Key Takeaways
- Faster pages convert better. A 0.1 second gain lifted retail conversions by 8.4% in a 30 million session study
- Lead generation sites are affected too, not just online stores
- Check mobile field data, not your best lab score
- Fix the biggest image on the page first, then caching and scripts
Discover Your Revenue Opportunity
See how your website speed could be affecting conversions and revenue. Get your estimate in about 30 seconds.
Most businesses treat website speed as a technical chore. Someone runs a PageSpeed test, sees an orange number, and files it under “later.” Meanwhile the research says something much more expensive. In one analysis of more than 100 million page views, B2B sites that loaded in one second converted at about three times the rate of sites that took five seconds. For ecommerce, the gap was about two and a half times.
This guide pulls together the speed research that holds up, flags the numbers that do not, and shows where every statistic comes from. It also includes two tools: a calculator that estimates what slow pages cost you, and a Core Web Vitals checker that tells you what to fix first.
The short version
- Faster pages convert better, and the effect shows up in small steps. Deloitte and Google found that a 0.1 second improvement lifted retail conversions by 8.4% and travel conversions by 10.1%.
- Lead generation sites are not exempt. The same study saw 21.6% more visitors reach the form page on faster sites.
- Fewer than half of mobile sites pass all three Core Web Vitals (48% in 2025), and WordPress sites pass less often than average (45%).
- Lab scores can mislead you. Google judges what real visitors experience at the 75th percentile, not your best test run.
- Fix the biggest image on the page first. Both real case studies below got their gains from loading the main content faster.
What the Research Actually Shows
Dozens of blog posts quote the same handful of speed statistics, often without a source, a year, or a sample size. Here are the studies worth knowing, with the details most summaries leave out. The last column matters: it tells you how much weight each result deserves.
| Study | Year | Sample | Finding | Evidence type |
|---|---|---|---|---|
| Portent | 2022 | 100M+ page views, 20 B2B and B2C sites | A 1 second load converted about 3x better than a 5 second load for B2B, and 2.5x better for ecommerce | Observational |
| Deloitte and 55, for Google | 2020 | 37 brand sites, 30M+ sessions | 0.1s faster: retail conversions +8.4%, travel +10.1%, retail order value +9.2%, lead gen form-page progress +21.6% | Observational, large sample |
| Vodafone | 2021 | Live A/B test, about 34K visits per day per version | LCP 31% faster (8.3s to 5.7s): 8% more sales, 15% better lead-to-visit rate, 11% better cart-to-visit rate | Controlled A/B test |
| Nuvemshop | 2026 | Platform-wide, Jan 2025 to Jan 2026 | Core Web Vitals pass rate rose from 48% to 72%; mobile organic conversion rose 8.9% | Before and after (several changes at once) |
| 2016 | 10,000+ mobile domains | 53% of mobile visits likely abandoned if a page takes longer than 3 seconds | Observational, older data |
If you only trust one row, trust Vodafone. They split live traffic between a fast version and a slow version of the same page, which removes most of the “maybe fast sites are just better sites” objection. Portent and Deloitte compare sites or sessions that differ in more than speed, so treat them as strong patterns rather than guarantees. Nuvemshop changed several things at once, so we cannot credit any single fix. The 2016 Google figure is the oldest and comes from the 3G era, so use it as a rough benchmark, not a precise one.
None of this says your site will gain 8%. It says that across different companies, industries, designs, and years, the direction is the same: faster pages convert more visitors.
Conversion Rate by Load Time
Portent measured conversion rates against page load time across millions of sessions. These are the figures for ecommerce and for B2B lead generation.
Source: Portent, site speed and conversion rate research (updated 2022), 100M+ page views across 20 sites. Bar lengths are relative to the 1 second value in each chart.
One caution on the B2B chart: those rates are what Portent measured on the pages in its study, and they are far higher than most businesses see site-wide. Your own baseline will almost certainly be lower. Look at the shape of the curve, not the absolute numbers. Every added second costs conversions, and in ecommerce the early seconds cost the most.
