CRO Ecommerce: Why Most Stores Fix the Wrong Things

CRO ecommerce is the practice of improving the percentage of website visitors who complete a purchase, and it is one of the highest-leverage activities available to any online retailer. A 1% improvement in conversion rate on a site doing £5 million in revenue is worth £50,000, without spending an extra penny on traffic. The problem is that most ecommerce teams spend their optimisation budget fixing things that look broken but are not, while ignoring the structural issues that are actually costing them money.

I have watched ecommerce teams spend months A/B testing headline copy on their homepage while their mobile checkout was throwing an error for 12% of users on a specific Android version. The homepage test moved conversion by 0.2%. The checkout bug, once fixed, recovered a material chunk of lost revenue overnight. That is not an unusual story. It is the norm.

Why Ecommerce CRO Is Different From General CRO

Ecommerce has a specific conversion architecture that separates it from lead generation or SaaS. The purchase funnel is longer, the decision-making process involves more friction points, and the emotional and rational triggers at play are different depending on category, price point, and audience. A tactic that works for a £15 impulse purchase will not work for a £400 considered purchase. That sounds obvious, but the volume of generic CRO advice being applied indiscriminately across both contexts is remarkable.

Ecommerce conversion optimisation also has to account for the fact that the same user may visit multiple times before converting, may convert on a different device than the one they browsed on, and may be influenced by email or paid retargeting between sessions. Single-session, single-device conversion rate is a useful operational metric, but it is an incomplete picture of how people actually buy. If you are not thinking about assisted conversions and multi-touch attribution alongside your on-site CRO work, you are optimising a partial view of the problem.

There is a broader body of thinking on this at The Marketing Juice conversion optimisation hub, which covers the full scope of CRO from testing methodology to commercial measurement. For ecommerce specifically, the principles are the same but the application is more granular.

Where Ecommerce Conversion Rate Losses Actually Happen

Before you can fix anything, you need to know where the money is leaving. Most ecommerce sites have three or four places where conversion rate drops sharply, and they are usually predictable.

The product detail page is the first major decision point. This is where a visitor decides whether the product is right for them, whether the price is justified, and whether they trust the seller enough to proceed. It is also where most ecommerce CRO programmes spend the least time. Teams will run extensive tests on homepage layouts and category page filters while leaving product pages with manufacturer copy, three low-resolution images, and no customer reviews. The product page is where the purchase decision is made or abandoned. Treating it as a pass-through is a structural mistake.

The cart is the second drop-off point. Visitors who add to cart have already signalled intent. Losing them at this stage is expensive because you have already paid to acquire them and they have already expressed interest. Cart abandonment is often a trust issue, a price shock issue (unexpected shipping costs are the most cited reason for cart abandonment), or a commitment-timing issue. Someone may genuinely want the product but not want to buy today. That is a retargeting problem as much as a CRO problem.

Checkout is the third and most commercially critical stage. Friction here is unforgivable. Forced account creation, long form sequences, limited payment options, unclear delivery information, and absence of security signals all drive abandonment at the point of maximum intent. Hotjar’s ecommerce CRO research consistently points to checkout complexity as one of the primary causes of lost revenue. This is not a new finding, but it remains one of the most under-addressed problems in ecommerce.

Site speed is the fourth variable, and it sits across all of the above. A slow product page, a slow cart update, a slow checkout step, each of these compounds the friction at every stage. Semrush’s analysis of page speed and performance makes clear that speed is not just a technical concern. It is a conversion variable with direct commercial consequences.

The Segmentation Problem: Why Blended Conversion Rates Mislead

One of the first things I do when reviewing an ecommerce CRO programme is ask to see conversion rate broken down by traffic source, device type, new versus returning visitors, and product category. The answer I usually get is a single blended number. That number is almost useless for diagnostic purposes.

I ran a review for a mid-size retailer a few years ago where the blended conversion rate looked acceptable, around 2.8%. When we segmented by device, mobile conversion was sitting at 1.1% against desktop at 4.9%. The mobile experience was broken in ways that were invisible if you only looked at the aggregate. The team had been running A/B tests on desktop layouts for six months while their fastest-growing traffic segment was converting at less than a quarter of the rate of desktop. The opportunity cost was significant.

Segmentation also matters by traffic source. Paid social traffic typically converts at a lower rate than branded search traffic, because the intent levels are different. If you are running heavy paid social acquisition, your blended conversion rate will look worse than your organic or email conversion rate. That is not a CRO problem. It is a traffic mix problem. Conflating the two leads to bad decisions.

