A client came to us with a mobile PageSpeed score of 38. The site was visually polished — a legitimate brand with real distribution, stocked in major retailers, shipping to 70+ countries. But the product pages were taking over 6 seconds to show anything useful on mobile.
They thought it was an SEO problem. They were partially right. But the revenue impact was coming from somewhere more immediate than search rankings.
The double hit you don't see in your dashboard.
A slow site costs you twice, and the two effects compound each other.
The first hit is direct: users leave before the page loads. Google's own research puts the numbers bluntly. A 1-second delay in mobile load time reduces conversions by 7%. At 3 seconds, 53% of mobile users abandon. At 6 seconds — which is where our client was — the majority of visitors who arrived from paid campaigns never saw the product. They'd already left.
The second hit is slower and more insidious: Google deprioritises slow sites in organic search. Core Web Vitals are a ranking signal. If your Largest Contentful Paint (LCP) is failing and your Cumulative Layout Shift (CLS) score is unstable, you're competing with a handicap on every keyword. You're paying for rankings you'd have organically if the site was fast.
So the maths for a slow e-commerce site looks like this: you're spending more on paid acquisition because organic is suppressed, and your paid acquisition is converting at a fraction of what it should because users bounce before the page loads. Both costs are invisible in most dashboards.
Where speed actually comes from.
The common misconception: speed is a plugin you install. Enable a caching layer, compress images, and you're done. This is true enough to be dangerous — it produces marginal improvements that don't move the needle, and teams walk away thinking speed is solved when it isn't.
Real PageSpeed improvement comes from addressing the root causes:
Images are usually the largest contributor to slow load times, and the most underestimated. Not because images are large — though they are — but because they're typically loaded badly. An image that renders at 400px on mobile served at 1400px original resolution with no lazy loading, in a format that's 3× larger than the modern equivalent, will kill your LCP by itself. Most sites have dozens of these.
Render-blocking resources are the second major culprit. JavaScript files that must execute before any content appears, CSS that blocks rendering, third-party scripts loaded synchronously in the head — these all push the point at which the browser can show anything useful to the user. Eliminating them or deferring them correctly is the difference between a 2-second LCP and a 6-second one.
JavaScript bundle size matters more on mobile than desktop. A 4G connection loading a 2MB JavaScript bundle takes longer than you expect, and JavaScript also blocks the main thread during parsing and execution, which delays interactivity. Tree shaking, code splitting, and removing unused dependencies are engineering tasks, not plugin settings.
Correct CDN configuration means that assets are served from an edge node close to the user — not from a single origin server. Most platforms have CDN support built in; most teams don't configure it correctly, or use it only for images while serving JavaScript and CSS from origin.
The case study in numbers.
For the client above, the fix took three weeks of focused engineering. No visual redesign. No new features. Just a systematic audit and rebuild of the performance layer: next-gen image formats and correct sizing, deferred JavaScript, critical CSS inline with the rest loaded asynchronously, CDN configuration corrected, and LCP element prioritisation on the above-the-fold product images.
Mobile PageSpeed: 38 → 91. LCP: 6.2s → 1.4s. The team called it "incredible" — which is a reasonable reaction when a metric that was dragging organic rankings and killing paid conversion improves by that margin.
We didn't track post-launch conversion data directly for this client, but the directional math on a site spending meaningfully on paid acquisition is not hard to do. If 53% of mobile users were abandoning at 3 seconds, and the site was well past 3 seconds, the recaptured revenue from users who now see the page is significant before you even consider the organic search improvement.
How to audit your own site.
This takes 10 minutes and the tools are free:
Run your site through PageSpeed Insights on mobile. If the score is below 50, you have a problem. If it's below 70, you likely have a problem that's affecting revenue. Look specifically at the LCP and CLS diagnostics — these are the metrics Google weighs most heavily for rankings.
Then look at the "Opportunities" and "Diagnostics" sections. PageSpeed will tell you exactly what's slow and how much time each issue is costing. The items with the largest estimated savings are where to start.
If the top items are image-related: you can make real progress with a developer who knows what they're doing in a few days. If the top items are JavaScript-related: you're likely looking at a week or more of careful engineering work to do it properly.
The businesses that grow their organic share over time are the ones that treat site performance as a product quality metric, not a technical nice-to-have. The fast site compounds. The slow one doesn't.