The Data Most Businesses Ignore
Everyone has heard “slow sites lose sales.” These six findings are the parts that get skipped, and they change what you should actually do.
1. Tiny improvements pay off, not just big ones
The Deloitte and Google study did not measure dramatic overhauls. It measured a 0.1 second improvement, a gap too small for a visitor to consciously notice, and still found measurable gains across industries.
Source: Deloitte and 55 for Google, “Milliseconds Make Millions” (2020), 37 brand sites and more than 30 million sessions.
The practical lesson is that across the ranges studied, speed work keeps paying. Waiting for a “big redesign” to fix performance means leaving gains on the table for months.
2. Service and lead generation sites are affected too
Most speed statistics come from online stores, which leads service businesses to assume the research does not apply to them. It does. In the Deloitte study, lead generation sites saw 21.6% more visitors progress to the form submission page. In Portent’s data, B2B lead generation pages converted about 3x better at one second than at five, compared with 2.5x for ecommerce. If your website exists to book calls, quotes, or appointments, slow pages are costing you leads, not just sales.
3. Mobile is where most sites fail
Pages are usually tested on a fast office computer. Customers use phones on whatever connection they have. The HTTP Archive’s Web Almanac, which analyzes millions of real websites, shows the gap clearly.
Source: HTTP Archive, Web Almanac 2025, Performance chapter.
Loading the main content (LCP) is the weakest spot on mobile, with only 62% of sites passing. That is also the metric most closely tied to what a visitor feels when they land on your page.
4. A good lab score does not mean a good field score
This is the one that trips up the most businesses. There are two kinds of speed data:
Lab data is a simulated test of one page load, on one pretend device, on one pretend connection. It is what you get from the big score at the top of a PageSpeed test. Field data is what real visitors actually experienced, collected from real Chrome users over the last 28 days. Google assesses Core Web Vitals using field data, at the 75th percentile, which means three out of every four visits need to meet the threshold for a page to pass.
So a site can score 95 in a lab test and still fail in the field because real visitors are on older phones, slower networks, or hitting pages the test never touched. To see your field data, open your page in PageSpeed Insights and look at the section near the top about what real users are experiencing, or check the Core Web Vitals report in Google Search Console. If your site has too little traffic to show field data, test on a throttled mobile connection and treat the result as a rough guide.
5. WordPress sites pass less often, and the build matters more than the platform
Source: HTTP Archive, Web Almanac 2025, CMS chapter.
Before you conclude WordPress is the problem, read the Almanac’s own explanation: implementation choices often matter more than platform defaults. A lean theme with a handful of plugins can be very fast. The same platform with a heavy page builder, a dozen plugins, and a sliding hero banner usually is not. The Almanac also notes that tightly managed platforms improved fastest, while WordPress improved by only about four points year over year. If you run WordPress, your theme, plugins, hosting, and caching setup are where the gains are. Our guide to why even small websites need a CDN covers one of the highest-impact fixes.
6. Speed supports rankings, but it is not a magic ranking lever
Google’s documentation says Core Web Vitals align with what its core ranking systems seek to reward, and it recommends that site owners achieve good Core Web Vitals. It does not say how much weight they carry, and it is clear that great content still matters more than raw speed. That makes the conversion argument stronger than the ranking argument. Even if speed never moved a single ranking, faster pages would still turn the traffic you already have into more customers.
