An Ecommerce CRO Audit That Ends In Pass Or Fail

Someone ran an ecommerce CRO audit on your store and sent back twenty-three recommendations: a smaller hero image, reviews moved up the page and a shorter checkout.

Most of those are probably worth doing. What the list withholds is which one costs you money today, and which one your developer should build on Monday.

That difference separates a review from an audit. A review reports what somebody thought when they looked at your store, while an audit returns a verdict against a condition written down before anyone looked.

I have spent eleven years running SEO and conversion work on other people's ecommerce stores, most of them on Shopify.

Every audit that changed revenue had the same shape: a handful of checks, each carrying a written threshold, worked through in a fixed order. This article sets out that shape, pass condition by pass condition.

The Order The Checks Have To Run In

An ecommerce CRO audit produces decisions only if two things hold. Every check carries a pass condition you could have written before opening the store, and the checks run in an order that dependency decides for you.

Get either wrong and you are back to a list of opinions with no way to rank them.

What A Pass Condition Has To Look Like

Here is that distinction in practice. “The rating is hard to see” is not a check, while “the star rating renders within the first 844 pixels on a 390-pixel-wide viewport” is one, because two people looking at that page return the same answer.

A condition has to be settleable by inspection. In practice that means a number, a presence test, or a yes-or-no about where something sits on the page.

Scoring pages against written criteria is long established, which stops the audit feeling like something an agency invented to sound rigorous. It goes back to the heuristic frameworks associated with Andre Morys of konversionsKRAFT.

For a tested standard to score against, the Nielsen Norman Group product-pages report runs to 108 design guidelines. That is far more than one store needs, but it settles the question of whether a pass condition can be written down at all.

One practical consequence of writing them down matters more than the provenance. Because every condition below is settled by looking, with no test involved, a store taking 300 visits a month reaches the same verdicts as a store taking 300,000.

So you are never waiting on statistical significance, which is why this audit works where A/B testing cannot. You are checking whether a thing is true.

Ranking What You Find: Theme-Level Before Page-Level

Two faults of identical severity are rarely worth the same to fix. Every audit we run starts by establishing which pages and products already carry the revenue, because that decides where the rest of the work is worth doing.

You can see how we structure an audit before any of it begins.

After that, exposure decides the ranking. A fault living in the theme beats the same fault on one page, because a single template edit clears it across the catalogue while a page fix clears one URL.

That is why a buy button sitting too low on your default product template outranks a missing delivery line on your best seller, even though the best seller feels more urgent.

Those two rules produce three passes in a fixed order. Verify the analytics first, because every number you are about to reason from comes out of it.

Then take the product page, where the decision to buy is made. Finish with the cart and checkout, where that decision gets executed.

Whether Your Analytics Is Telling You The Truth

No funnel number can be acted on until the events behind it have been verified. There is also a platform change only days away now that will silently break conversion tracking which works perfectly well on your store today.

This pass is short, and it is not optional.

The Events That Have To Be Firing Before A Funnel Means Anything

Open your analytics and confirm the product view, add to cart, checkout start and purchase events all arrive with the right parameters. On Shopify, Settings then Customer events is the fastest honest way to see it.

Alongside that, run GA4 DebugView with your own phone session going. You are not looking at traffic volume here.

That check is narrower: whether an add to cart tapped on a real handset produces the event you believe it does, with the value attached. Our guide to conversion tracking covers the setup itself.

Then check what you measure yourself against, because a half-remembered benchmark beats nothing by very little. Market data from IRP Commerce for June 2026 put the global average ecommerce conversion rate at roughly 2.0%.

That figure moves by month and by traffic source, so treat it as a reference point. Chasing the average is wasted effort.

Your pass condition is a baseline you can name along with the month it was measured. Our piece on ecommerce conversion rate covers choosing one.

The 26 August Checkout Deadline And What It Does To Your Data

Here is the check that did not exist last year. Script tags on the Thank you and Order status pages, per Shopify's own documentation, are sunset for non-Plus stores on August 26, 2026.

Put concretely: Non-Plus stores keep checkout.liquid, Additional Scripts and Thank You page script boxes until 26th August 2026, after which Shopify auto-upgrades and drops everything living in them.

That matters because plenty of stores put their conversion pixel in exactly that box years ago and never revisited it. Anything sitting there stops reporting on the twenty-sixth of this month, and nothing will announce it.

That replacement is Web Pixels, under Customer events. Agencies handling the migration, Fudge among them, report conversion losses reaching 14% where it is done badly.

A loss that size would take months to spot in a monthly report, which is why this check moves to the front of the audit.

The Product Page: Four Conditions With A Pass Mark

Four conditions decide whether a product page lets somebody buy without effort. All four are checkable in about ninety seconds a page, which makes this the fastest pass in an ecommerce CRO audit.

Rather than assert how often stores clear them, I went and measured it on 6th August 2026. We attempted 48 DTC domains, confirmed 40 were on Shopify and rendered 38 product pages. A buy box could be located on 28 of those. Each page was rendered at a 390×844 mobile viewport with a real iOS user agent.

Those ten we could not measure are a finding in their own right. On nine of the 38 stores no buy-box control could be located at all, and a tenth reported it off-canvas at -804 pixels.

A buy button automated tooling cannot identify is worth checking by hand, because assistive technology may not find it either. All ten are excluded from the percentages below.

The Buy Box And The 844-Pixel Line

Your condition: a shopper landing on a phone reaches the add-to-cart control without scrolling. Either it sits inside the first screen, or a persistent bar keeps it there.

Of the 28 Shopify product pages where we could locate a buy box, 18 placed the add-to-cart control below the 844-pixel fold of a 390×844 mobile viewport. Two thirds of the pages we could measure ask for a scroll before they will take the money.