New versus returning visitor conversion is another important split. Returning visitors who have already browsed or added to cart convert at significantly higher rates than cold traffic. If your returning visitor rate is increasing (which it will if your retargeting and email programmes are working), your blended conversion rate will improve even if your on-site experience has not changed. The inverse is also true. An aggressive new customer acquisition push will dilute your blended rate even if your actual conversion experience is improving. Always segment before you draw conclusions.

Product Page Optimisation: The Most Undervalued Lever in Ecommerce CRO

Product pages are where most ecommerce revenue is won or lost, and they are systematically underinvested compared to the attention given to homepage design, category navigation, and paid media creative. The irony is that the homepage is often the page with the lowest direct commercial impact. Most visitors do not convert from the homepage. They convert from product pages, after arriving via search, a paid ad, or a direct link.

The copy on a product page has to do several things simultaneously. It has to answer the specific questions a buyer has about this product. It has to address the objections that would prevent a purchase. It has to communicate value in a way that justifies the price. And it has to do all of this without being long-winded, because most people will not read every word. The gap between a product description that accomplishes this and one that just lists technical specifications is often the gap between a conversion and a bounce.

When I was at iProspect, we worked with a retail client whose product pages were entirely manufacturer-supplied. The copy was accurate but it was written for a buyer who already understood the product category. It answered no questions, addressed no objections, and gave the visitor no reason to buy from this retailer rather than a competitor. Rewriting the copy for the actual buyer, with language that reflected how customers described the product in reviews and support tickets, produced a measurable uplift in conversion. Not from a test, but from a straightforward before-and-after comparison with a clean enough traffic baseline to be meaningful.

Images and social proof sit alongside copy as the other critical product page variables. Multiple images showing the product in use, at different angles, in context, reduce the uncertainty that drives abandonment. Reviews and ratings provide the social validation that many buyers need before committing. Neither of these is a novel insight. But the number of ecommerce sites still operating with single product images and no reviews is higher than it should be.

Moz’s CRO playbook covers product page optimisation as one of the core ecommerce levers, and the principles there align with what I have seen work in practice: specificity in copy, quality in imagery, and trust signals placed where the buyer’s eye goes before they make a decision.

Checkout Optimisation: Removing Friction at Maximum Intent

Checkout abandonment is the most commercially painful form of conversion loss in ecommerce, because it happens at the point of maximum purchase intent. A visitor who has reached checkout has already decided they want the product. Losing them here is not a persuasion failure. It is an experience failure.

The most common checkout friction points are well-documented and have been for years. Forced account creation before purchase remains one of the most persistent causes of abandonment. Offering guest checkout is not a new idea, but there are still ecommerce sites requiring registration before purchase. Unexpected costs at checkout, most often shipping, are another primary driver. If your shipping cost is not visible until the final checkout step, you will lose buyers who have already invested time in the purchase process and feel surprised or misled.

Payment method coverage matters more than it used to. Buy now, pay later options have moved from a niche offering to a mainstream expectation in many categories. Digital wallets reduce form friction significantly. If your checkout requires manual card entry when a significant portion of your mobile traffic uses Apple Pay or Google Pay, you are adding unnecessary steps at the worst possible moment.

Trust signals in checkout are easy to overlook because they feel like table stakes. SSL certificates, recognisable payment logos, clear returns policies, and visible customer service contact details all contribute to the sense that a transaction is safe. The absence of these signals, particularly on smaller or less well-known retailers, is a genuine conversion inhibitor. Buyers who are uncertain about a brand will look for reasons to abandon, and a checkout page that does not signal trustworthiness gives them one.

CrazyEgg’s CRO statistics provide useful benchmarks across checkout abandonment rates and the impact of specific friction points. The numbers vary by category and price point, but the directional findings are consistent: fewer steps, clearer costs, and more payment options all improve checkout completion.

The Over-Engineering Problem in Ecommerce CRO

There is a version of ecommerce CRO that involves sophisticated personalisation engines, dynamic pricing experiments, AI-driven product recommendations, real-time behavioural targeting, and multi-variate testing across dozens of variables simultaneously. Some of that has genuine value at scale. Most of it is overkill for the majority of ecommerce businesses, and some of it actively gets in the way.