Core Web Vitals in Plain English
Core Web Vitals are three measurements Google uses to describe how a page feels to a real person. Here is what each one means and where the lines are.
| Metric | What it means | Good | Needs improvement | Poor |
|---|---|---|---|---|
| LCP Largest Contentful Paint |
How long until the main content (usually the hero image or headline) appears | 2.5s or less | 2.5s to 4s | Over 4s |
| INP Interaction to Next Paint |
How quickly the page responds when someone taps or clicks | 200ms or less | 200ms to 500ms | Over 500ms |
| CLS Cumulative Layout Shift |
How much the page jumps around while loading | 0.1 or less | 0.1 to 0.25 | Over 0.25 |
A page passes only when all three are in the “good” range at the 75th percentile of real visits. For a deeper look at how to check yours, see how to check if your Core Web Vitals suck and our write-up on the latest Core Web Vitals changes.
Check Your Core Web Vitals in 30 Seconds
Pull your numbers from PageSpeed Insights (use the field data section, mobile tab) or Search Console, then enter them below. The checker tells you which metrics pass and which one to fix first.
Core Web Vitals Checker
Enter your mobile field data at the 75th percentile. Nothing you type leaves your browser.
What Actually Moved the Needle: Two Real Case Studies
Vodafone: 31% faster LCP, 8% more sales
Vodafone ran a proper A/B test on a landing page, sending half of its traffic to a version optimized for Web Vitals and half to the original. The optimized version had a 31% better Largest Contentful Paint (5.7 seconds against 8.3 seconds). The result was 8% more sales, an 11% better cart-to-visit rate, and a 15% better lead-to-visit rate. The team got there by rendering critical page content on the server and optimizing images: resizing them, optimizing SVGs, lazy loading, and compressing PNGs.
Nuvemshop: pass rate from 48% to 72%, conversions up 8.9%
Nuvemshop, an ecommerce platform, raised the share of its stores passing Core Web Vitals from 48% to 72% between January 2025 and January 2026. Mobile visitors arriving from organic search converted 8.9% more often. Their fixes were small and specific:
- Removed CSS transitions from the first sections of the page
- Stopped lazy loading images that sit above the fold
- Marked the main image as high priority so the browser fetched it first
- Added edge caching, while protecting live pricing and inventory data
Notice the pattern. Neither company rebuilt its site. Both focused on getting the most important content in front of the visitor sooner. Whatever else you do, start with the biggest image or headline a visitor sees when they land.
Speed Fixes, Ranked by Effort and Payoff
This is a practical order of attack for a typical small business website. The effort ratings are our judgment from the kinds of changes involved, not a published benchmark.
| Fix | Helps | Effort | Notes |
|---|---|---|---|
| Compress and resize the hero image | LCP | Low | Use a modern format and serve the size the screen actually needs |
| Never lazy load the top image | LCP | Low | Lazy loading belongs below the fold only |
| Prioritize the main image | LCP | Low | Preload it or mark it high priority so it loads first |
| Add caching and a CDN | LCP, server response | Low to medium | Often the biggest single win for WordPress sites |
| Remove unused plugins and scripts | LCP, INP | Medium | Every plugin you delete is code visitors never download |
| Delay third-party scripts | INP | Medium | Chat widgets, tag managers, and heat maps are common culprits |
| Reserve space for images, embeds, and ads | CLS | Low to medium | Set width and height so the layout does not jump |
| Upgrade hosting or PHP version | Server response | Medium | Worth it if your server is slow before anything else loads |
What Is a Slow Site Costing You?
This calculator turns the research into your numbers. Enter your traffic, your conversion rate, what a conversion is worth, and your current and target load times. It estimates the extra revenue a faster site could produce.
Speed Revenue Calculator
Use your own numbers for the pages that earn you leads or sales. Nothing you type leaves your browser.