Sitting below the fold is survivable when something brings the button back. Frequently nothing does. On 14 of those 28 pages the add-to-cart sat below the fold with no persistent bar to bring it back.

Scrolling took the buy button off screen for good on exactly half the pages we could measure. The median add-to-cart on the pages we measured sat at 996.5 pixels, a clear 152 pixels below the fold it needed to clear.

Stores that pass do it two different ways, and the distinction decides what you copy. On its Fresh Kiss trio page, ColourPop puts the add to bag at 823 pixels, inside the fold, on a theme carrying years of accumulated apps.

That is a layout solution. It also disposes of the excuse I hear most often, because a crowded theme and a catalogue that size plainly did not force the button down the page.

The second route is to carry the button with the shopper instead. On its poop bag carrier page, Wild One has no inline add-to-cart in the first screen at all, and the control you reach at 784 pixels is a bar pinned to the bottom of the viewport.

So both routes pass, and a shopper cannot tell them apart. Knowing which one you are copying still matters, because a persistent bar is something your developer can ship this week and a template rebuild is not.

That bar rescues failures too. On its edit page, Tropic Skincare leaves its real ADD TO BAG below the fold, and passes this condition only because the persistent bar is sitting there to catch it.

Take the bar from that page, and leave the layout where it is. Where the buy button sits is a call to action decision before it is a design one.

The Rating A Shopper Never Scrolls To

This condition splits in two. Reviews exist on the page at all, and a star rating is visible in the first screen beside the price.

Seven of the 38 pages carried no reviews at all, and on 12 of the 28 measurable pages the star rating never appeared in the first screen. That second failure is the quieter and more common one.

An owner knows the reviews are there and never checks whether a shopper sees them before deciding. Our write-up on ecommerce UX goes further into where proof belongs on the page.

It is also the failure with the least excuse behind it. Fitting a star rating beside a price costs no layout space that the price does not already occupy.

Where I see it go wrong is on pages with a great deal to say about the product. Specification copy pushes the review block down, the rating goes with it, and nobody notices because the reviews are still technically on the page.

The Delivery Cost You Leave Them To Find At Checkout

Your condition: the delivery cost or free-delivery threshold appears on the product page, well before the payment step.

Nine of the 38 product pages we rendered stated no delivery cost or free-delivery threshold anywhere on the page, leaving it to be discovered at checkout. Roughly a quarter of the sample therefore reveals the price increase at the worst available moment.

What that costs is the subject of the next pass, and it is the largest single number in this article.

Clearing every condition at once turns out to be rare. Only 6 of the 28 pages cleared all four conditions at once, a little over a fifth of the stores I crawled.

That makes doing so a competitive position rather than table stakes. Twelve of the 28 pages missed exactly one condition, and the misses were not evenly spread: six on the buy box, four on the rating, two on delivery.

The buy box is therefore both the most commonly failed condition and the most commonly near-missed, which is where I would start on almost any store.

Category explains none of it either. The six that cleared all four span skincare, activewear and pet accessories. Cosmetics, bedding and bathroom fixtures complete the set.

Speed, by contrast, was the check fewest stores failed. Across those 38 stores the median product page finished loading its document in 940ms and the slowest took 2,986ms.

The Cart And Checkout Pass: Surprise Costs And Forced Accounts

Two faults account for more abandoned orders than everything else an ecommerce CRO audit turns up, and both are found by putting a real order through to the payment step. Your theme settings will not tell you.

Owners skip this pass because the checkout looks finished.

Surprise Costs At The Payment Step

Baymard Institute, which has run moderated checkout testing for over a decade, maintains a rolling analysis of 50 cart-abandonment studies putting the overall rate at 70.22% and attributing 39% of abandonments to surprise costs at checkout.

Read those two numbers together and seven in ten carts are lost, the largest identified slice of them to a cost the shopper never saw coming.

That single cause outweighs every layout question in this article put together, and fixing it is a merchandising decision, and a cheap one.

Timing is what makes it so expensive. The shopper has already chosen the product, so a cost added at the payment step lands as a penalty.

Your pass condition follows directly. Every cost the shopper will pay is visible before they reach the payment step.

Run one order on your own store, on a phone, to the point where the card field appears. Note the moment the total changes. If it changes there, you have found the fault.

Forced Account Creation

Here the condition is whether a guest can complete an order without ever creating an account. The same Baymard analysis attributes 19% of abandonments to mandatory account creation, which is a fifth of your lost carts sitting behind one toggle.

On Shopify that toggle is a single setting under checkout configuration. It is one of very few conversion faults that takes longer to describe than it does to fix.

Treating the checkout as a whole pays better than most owners expect. A separate piece of Baymard work, its checkout flow guide, finds the average large ecommerce site can gain a 35.26% increase in conversion rate through better checkout design alone.

Treat that as a ceiling. It describes an average across sites with far more to fix than yours.

It is still why this pass comes last and never gets dropped, and why the ecommerce funnel repays auditing end to end.

Have The Audit Run For You

Suppose you have run this ecommerce CRO audit and come out holding a list of failed conditions. Checking was never the hard part.

That ordering is where stores stall. A buy box sitting 152 pixels too low on the default template and a purchase event that stops firing on the twenty-sixth are not the same kind of job, and only one of them has a deadline attached.

Separating those two is the work I do personally on every project.

After eleven years and more than 3 million organic sessions driven for ecommerce stores, an outside read on that list is usually worth more than another opinion on the hero image. Most of that time is spent arguing stores out of the fix that feels urgent and into the one that touches every template.

Bring me the conditions your store failed. Book a free 30-minute consultation and we will put them into the order I would work them in on a store like yours, at no cost.