Website Speed and Conversions: What the Research Actually Says

Everyone agrees a fast website is "better." But is speed worth investing in compared with a redesign, new content or more advertising? The research says yes, and that improvements as small as a tenth of a second can show up in the numbers.

Here's what the best-known studies found, what they don't tell you and where to start if your site feels sluggish.

The research

"Milliseconds Make Millions" (Google and Deloitte, 2020)

The most cited study on speed and revenue was commissioned by Google and carried out by Deloitte with the data agency Fifty-Five. The team analyzed mobile site data from 37 retail, travel, luxury and lead generation brands across Europe and the US over four weeks, covering more than 30 million user sessions. No redesigns happened during the study, so the researchers could look at how natural fluctuations in speed related to user behavior.

Based on a 0.1-second improvement in mobile site speed, they observed:

• Retail: conversions up 8.4% and average order value up 9.2%.
• Travel: conversions up 10.1% and average order value up 1.9%.
• Luxury: page views per session up 8.6%.
• Lead generation: bounce rate on information pages improved by 8.3%.

(Deloitte, Milliseconds Make Millions, 2020; summary on web.dev)

One important caveat: this is a correlation study across many brands, not a guarantee that shaving 0.1 seconds off your site will lift conversions by 8%. But it's consistent with a large body of evidence pointing the same direction.

Bounce rates rise quickly as pages slow down

Google research cited in the same report found that as page load time goes from 1 second to 3 seconds, the probability of a visitor bouncing increases by 32%. (cited in Milliseconds Make Millions, 2020)

An earlier Google analysis, published in 2017, reported that 53% of mobile site visitors leave a page that takes longer than three seconds to load. (AMP Project blog, 2017) Networks and phones have improved since then, but user expectations have only grown.

Vodafone: a controlled A/B test (2021)

Correlation studies are useful, but controlled experiments are more convincing. Vodafone ran an A/B test on a landing page where the only difference between the two versions was performance optimization; they were visually and functionally identical. The optimized version had a 31% better Largest Contentful Paint (a measure of how quickly the main content appears). The result: 8% more sales, a 15% improvement in lead-to-visit rate and an 11% improvement in cart-to-visit rate. (web.dev case study, 2021)

What did Vodafone change? Mostly unglamorous fundamentals: moving some rendering from the browser to the server, reducing render-blocking JavaScript and optimizing images.

Why speed affects conversions

It's not mysterious. A slow page:

• Loses impatient visitors before they see your offer, especially on mobile and slower connections.
• Feels untrustworthy. Sluggish or jumpy pages make a business seem less professional.
• Adds friction at every step. If each page is slow, the whole journey from landing page to contact form feels like work.
• Wastes ad spend. You pay for the click whether or not the page loads in time.

Speed is also part of how Google evaluates page experience. Google states that Core Web Vitals, its set of speed and stability metrics, are used by its ranking systems, while cautioning that good scores alone don't guarantee top rankings. (Google Search Central) We explain those metrics in Core Web Vitals in plain English.

How to check your own site

1. PageSpeed Insights (pagespeed.web.dev): enter a URL to see both real-user data (if your site has enough traffic) and a lab test with specific recommendations.
2. Google Search Console's Core Web Vitals report: shows which groups of pages are slow based on real visitor data.
3. Test on a real phone, on mobile data rather than office Wi-Fi. It's the experience many of your visitors actually have.

Focus on your most important pages first: the homepage, top landing pages, service or product pages and the contact or checkout flow.

The usual culprits (and fixes)

• Oversized images. Resize to the dimensions actually displayed, use modern formats like WebP or AVIF and lazy-load images below the fold. This is often the single biggest quick win.
• Too many third-party scripts. Chat widgets, old tracking pixels, heatmap tools and social embeds each add weight. Audit them and remove what nobody uses.
• Heavy themes and page builders. Some themes load large amounts of code on every page. Sometimes a lighter theme or rebuild is the most cost-effective fix.
• Slow hosting or no caching. A slow server delays everything. Page caching and a content delivery network (CDN) help a lot.
• Render-blocking CSS and JavaScript. Defer non-essential scripts and inline critical styles so content appears sooner.
• Web fonts. Limit font families and weights, preload the important ones and use font-display: swap.
• Layout shifts. Reserve space for images, ads and embeds so content doesn't jump around while loading.

Prioritize by business impact

Not every speed fix is worth the same. A sensible order:

1. Fix your highest-traffic and highest-value pages first.
2. Take the quick wins such as images, unused scripts and caching.
3. Tackle structural issues like the theme, hosting or architecture if quick wins aren't enough.
4. Measure before and after. Track conversion rate and bounce rate on the pages you changed, not just speed scores.

For example, a local service business might find its homepage loads a 4 MB uncompressed hero photo and five tracking scripts it no longer uses. Compressing the image and removing the dead scripts could noticeably speed up the page in an afternoon. This is illustrative, but it's a very common pattern.

The bottom line

The evidence from correlation studies and controlled experiments points the same way: faster pages tend to convert better. You don't need a perfect score. You need your important pages to load quickly and smoothly for real visitors on real phones. Start with the biggest, cheapest fixes and measure what changes.