Website Speed Cost Calculator
Enter your homepage or the landing page your ads point to, to see how fast it loads for real visitors. Add your ad numbers and we'll estimate what a faster page would bring in each month, as a range, with every assumption and source shown.
Your results will appear here
- How fast your site loads for real visitors, on phones and desktops
- What the load time costs you in leads and dollars, as a range
- How much a faster site would bring in, with a slider to try different speeds
- What would make your site faster
How much does a slow website cost?
The best evidence we found suggests each second faster brings roughly 3% to 13% more conversions. A randomised test on paid traffic found about 3% per second; a regression over 10 million visits found about 13%. Gains continue below Google's 2.5-second "good" line, though the evidence there is thinner. A site that loads in 4.5 seconds on phones could be missing 6% to 26% of the leads or sales its ad budget pays for, if it loaded in 2.5.
How we calculate this
- We read the page's load time from Chrome's real-visitor data (the Chrome UX Report): the time it takes the main content to appear (Largest Contentful Paint, LCP) for 75% of visits, on phones and desktops. Chrome only has figures for pages with enough visits. When it has none for the page you entered, which is common for landing pages, we use a Google PageSpeed test of that page instead, and say so. We don't use whole-site figures for a specific page: a landing page can load very differently from the rest of a site.
- We take the seconds between your load time today and the load time you pick with the slider, which starts at Google's 2.5-second line, and weight phone and desktop by your mobile share. We cap the gap at 4 seconds, because no study covers a bigger gain.
- We multiply that gap by two rates of conversion gain per second, 3% and 13%, one from each of the two studies below, and show the whole range between them. They're our own reading of those studies, not published constants, and the research doesn't support picking one number in between.
- From your spend, cost per click and conversion rate we work out the conversions you get today, how many more the same budget would buy at the target speed, and your cost per conversion before and after.
The two rates
| Estimate | More conversions per second faster | Based on |
|---|---|---|
| conservative | 3% | Based on Vodafone's randomised A/B test on a paid-traffic landing page. |
| optimistic | 13% | Based on Renault's regression over 10 million landing-page visits. |
Sources we use
Vodafone: a 31% improvement in LCP increased sales by 8% · web.dev (Google), 2021
Cutting LCP from 8.3s to 5.7s on a paid-traffic landing page raised sales 8% and leads per visit 15%.
A randomised A/B test, 50/50, on one landing page receiving paid traffic. Several changes shipped together.
Used for the conservative estimate.
Renault: how LCP relates to bounce and conversion · web.dev (Google), 2021
Each second of faster LCP went with about 13% more conversions and 14 points less bounce.
A regression over 10 million+ landing-page visits in 33 countries, December 2020 to March 2021. A correlation, for one brand.
Used for the optimistic estimate.
Site speed is (still) impacting your conversion rate · Portent, 2022
On B2B lead-generation sites, goal conversion was about 40% at 1s, 34% at 2s and 29% at 3s of load time.
About 100 million page views across 20 sites, 14 of them B2B lead generation, over 30 days. Compares sites with each other; the load-time metric isn't defined.
The science behind Web Vitals · Chromium Blog (Google), 2020
Visitors to sites meeting the Core Web Vitals thresholds were 24% less likely to leave a shopping page before it loaded, and 22% less likely on news sites.
Chrome's analysis of millions of page impressions, where abandoning means leaving before any content appears. Published before INP replaced FID as a Core Web Vital.
Numbers we don't use, and why
Most speed calculators and pitches rely on these. We checked each against its original source.
53% of mobile visits are abandoned if a page takes longer than 3 seconds to load · Google (DoubleClick), 2016
53% of mobile site visits were abandoned when a page took longer than 3 seconds to load.
A one-line claim from data on 10,000+ mobile sites. No method was published.
Why we don’t use it: Ten years old, with no published method, and it measures leaving before load, not sales. It's often quoted without the word "mobile".
Bounce probability rises 32% as load time goes from 1s to 3s · Google and SOASTA, 2017
Bounce probability rises 32% from 1s to 3s, 90% at 5s and 123% at 10s.
A machine-learning model's prediction over about 900,000 mobile landing pages, not observed behaviour.
Why we don’t use it: It's a model's prediction, not a measurement, and we couldn't find the original method published anywhere.
Milliseconds Make Millions · Deloitte, for Google, 2020
A 0.1s faster mobile site went with 8.4% more retail conversions and 10.1% more travel conversions.
37 brands, about 30 million mobile sessions, over four weeks. The report calls it a correlation, and the 0.1s is an improvement in four speed metrics at once.
Why we don’t use it: It's a correlation for a 0.1s change and can't be scaled up: read as a rate, one second faster would mean 84% more sales. It has no conversion figure for lead generation.
Every 100ms of latency cost Amazon 1% of sales · Amazon (conference talk), 2006
100ms of added latency cost 1% of sales.
A remark in a talk. The slides are no longer online and no data was published.
Why we don’t use it: An anecdote with no published data, from a 2006 website.
A 100-millisecond delay can hurt conversion rates by 7% · Akamai, 2017
A 100ms delay in load time went with conversion rates 7% lower.
One month of real-user data, about 10 billion visits to retail sites.
Why we don’t use it: A real dataset, but nine years old and retail only. It's often misquoted as 7% per second.
Evidence last reviewed October 14, 2026.
Frequently asked questions
Is the 53% figure about mobile visitors still true?
The claim that 53% of mobile visits are abandoned after 3 seconds comes from a 2016 Google announcement with no published method, and it measures people leaving before the page loads, not lost sales. We don't use it.
What is a good load time?
Google counts a page as good when its main content appears within 2.5 seconds (Largest Contentful Paint) for at least 75% of visits. Above 4 seconds is poor.
Why do my results show a range instead of one number?
The research doesn't support one precise figure. The two best studies, a randomised test and a large regression, found about 3% and 13% more conversions per second faster. Any calculator that gives you a single number is hiding that uncertainty.
Why doesn't my site have real-visitor data?
Chrome only publishes data for sites with enough visits from Chrome users who share usage statistics. Many smaller sites don't qualify. We then fall back to a Google PageSpeed test, which is a single simulated load and less reliable.
Do you store my website or my numbers?
Speed results are cached for a day so a shared link loads quickly. Your ad numbers are saved in your own browser so you don't have to type them again. They're sent to us only if you request the PDF report, which is built from them, and they otherwise leave your browser only if you copy a link with your numbers and send it to someone.
