
The latest field data says 48% of mobile websites and 56% of desktop websites pass all three Core Web Vitals, per the HTTP Archive Web Almanac 2025 Performance chapter, which reads Chrome UX Report (CrUX) data from July 2025. In 2022 the same report counted 39% of mobile and 44% of desktop sites passing, back when First Input Delay (FID) was still a Core Web Vital. Recalculated with today’s metric (INP), 2022 mobile drops to 31%, so mobile pass rates have risen 17 points in three years.
Below: the 2025 numbers by metric, device, page type and CMS; the full 2022 data set for anyone searching “website speed stats 2022”; the exact thresholds and INP switchover date; the classic speed and conversion studies traced back to what they actually measured; what Core Web Vitals consulting costs at published prices; and a fix checklist with pass/fail criteria.
Latest website speed statistics: Core Web Vitals pass rates in 2025
A site “passes” when at least 75% of real Chrome visits hit the “good” threshold on LCP, INP and CLS. These are the shares of websites rated good in the 2025 Web Almanac (CrUX, July 2025, global).

| Metric (good threshold) | Mobile good | Desktop good | Change vs 2024 |
|---|---|---|---|
| All three Core Web Vitals | 48% | 56% | Mobile +4 pts (44%), desktop +1 pt (55%) |
| LCP, Largest Contentful Paint (2.5 s or less) | 62% | 74% | 13% of mobile sites rated poor |
| INP, Interaction to Next Paint (200 ms or less) | 77% | 97% | Mobile +3 pts (74%) |
| CLS, Cumulative Layout Shift (0.1 or less) | 81% | 72% | Both devices were 62% in 2021 |
| FCP, First Contentful Paint (diagnostic) | 55% | 70% | Mobile +4 pts, desktop +2 pts |
| TTFB, Time to First Byte (0.8 s or less, diagnostic) | 44% | 55% | Mobile +2 pts, desktop +1 pt |
Three patterns matter for anyone deciding where to spend on speed:
- LCP is the binding constraint on mobile. At 62% it is the metric fewest mobile sites pass, so it is where most mobile failures start. Fewer than half of mobile sites (44%) have a good server response time, and a slow TTFB eats into the 2.5 second LCP budget before anything renders.
- Desktop is a CLS problem, not a speed problem. Desktop beats mobile on every metric except layout stability (72% vs 81%).
- JavaScript is getting heavier while INP improves. Median mobile Total Blocking Time in the lab rose 58%, from 1,209 ms to 1,916 ms, between the 2024 and 2025 editions. Field INP still improved, but the headroom is shrinking on script-heavy sites, which is why JavaScript SEO and rendering choices now show up in performance audits too.
Home pages pass less often than inner pages
| Segment (2025) | Mobile good CWV | Desktop good CWV |
|---|---|---|
| Home pages | 45% | 47% |
| Secondary pages | 56% | 61% |
| Top 1,000 sites by traffic | 51% | 59% |
| Sites ranked 1,001 to 10,000 | 42% | 55% to 57% (all tiers below the top 1,000) |
| Sites ranked 10,001 to 100,000 | 37% | 55% to 57% |
Source: Web Almanac 2025, Performance. The practical read: home pages fail most often, so a test of inner pages alone flatters you, and a test of the home page alone hides template problems. Test the home page, one service page and the contact page separately.
Core Web Vitals pass rates by year, 2020 to 2025
Two series exist, and most stat roundups mix them up. Each year’s Web Almanac reported pass rates using the metrics that were official at the time (FID until 2024). The 2025 edition recalculated history with INP, which is stricter on mobile. Use the INP column when you compare years.
| Year | Mobile, as reported that year | Desktop, as reported that year | Mobile, INP basis (2025 edition) | Desktop, INP basis (2025 edition) |
|---|---|---|---|---|
| 2020 | 24% | 34% | n/a | n/a |
| 2021 | 29% | 41% | 32% | 41% |
| 2022 | 39% | 44% | 31% | 44% |
| 2023 | 36% | 48% | ||
| 2024 | 44% | 55% | ||
| 2025 | 48% | 56% |
Sources: 2020 to 2022 “as reported” columns from the Web Almanac 2022, Performance; INP-basis columns from Figure 7.1 of the Web Almanac 2025, Performance. The 2022 chapter itself predicted the gap: it estimated that with INP instead of FID, mobile would fall from 40% to 31% and desktop from 44% to 43%.
