopify Plus.
- Humblytics, Shopify integration and llms.txt, retrieved October 6, 2026. Support the product capabilities and limits listed above.
The failing signals are diagnostic rules of thumb, not benchmarks. Compare each number with your own store over time rather than with an industry average.