I have seen this pattern repeatedly across agency work: a brand invests in a complex personalisation platform before it has fixed its checkout flow, before it has written decent product copy, before it has addressed the fact that its mobile site is slower than its desktop site by three seconds. The platform costs six figures annually. The checkout fix would have cost a fraction of that and moved revenue faster.

This is not an argument against technology or sophistication. It is an argument for sequencing. The highest-return CRO work in ecommerce is almost always structural: fixing what is broken, reducing friction in the purchase flow, improving the quality of product information, and ensuring the site loads fast enough that visitors do not leave before they see anything. Once those foundations are in place, personalisation and advanced testing have something to work with. Without them, you are optimising the decoration on a building with structural problems.

When I launched a paid search campaign for a music festival at lastminute.com, we generated six figures of revenue within roughly a day from a campaign that was, by modern standards, straightforward. Clean landing page, clear value proposition, minimal friction between click and purchase. No personalisation engine, no dynamic content, no multi-variate test running in the background. The fundamentals were right and the commercial outcome followed. That experience shaped how I think about CRO: get the basics right first, then layer in sophistication.

Unbounce’s thinking on CRO and SEO working together touches on this principle from a different angle. The argument that organic traffic and conversion optimisation should be treated as connected rather than separate disciplines is one I agree with. Traffic quality affects conversion rate. Conversion rate affects the commercial return on SEO investment. They are not separate programmes.

How to Build an Ecommerce CRO Programme That Produces Commercial Results

A CRO programme that produces commercial results starts with a diagnostic, not a test. Before you run a single experiment, you need to understand where your conversion rate is lowest relative to where it should be, and why. That means segmented analytics, session recording review, heatmap analysis, and if your traffic volumes allow it, user testing with people who match your buyer profile.

The diagnostic should produce a prioritised list of hypotheses. Not “test button colour on the product page” but “the add-to-cart button is below the fold on mobile for 60% of our product pages, and mobile conversion is 3x lower than desktop, so moving the button above the fold on mobile is likely to reduce the gap.” That is a hypothesis with a mechanism, a supporting data point, and a specific expected outcome. Test that.

Prioritisation should be driven by expected commercial impact, not by ease of implementation or personal preference. The tests that are easiest to run are rarely the ones with the highest upside. The checkout flow is harder to test than the homepage, but the commercial stakes are higher. Allocate your testing capacity accordingly.

Statistical rigour matters, but it should not become an excuse for paralysis. You need enough traffic to reach statistical significance before calling a test, but you do not need to run every test for three months before acting on clear directional evidence. Unbounce’s CRO resource roundup includes useful guidance on test duration and significance thresholds that is worth reading if you are building out your testing methodology.

Document everything. The value of a CRO programme compounds over time, but only if you are building institutional knowledge about what works for your specific audience on your specific site. A test that fails is not a waste. It is information. A test that fails and is not recorded is a waste, because you will run the same test again in 18 months and get the same result.

There is more on the commercial measurement side of CRO programmes, including how to frame the value of optimisation work to senior stakeholders, in the conversion optimisation section of The Marketing Juice. The commercial case for CRO is straightforward, but it needs to be made clearly to get the internal investment the work deserves.

The Role of Speed in Ecommerce Conversion

Page speed deserves its own section because it is both well-understood and chronically under-addressed in ecommerce CRO programmes. The relationship between load time and conversion rate is not subtle. Slow pages lose visitors before they have seen the content, and the visitors most likely to leave are those on mobile connections, which is an increasingly large proportion of ecommerce traffic.

The technical causes of slow ecommerce sites are usually predictable: unoptimised images, too many third-party scripts loading on every page, poorly configured caching, and hosting infrastructure that has not kept pace with traffic growth. None of these are exotic problems. They are fixable with standard web performance work. But they require development resource and a willingness to prioritise performance over the accumulation of tracking pixels and marketing technology tags that tend to build up on ecommerce sites over time.

Core Web Vitals have made page speed a measurable, standardised concern. Largest Contentful Paint, First Input Delay, and Cumulative Layout Shift are not just SEO metrics. They are proxies for the experience your visitors are having. A product page with a poor LCP score is a product page where buyers are waiting for content to appear before they can make a decision. That wait has a commercial cost.

If I were auditing an ecommerce site today and had to choose between running a homepage A/B test and fixing a page speed issue that was adding two seconds to mobile product page load times, I would fix the speed issue without hesitation. The expected return is higher, the fix is permanent rather than experimental, and the benefit accrues to every visitor rather than just the test cohort.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions