Site Speed & CRO

Here’s the uncomfortable truth about site speed and conversions: the data is scary, and the fixes are boring. Study after study shows slow pages bleeding sales, which makes for great headlines — but the actual work is a checklist of unglamorous engineering tasks most marketers keep putting off. This post does two things honestly: it tells you what the speed-versus-conversion research really shows (and where it overstates the case), then gives you the high-impact fixes to ship first, in priority order.

What the studies really show — and their limits

The widely cited research points one direction. A large Portent study spanning many sites and a huge volume of pageviews found conversion rates highest for the fastest pages and falling steadily as load time grew — pages loading in around a second converted at roughly the highest rate, with conversion dropping notably by two and three seconds, and sites taking more than five seconds converting at around half the rate of fast ones. Google has long cited that even a one-second delay is associated with meaningfully fewer conversions — in the neighborhood of 7%.

Those numbers are real, but read them honestly: they’re correlational, not causal. A page that loads in six seconds is often also bloated, poorly built, badly hosted, and cluttered — all things that independently hurt conversions. Speeding it up to two seconds won’t magically fix a weak offer or a confusing layout. So treat speed as a necessary condition, not a silver bullet: slow pages reliably lose visitors before they can convert, and removing that friction lets everything else you’ve done actually get a chance to work.

Core Web Vitals, plainly

Google measures real-world speed and stability through Core Web Vitals — three metrics worth understanding without the jargon:

• LCP (Largest Contentful Paint) — how long until the main content loads. This is your “does it feel fast” number. Aim for under 2.5 seconds.
• INP (Interaction to Next Paint) — how quickly the page responds when someone taps or clicks. Laggy, unresponsive pages feel broken. Aim for under 200 milliseconds.
• CLS (Cumulative Layout Shift) — how much the layout jumps around as it loads. Jumpy pages cause mis-taps and frustration. Aim for under 0.1.

These matter twice over: they’re a ranking signal for SEO and a direct proxy for the experience that drives conversions. The same fixes serve both.

How to measure your speed

Don’t guess — measure. Google PageSpeed Insights gives you both lab scores and real-user field data (from the Chrome User Experience Report), plus a prioritized list of what’s slowing you down. Run it on your key pages, on mobile first — mobile is where most traffic and most speed problems live, and it’s usually slower than desktop. Lighthouse (built into Chrome) gives you the same lab audit locally. Fix for your real users, not for a perfect score on a fast connection.

The data is scary and the fixes are boring. Do them anyway — every second you shave removes a real reason a ready-to-buy visitor left before your page even loaded.

The high-impact speed fixes, in priority order

Most sites get the majority of their speed back from a handful of fixes. Work them in this order:

Images first

Images are the number-one cause of slow pages, and the easiest win. Compress every image, serve modern formats like WebP, size them to the dimensions they actually display at (don’t ship a 4000px image into a 400px slot), and lazy-load anything below the fold so it only loads when needed. On most sites this single category of fixes is the biggest LCP improvement available.

Caching and a CDN

Caching stores a ready-built version of your pages so the server isn’t rebuilding them on every visit. A CDN (content delivery network) serves your files from a location physically near each visitor instead of one distant origin server. Together they cut load time dramatically, especially for repeat visitors and geographically spread audiences, and they’re mostly set-and-forget once configured.

Render-blocking JavaScript, CSS, and third-party scripts

Scripts that must load before the page can display will hold your content hostage. Defer or async non-critical JavaScript, minify and streamline CSS, and audit your third-party scripts — chat widgets, analytics, ad pixels, A/B tools. Each one is someone else’s code running on your page, and a slow third party drags your whole site down. Remove the ones you don’t truly need.

Hosting and TTFB

If your time to first byte is slow, your server is the bottleneck and no amount of front-end tuning fully fixes it. Cheap shared hosting is a common culprit — your site shares resources with hundreds of others, and when they spike, you slow down. Upgrading to faster, properly resourced hosting raises the ceiling on everything else — it’s less exciting than tweaking images, but for a genuinely slow server it’s the real fix. A good tell: if PageSpeed Insights shows your content rendering fine once it starts, but there’s a long delay before anything happens, that gap is usually TTFB, and that’s a hosting conversation, not a front-end one.

Speed is necessary, not sufficient

Here’s the honest framing: a fast page doesn’t cause conversions — it removes a reason they don’t happen. Shave your load time and you stop losing the impatient visitors, but you still need a clear value proposition, a single strong CTA, and a short form to convert the ones who stay. Pair your speed work with the fundamentals in how to increase your website conversion rate and landing page best practices, then confirm your wins with A/B testing. Speed clears the runway; CRO flies the plane.

Frequently asked questions

Does site speed really affect conversions?

The correlation across large studies is strong and consistent — faster pages convert better, and sites taking more than five seconds convert at roughly half the rate of fast ones. Just read it honestly: it’s correlational, so speed removes friction rather than guaranteeing sales on its own. Fast pages let the rest of your optimization actually work.

What is a good page load time?

Aim for your main content to load in under about two and a half seconds (that’s the LCP target), with faster being better. The single-second range is where the top conversion rates show up in the research, so treat every second you shave as removing a reason visitors left.

What are good Core Web Vitals scores?

Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200 milliseconds, and Cumulative Layout Shift under 0.1. Hitting those “green” thresholds serves both your rankings and your conversions.

Is mobile or desktop speed more important?

Mobile, for most sites — it’s where the majority of traffic and the worst speed problems live, and it’s typically slower than desktop. Measure and optimize for mobile first, then confirm desktop holds up.

Just ship the boring fixes

You don’t need a perfect score — you need to stop losing visitors to a loading spinner. Measure with PageSpeed Insights, fix images, caching, scripts, and hosting in that order, and keep your Core Web Vitals in the green. It’s unglamorous, and it works. For the rest of the conversion playbook, browse the CRO resources. And if you want to learn where technical performance and conversion actually meet — from practitioners, with real numbers, no pitches — come to SEO Spring Training in Chandler, Arizona. Grab your ticket.