Website speed stats 2022: the full data set
If you need 2022 numbers specifically (for a report, a before/after comparison or an old slide), these come from the 2022 Web Almanac Performance chapter, which uses CrUX field data, and the 2022 CMS chapter, which uses June 2022 data. All figures are the share of websites rated good.
| Metric, 2022 | Mobile good | Desktop good | Mobile change vs 2021 |
|---|---|---|---|
| All Core Web Vitals (LCP, FID, CLS) | 39% | 44% | +10 pts |
| LCP (2.5 s or less) | 51% | 63% | +6 pts |
| CLS (0.1 or less) | 74% | 65% | +12 pts |
| FID (100 ms or less), retired in 2024 | 92% | 100% | +2 pts |
| INP (200 ms or less), experimental in 2022 | 55% (8% poor) | 95% (1% poor) | first year measured |
| FCP (1.8 s or less) | 49% | 64% | +11 pts |
| TTFB (0.8 s or less) | 40% | 52% | +1 pt |
The 2022 diagnostics are where the story is, because they show what was slowing pages down:
- Images were the LCP element on 72% of mobile pages and 82% of desktop pages.
- 18% of mobile pages with an image LCP delayed that image with lazy-loading (9.8% native
loading=lazy, 8.8% custom scripts). WordPress ran on 72% of the pages lazy-loading their LCP image natively. - Only about 0.56% of mobile pages preloaded the LCP image, and 0.03% used
fetchpriority=highon it. - LCP images were 67% JPG, 26% PNG and 4% WebP. Median LCP image size on mobile was 95 KB.
- 72% of mobile pages had at least one image without width and height set, the classic cause of layout shift.
- Median JavaScript shipped to mobile was 462 KB, and median lab Total Blocking Time was 1.7 seconds.
Three years later, per the 2025 edition: fetchpriority=high on the LCP image went from 0.03% to 17.3% of mobile pages, unsized images fell from 72% to 62% of mobile pages, and roughly 16% to 17% of pages still lazy-load their LCP image. That last number has barely moved, which makes it the cheapest fix most sites still have.
Core Web Vitals pass rates by CMS, 2022 vs 2025
Mobile only, because the 2025 CMS chapter reports only mobile. Hosted builders that control the whole stack pass far more often than WordPress, where performance depends on the theme, plugins and host you choose.
| CMS | Mobile good CWV, 2022 | Mobile good CWV, 2025 | Mobile good LCP, 2025 |
|---|---|---|---|
| Duda | 67% | 85% | 94% |
| TYPO3 | not in top 10 | 79% | 89% |
| Wix | 39% | 74% (55% in 2024) | 81% |
| Weebly | not in top 10 | 47% | 52% |
| WordPress | 30% | 45% | 53% |
| Drupal | 50% | not stated (about +4 pts vs 2024) | not stated |
| Squarespace | 34% | not stated (+8 pts vs 2024) | not stated |
| Joomla | 38% | not stated (+7 pts vs 2024) | not stated |
Sources: Web Almanac 2022, CMS (June 2022 data) and Web Almanac 2025, CMS. WordPress powers about 64% of CMS-built sites in the 2025 chapter, so its 45% pass rate drags down the web-wide average more than any other platform. If you are rebuilding, the platform decision is a performance decision, which is the core argument for planning web development and SEO together instead of bolting speed on afterward.
Core Web Vitals thresholds, and the date INP replaced FID
| Metric | What it measures | Good | Needs improvement | Poor |
|---|---|---|---|---|
| LCP | Loading: when the largest element renders | 2,500 ms or less | 2,501 to 4,000 ms | over 4,000 ms |
| INP | Responsiveness: latency of clicks, taps and key presses | 200 ms or less | 201 to 500 ms | over 500 ms |
| CLS | Visual stability: unexpected layout movement | 0.1 or less | above 0.1, up to 0.25 | over 0.25 |
Thresholds from Google’s Defining the Core Web Vitals metrics thresholds; all are judged at the 75th percentile of page loads, split by mobile and desktop, per web.dev.
INP replaced FID on 12 March 2024. Google’s launch post reads: “Interaction to Next Paint is now a stable Core Web Vital metric, replacing First Input Delay.” PageSpeed Insights switched its assessment to INP at the same time, and developers pulling FID from the CrUX or PageSpeed Insights APIs had until 9 September 2024 to migrate. Any stat page still listing FID as a Core Web Vital is working from pre-2024 material.
On rankings, Google’s Search Central documentation (updated 10 December 2025) says it “highly recommend[s] site owners achieve good Core Web Vitals” and that page experience “aligns with what our core ranking systems seek to reward.” It does not call Core Web Vitals a dominant ranking factor, and the metric set on that page is still LCP, INP and CLS.
Speed and conversion studies, traced to the original source
Every page speed roundup repeats the same five studies. Here is what each one actually measured, so you can quote them without overstating. None are controlled experiments: they correlate natural speed variation with behavior.