Your 30-Day Speed Plan
| When | Do this | Done when |
|---|---|---|
| Week 1 | Pick your 5 most valuable pages (the ones that earn leads or sales). Record mobile field data for each from PageSpeed Insights or Search Console. Note your current conversion rate for each. | You have a before snapshot in a spreadsheet |
| Week 2 | Fix the LCP image on each page: compress, resize, prioritize, and remove lazy loading from it. Turn on caching and a CDN if you have not. | Lab LCP is under 2.5s on a throttled mobile test |
| Week 3 | Audit plugins and third-party scripts. Remove what you do not use and delay what you cannot remove. | Fewer scripts load before the page is usable |
| Week 4 | Reserve space for images and embeds to stop layout shifts. Run the lab tests again and annotate the date in your analytics. | CLS is 0.1 or less on every key page |
| Weeks 5 to 8 | Wait. Field data uses a rolling 28 day window, so real-user numbers update slowly. Then compare conversion rates against your Week 1 snapshot. | Field data shows “good” and conversion has been measured |
How to Measure It Without Fooling Yourself
- Use field data first. The real-user section of PageSpeed Insights and the Search Console Core Web Vitals report show what Google sees.
- Test mobile, throttled. A test on fast office Wi-Fi flatters you.
- Measure by landing page. Sitewide averages hide the pages that matter. Track conversion rate per page before and after changes.
- Change one batch at a time. If you rebuild the whole site and conversions move, you will not know why.
- Allow for seasonality. Compare the same weeks in different months or years, not a holiday week against a quiet one.
- Do not chase 100. A lab score of 100 is not the goal. Passing in the field on the pages that earn money is.
Common Speed Mistakes
| Mistake | Why it hurts | Do this instead |
|---|---|---|
| Chasing a perfect lab score | Lab results differ from what real visitors experience | Aim to pass Core Web Vitals in field data on your money pages |
| Testing on desktop only | Mobile is where most sites fail | Check mobile field data first |
| Lazy loading every image | It delays the hero image, which is usually the LCP element | Lazy load below the fold only |
| Installing a plugin to fix speed | Extra code can slow the site more than it helps | Remove first, then add only what earns its place |
| Optimizing only the homepage | Many visitors land on service, product, or blog pages | Start with the pages that bring in leads or sales |
| Checking once and moving on | New plugins, images, and scripts slow sites down over time | Re-check monthly |
Frequently Asked Questions
How fast should a website load?
Aim for a Largest Contentful Paint of 2.5 seconds or less at the 75th percentile of mobile visits. The research suggests conversions keep improving at lower load times, so treat 2.5 seconds as the minimum target rather than the finish line.
Does website speed affect conversion rate?
Yes. Portent, Deloitte and Google, and Vodafone all found faster pages converting better, using different methods and sample sizes. Vodafone’s controlled test saw 8% more sales from a 31% faster Largest Contentful Paint.
How much does a one second delay cost?
It depends on your traffic, conversion rate, and what a conversion is worth. In Portent’s data, ecommerce conversion rates fell from 3.05% at one second to 1.68% at two seconds. That is a pattern across the sites studied, not a promise for yours. Use the calculator above to estimate your own number.
Is a PageSpeed score of 90 or higher enough?
Not by itself. The score comes from a lab test, while Google assesses real-user field data. A site can score 95 in the lab and still fail Core Web Vitals in the field.
Does page speed affect SEO?
Google says Core Web Vitals align with what its core ranking systems seek to reward, and recommends achieving good scores. It does not publish how much weight they carry, and content relevance still matters more. The conversion benefit is easier to prove than the ranking benefit.
Why is my WordPress site slow?
Usually because of the build, not the platform. Heavy themes, page builders, too many plugins, unoptimized images, and missing caching are the common causes. Only 45% of WordPress sites passed Core Web Vitals on mobile in 2025, so you are far from alone.
Sources
- Portent: site speed and conversion rate research (updated 2022)
- Deloitte and 55 for Google: Milliseconds Make Millions (2020)
- web.dev: Vodafone Web Vitals A/B test (2021)
- web.dev: Nuvemshop performance case study (2026)
- HTTP Archive Web Almanac 2025: Performance
- HTTP Archive Web Almanac 2025: CMS
- web.dev: Web Vitals
- Google Search Central: Core Web Vitals
- Google: The need for mobile speed (2016)