The Checkout That Loaded Too Slowly

Mobile checkout completions fell 28% over six weeks. The team blamed the new design. GA4 told a different story.

By WebSignalytics Inc.  ·  26 Jul 2026  ·  6 min read

The scenario

Ridgeback & Co is a small homeware and gifts e-commerce shop run by two founders out of Halifax. They sell hand-sourced ceramics, textiles, and seasonal gift bundles — the kind of products that photograph well and attract a loyal following on Instagram. Most of their traffic is mobile. Priya handles marketing and digital; her co-founder handles buying and operations. They have one dev contractor they bring in for bigger changes.

In late spring, they launched a redesigned checkout experience. The brief was simple: fewer clicks, cleaner layout, inline address autocomplete instead of separate fields. The new design looked sharp on every device they tested. They'd been using an older checkout flow that customers sometimes abandoned partway through, and the redesign was supposed to fix that. Priya was optimistic.

Six weeks after launch, she opened GA4 and pulled up the purchase key event report. Mobile purchase completions had fallen by 28% compared to the same six-week window before the redesign. On a business where most revenue comes through mobile, that number was impossible to ignore. She immediately assumed the new checkout was the problem.

The confident wrong conclusion

The logic felt airtight: they'd launched a new checkout, and mobile conversions had dropped. The new design had more interactive elements — inline step validation, a live address lookup, animated progress indicators. Priya's first instinct was that mobile users were finding the new interface confusing or overwhelming. Maybe the inline validation messages were appearing at the wrong moment. Maybe the autocomplete was triggering keyboard issues on certain Android browsers. The design that looked elegant on a test device might be friction in disguise for real customers.

She started drafting a brief to revert to the old checkout flow. The plan was to roll back the design, stabilise the conversion numbers, and then re-approach the redesign more cautiously. The dev contractor was on standby. She was a day away from sending that brief when she decided to spend another hour in GA4 first.

The overlooked device asymmetry

Priya pulled up the device category breakdown in GA4's Tech report — a straightforward segmentation by desktop, mobile, and tablet. What she found stopped her: desktop checkout completions were entirely unchanged. Not down slightly. Flat. The same customers completing purchases on desktop were still completing purchases at exactly the same rate as before. The 28% drop was mobile-only.

That asymmetry changed everything. A UX problem with the checkout flow — confusing validation, a problematic autocomplete — would have affected every user regardless of device. The fact that desktop was completely unaffected meant the issue wasn't the design logic. It was something that behaved differently between desktop and mobile. She opened the Funnel Exploration report in GA4 and mapped the checkout steps. Users on mobile were reaching the payment input step — they weren't dropping off at the cart page or the address step. They were getting all the way to payment and then abandoning. Not earlier. Not at the new autocomplete field. Specifically at payment entry on mobile.

That was enough to make her pause the revert. She asked the dev contractor to run a PageSpeed Insights test on the checkout page — specifically on the mobile simulation. The result: 7.4 seconds to interactive on a mid-tier mobile connection. The redesign had added a high-resolution animated product thumbnail gallery — a nice touch that showed customers a visual summary of what they were buying — but it loaded synchronously, blocking the payment form from rendering until all the images had downloaded. On desktop and fast wifi connections, it was invisible. On a typical 4G mobile connection, it was a seven-second wall between the customer and the ability to enter their card details.

The corrected interpretation

The checkout design wasn't confusing anyone. The checkout was slow — specifically on mobile, specifically at the payment step, because a synchronous image gallery was blocking the render of the most important field on the page. Customers weren't abandoning because they were uncertain. They were abandoning because the page appeared broken. Seven seconds without a payment form is enough for almost anyone to close a tab and assume something went wrong.

The fix took the contractor about three hours: lazy-loading the thumbnail gallery so it rendered after the payment form, compressing the images, and deferring the animation until after the critical path had loaded. Within two weeks of pushing that change, mobile checkout completions had recovered to within 4% of the pre-redesign baseline. The design itself was never the issue. If Priya had reverted, she would have lost the cleaner checkout and still eventually hit the same performance problem the next time a heavier asset was added.

What to do next

If you see a conversion drop that's isolated to one device category, don't assume it's a design problem before you've checked performance. Here's how to work through it systematically:

• Never interpret a mobile-only conversion drop as a design problem before checking page speed — asymmetric drops by device almost always point to a technical or performance difference, not a UX one.
• Use GA4's Tech report (Audience > Tech > Device category) to isolate whether a drop is affecting all devices equally or one segment in particular — this is the single fastest way to rule out a UX explanation.
• Run PageSpeed Insights on mobile simulation (not just desktop) on any revenue-critical page; Google's desktop score will almost never surface an image loading issue that only shows up on slower mobile connections.
• Use GA4's Funnel Exploration report to identify exactly which step in a checkout or form flow is losing users — a drop at the payment step specifically, rather than at the first field, is a performance signature, not a confusion signature.
• Watch for the combination of low mobile engagement rate and high exit at a specific funnel step — together, those two signals are a reliable indicator that a page is loading slowly and users are abandoning before the critical content appears.

For a guide to finding slow pages across your site using GA4, see How to Find Slow Pages in GA4.

Ridgeback & Co and Priya are illustrative — a composite created to demonstrate a real and common pattern.