“53% of mobile visitors leave after 3 seconds”: where it comes from
The figure first appeared in Google’s “The Need for Mobile Speed” post on 8 September 2016. The exact wording is “53% of visits are likely to be abandoned if pages take longer than 3 seconds to load.” The footnote gives the base: aggregated Google Analytics data from 3,700 mobile sites that opted into benchmark sharing, March 2016. The same post put the average mobile site load at 19 seconds on 3G. Akamai’s April 2017 retail report release repeated a 53% figure, which is why it is often credited to both. Honest use: “Google estimated in 2016 that 53% of mobile visits were likely abandoned after 3 seconds.” It is a 3G-era, decade-old estimate, not a current measurement.
Google/SOASTA (2017): bounce probability by load time
Google’s Daniel An published “Find out how you stack up to new industry benchmarks for mobile page speed” in February 2017 and updated it in February 2018. A deep neural network trained on bounce and conversion data, with 90% prediction accuracy, estimated that as load time goes from 1 to 10 seconds, bounce probability rises 123%. The intermediate steps (32% at 3 seconds, 90% at 5, 106% at 6) are credited to “Google/SOASTA Research 2017” in the Milliseconds Make Millions report. The 2018 update analyzed 11 million mobile ad landing pages in 213 countries: full load still averaged about 15 seconds, and 79% of pages were over 1 MB.
Google/Deloitte “Milliseconds Make Millions” (published 2020)
Commissioned by Google, prepared by Deloitte with analytics firm Fifty-Five. 37 brands (15 retail, 10 luxury, 6 travel, 6 lead generation) in Europe and the US, about 30 million mobile sessions over 4 weeks. Speed came from Lighthouse across four metrics (max server latency, first meaningful paint, estimated input latency, observed load), and “0.1 seconds faster” means a combined 0.1 second improvement across all four. The model was a logarithmic regression, so results are correlations at 95% significance.
| Vertical | Finding with a 0.1 s mobile improvement |
|---|---|
| Retail (20.5M sessions) | Conversions +8.4%, average order value +9.2%, product listing to product page progression +3.2% |
| Travel (7.4M sessions) | Conversions +10.1%, average order value +1.9%, home page bounce rate improved 6.5% |
| Luxury (2.1M sessions) | Page views per session +8%, product page to add-to-basket +40.1% |
| Lead generation (505K sessions) | Form step 1 to submission +21.6%, mobile bounce rate improved 8.3%, but overall conversion rates decreased when speed improved |
Two caveats almost nobody repeats. First, desktop was excluded because the authors “found a lot of contradicting parameters.” Second, the lead generation result is mixed, which matters if you sell services rather than products: faster pages moved people through forms, yet the report says conversion rates fell in that vertical.
Akamai State of Online Retail Performance (April 2017)
One month of anonymous data from top online retailers, about 10 billion visits. Headlines from the 19 April 2017 release: a 100 millisecond delay in load time was associated with a 7% lower conversion rate, and a 2 second delay with a 103% higher bounce rate. About half of consumers browsed on smartphones, but only about one in five purchases were completed on them.
Portent site speed and conversion study (April 2022)
Portent’s “Site Speed is (Still) Impacting Your Conversion Rate” (20 April 2022) covers about 100 million page views across 20 sites (6 B2C ecommerce, 14 B2B lead generation) in a 30-day window. It does not say which load metric it used, so treat the seconds as approximate.
| Load time | B2C ecommerce conversion rate | B2B lead gen conversion rate |
|---|---|---|
| 1 second | 3.05% | about 40% |
| 2 seconds | 1.68% | 34% |
| 3 seconds | 1.12% | 29% (where it begins to level off) |
| 4 seconds | 0.67% | not stated |
Portent’s summary: a 1 second site converts 2.5 times better than a 5 second site in ecommerce and 3 times better in B2B. Its own worked example, 1,000 visitors and a $50 product: 1,000 × 3.05% × $50 = $1,525 at 1 second, versus 1,000 × 1.12% × $50 = $560 at 3 seconds, a $965 gap per 1,000 visitors. The B2B percentages are goal conversions (any tracked goal), not qualified leads, so do not compare them with your form fill rate. For realistic lead-gen baselines, use conversion rate benchmarks by industry instead. Portent also found 82% of B2B pages and 86% of B2C pages loaded in 5 seconds or less in 2022.
