A slow website rarely fails loudly. Nobody emails to say a page took too long; they tap back and choose someone else, and your analytics may not even record the visit. That is why the link between page speed and conversions is easy to believe in and hard to put a figure on.
This guide covers both halves: how slowness costs leads and sales, and how to measure what it costs on your own site before you budget for performance work.
The short answer
Page speed affects conversions through a few concrete mechanisms, not a universal formula.
- Slow first screens lose visitors before they read a word, often before analytics records them.
- Paid clicks cost the same whether or not the page loads, so slow landing pages waste ad spend directly.
- Laggy buttons (poor INP) and jumping layouts (poor CLS) make enquiries and checkouts harder to finish.
- The size of the effect varies by site. To know what slowness costs you, segment your conversions by device and page, compare them with Chrome’s real-user speed data, and measure before and after each fix.
Four ways page speed affects conversions
1. Visitors leave before the page appears
Someone taps your Google result or social ad and sees a white screen. On office Wi-Fi it lasts a moment; on a mid-range phone using mobile data it can last several seconds, each one an invitation to try the next result.
Why it’s hard to see. Analytics only records a visit once its tag has loaded and run. Someone who gives up before then never becomes a session, so the worst speed losses are invisible in the reports you would naturally check.
The metric to watch is Largest Contentful Paint (LCP), which Google rates as good at 2.5 seconds or less for at least 75% of real visits. Core Web Vitals explained covers each metric and its usual causes.
2. Paid clicks are wasted on slow landing pages
With pay-per-click advertising you pay when someone clicks, not when your page finishes loading. Everyone who gives up on a blank screen is a visit you bought and never received.
Google Ads also counts landing page experience (how relevant and useful the page is to people who click) as one of three Quality Score components. Google’s advertiser help recommends fast-loading mobile pages but doesn’t publish how much speed feeds that rating, so treat any Quality Score gain as a bonus. The wasted clicks are the certain cost.
3. Buttons and forms that lag
A page can look ready and still not respond: the visitor taps “Send enquiry” and nothing happens, because the browser is busy running scripts. Some tap again and send a duplicate; others assume the form is broken and leave. On a shop, the same delay hits filters and “Add to basket”.
Interaction to Next Paint (INP) measures this, and Google rates 200 milliseconds or less as good. Chat widgets, piles of tracking tags and heavy page builders are common culprits. Forms sit at the exact point of conversion, so web form design deserves a look too.
4. Layout shifts that cause mis-taps
A late-loading banner or image pushes the text down. The visitor reaches for “Call us” and hits a different link, at the very moment they were ready to act.
Cumulative Layout Shift (CLS) measures this, with 0.1 or less rated good. It is often the cheapest fix of the three: reserve space for images, embeds and banners before they load.
What published research shows, and how to read it
The most useful public evidence comes from controlled experiments. In Vodafone’s case study on web.dev (2021), Vodafone Italy split traffic between two versions of a landing page: the one with a 31% better LCP in field data generated 8% more sales and a 15% uplift in its lead-to-visit rate. Web.dev’s round-up of the business impact of Core Web Vitals collects results from other large sites.
Observational research points the same way. Milliseconds Make Millions, a 2020 study commissioned by Google and carried out by Deloitte and fifty-five, tracked 37 retail, travel, luxury and lead-generation brand sites and found that a 0.1-second improvement in mobile speed was associated with more visitors moving through each funnel.
Before applying any of it to your business, ask:
- Who ran it? Much of the evidence comes from Google or companies it worked with: not wrong, but not neutral.
- Experiment or correlation? An A/B test isolates speed; a study that only observes can’t rule out that faster sites, or faster hours, differ in other ways too.
- What kind of site? A telecoms brand with huge traffic behaves differently from a manufacturer whose buyers take months to decide.
- Where did they start? Very slow to acceptable usually pays far more than fast to slightly faster.
Published results show the effect is real, not its size on your site.
How to measure what slowness costs on your own site
This is where the business case gets built, with free tools: Google Analytics 4, Search Console and PageSpeed Insights.
Step 1: Make sure you trust your conversion data
Check that GA4 records enquiries, calls, messaging clicks or purchases as key events, and that last month’s count roughly matches what your team actually received. If not, fix tracking first; tracking enquiries in GA4 walks through it.
Step 2: Segment conversions by device and landing page
In GA4, open Explore and build a free-form exploration with Device category and Landing page + query string as rows, and Sessions, Key events and Session key event rate as values, over at least three months. Look for two patterns:
- A large mobile–desktop gap on the same pages. Some gap is normal, since many people research on a phone and enquire from a computer, but a mobile rate far below desktop on phone-heavy pages deserves a closer look.
- Key pages that convert worse than similar ones. If two service pages attract similar visitors, but one converts at half the rate and is also much slower, follow that lead.
Step 3: Compare with real-user speed data
Put Chrome’s real-user data, the Chrome UX Report (CrUX), beside those segments. PageSpeed Insights shows it per URL where traffic allows (for the whole origin otherwise), for mobile and desktop separately. Search Console’s Core Web Vitals report groups similar URLs, so you can see whether a problem belongs to one template or the whole site. How to read PageSpeed Insights explains which parts to act on.
Line them up in one table: page or template, device, sessions, conversion rate, and LCP, INP and CLS status. If the pages that fail Core Web Vitals also convert worst, you have a hypothesis to test, not proof: a slow page may also be a weak one. CrUX Vis, Google’s viewer for CrUX data, charts weekly field data going back several months, so you can check whether a dip in enquiries followed a slowdown after a plugin update or a new tag.
Step 4: Connect speed to individual visits
The strongest evidence links speed to each visit. Google’s open-source web-vitals library can send LCP, INP and CLS from real visitors to GA4 as events; web.dev’s guide to measuring performance with GA4 and BigQuery shows how. Then compare conversion rates for visits with good and poor LCP within the same device type and country, because slow visits skew towards older phones and weaker connections.
Step 5: Check paid clicks against landing sessions
Compare each campaign’s ad-platform clicks with its GA4 sessions over the same dates. Consent choices, ad blockers and tagging gaps mean they never match exactly, but if one slow landing page loses far more between click and session than the others, the difference is a rough proxy for people who gave up waiting.
Step 6: Measure fixes before and after, properly
- Record a baseline of at least four weeks: conversion rate by device for the affected pages, plus their field data.
- Change one thing at a time where you can, and note the release date.
- Keep a comparison group. Fix one template first and leave a similar one untouched; if only the fixed pages improve over the same weeks, seasonality and campaigns are unlikely explanations.
- Allow about a month. CrUX figures in PageSpeed Insights and Search Console cover the previous 28 days.
- Compare like with like: equal periods, the same mix of weekdays, and year-on-year if your business is seasonal.
With enough traffic, an A/B test like Vodafone’s is the cleanest answer; A/B testing landing pages covers when that is realistic.
Putting a number on the cost of a slow website
With segments in hand, a rough estimate is simple arithmetic. A hypothetical example:
| Input | Example |
|---|---|
| Monthly mobile sessions on slow service pages | 3,000 |
| Their conversion rate | 1.0% |
| Rate on comparable fast pages, same device | 1.4% |
| Enquiries that become customers | 20% |
Closing that gap would mean about 12 more enquiries a month and two or three more customers. Multiply by your average customer value, then add lost paid clicks times your cost per click. Treat the total as an upper bound, since part of the gap may not be about speed, but it turns “the site feels slow” into a figure you can set against the cost of fixing it.
When speed work is worth paying for
| Situation | Likely verdict |
|---|---|
| Key templates fail Core Web Vitals on mobile, and most visitors use phones | Worth prioritising |
| Paid campaigns land on slow pages | Worth prioritising: the waste is direct |
| Pages pass Core Web Vitals but convert poorly | Speed isn’t the bottleneck; look at message, proof and forms |
| Too little traffic for field data | Test on a real phone and in the lab; make the cheap fixes first |
| An online shop with many product pages | Measure by template; see e-commerce conversion rate |
If speed isn’t the constraint, the conversion rate optimisation guide helps you find what is. If it is, why is my website slow? pins down the cause, and the website speed optimisation guide ranks the fixes by impact and effort.
A documented example, with a caveat
On a property developer’s website we redesigned, load time fell from about nine seconds to two, and buyer enquiries rose 40% in the three months after launch. Speed was one change among several, including a buyer-journey structure, phone-friendly layouts and tracked enquiry forms, so it was part of a package that worked, not the sole cause. What a property developer’s website needs covers the rest.
Frequently asked questions
Does website speed affect sales?
It can, especially on phones and paid landing pages, where slow first screens lose visitors before they see the offer. Controlled tests such as Vodafone’s have linked faster pages to more sales, but the size of the effect depends on your site, audience and starting point.
Why doesn’t my bounce rate show the effect of a slow site?
In GA4, bounce rate is the share of sessions that weren’t engaged. Visitors who leave before the analytics tag loads never become sessions, so the heaviest losses from a slow first screen never reach that report.
Does page speed affect Google Ads costs?
It raises your cost per lead, because you pay for every click whether or not the page loads. Its effect on the price of each click is less clear: landing page experience is part of Quality Score, but Google doesn’t say how much speed contributes to it.
What to do next
Speed is worth paying for when your own data shows slow pages and weak conversions in the same places, especially on phones and paid landing pages. Steps 2 and 3 above can tell you in an afternoon whether speed is a likely cause or a distraction.
If you’d like a second opinion before you budget for anything, ask for a free website audit. A real person checks how your key pages load and behave on a phone, and points out where slowness is most likely getting between visitors and an enquiry. Keep the report: it doubles as the “before” picture for any fixes you make.