What each study lets you claim
| If you want to say… | Safe source and wording | Do not say |
|---|---|---|
| Speed affects revenue | “Deloitte found 0.1 s faster mobile pages correlated with 8.4% more retail conversions (2020).” | “Every 0.1 s adds 8% revenue.” |
| Slow pages lose people | “Google estimated bounce probability rises 32% from 1 to 3 seconds (2017).” | “32% of visitors leave after 3 seconds.” |
| Most sites are slow | “48% of mobile sites passed Core Web Vitals in mid-2025 (Web Almanac).” | “Half the web fails Google’s ranking test.” |
| B2B lead gen benefits | “Portent saw goal conversions near 40% at 1 s vs 29% at 3 s (2022, 14 B2B sites).” | “Faster pages double your leads.” |
Core Web Vitals consulting: what you get and what it costs
“Core Web Vitals consulting” covers three very different products: a one-off audit, an audit with fixes shipped, and embedded consultant time working with your developers. Published prices, checked on the vendors’ own pages in November 2026:
| Provider | Offer | Published price | What is included |
|---|---|---|---|
| SpeedVitals | Essential Speed Boost | from $499, one-time | CWV optimization, image compression, caching and CDN, before/after report, 30-day support, 1 to 2 weeks |
| WP Speed Fix (WordPress) | Speed Audit & Plan / Audit + Implementation / Full Optimization Sprint | $795 / $995 / $2,495, one-time | Audit and prioritized plan; plus 1 to 2 implementation sessions; plus a 4 to 6 week sprint with 90 days support |
| GoGoChimp (UK) | Speed Sprint | £1,500 + VAT, fixed | Audit of 5 landing pages, 3 fixes shipped, 30-day conversion comparison, 14 days end to end |
| CoreWebVitals.io (Netherlands) | Consultancy: Steady / Focused / All in | €6,000 / €12,000 / €18,000 per month (1, 2 or 3 days a week); retainer from €1,500 per month | Template-by-template audit, tickets in your system, pairing with your developers, budgets and alerts; first findings within 5 working days |
Worked comparison: a 6-week WordPress sprint at $2,495 buys a fixed scope. Three months of CoreWebVitals.io’s Steady plan costs €18,000, the same as one month of All in, and buys senior time inside your engineering process. The first suits a brochure site with a theme and plugin problem; the second suits a custom stack where fixes need code review.
Core Web Vitals bureau in Belgium (and the Benelux)
We found no Belgian agency publishing Core Web Vitals prices. The nearest published Benelux pricing is CoreWebVitals.io, run by Arjen Karel from Bergen op Zoom in the Netherlands, with a Dutch-language page. Prices are in euros because that is the vendor’s own currency. For a Belgian or Dutch site, the questions below matter more than the agency’s location, because the work is done in your code and measured in CrUX, which covers Belgian Chrome users the same way it covers everyone else.
Seven questions to ask any Core Web Vitals consultant
- Will you work from field data (CrUX, Search Console, RUM) or only Lighthouse scores? Rankings and pass rates use field data.
- Do you guarantee a score? CoreWebVitals.io’s page answers: “Anyone who guarantees a score before looking at your site is guessing.” Some vendors do promise top PageSpeed scores or refunds, so read what the guarantee measures.
- Which templates will you audit? Home pages pass less often (45% mobile) than inner pages (56%), so a home-page-only audit misses the pattern.
- Who ships the fixes: you, my developers, or both?
- How will we verify? PageSpeed Insights field data covers the previous 28 days, and Search Console validation runs a 28-day monitoring session, so field results lag a deployment by about four weeks.
- What do you leave behind: performance budgets, alerts, documentation?
- Will fixes survive the next plugin update or redesign?
If your speed problem is really a platform or redesign question, that is a strategy call before it is a developer ticket, and it is the kind of decision I help firms make in a consultation.
Core Web Vitals fix checklist with pass/fail criteria
Run this per template (home, service page, blog post, contact), on mobile first. Each row names the share of sites that fail it in the 2025 Web Almanac, so you know how common the problem is.
| Check | Pass if | How common the failure is |
|---|---|---|
| Field LCP at p75, mobile | 2.5 s or less in the PageSpeed Insights field data section | 38% of mobile sites fail |
| LCP image is not lazy-loaded | No loading="lazy" or lazy-load script on the hero image | About 16% to 17% of pages lazy-load the LCP image |
| LCP image has priority | fetchpriority="high" on the hero image | Only 17.3% of mobile pages use it |
| LCP image is on your own host | Hero image served from your domain or your CDN, not a third-party host | 16% of mobile pages load it cross-host |
| Server response | TTFB 0.8 s or less at p75 | 56% of mobile sites miss it |
| Images reserve space | Every img has width and height (or CSS aspect-ratio) | 62% of mobile pages have at least one unsized image |
| Field CLS at p75 | 0.1 or less | 28% of desktop sites fail |
| Field INP at p75, mobile | 200 ms or less | 23% of mobile sites fail; median mobile TBT is 1,916 ms |
| Back/forward cache eligible | No unload handlers and no Cache-Control: no-store on HTML | 10% of mobile sites use unload; 23% send no-store |
| Search Console | Core Web Vitals report shows no “Poor” URL groups on mobile | The report groups URLs with a similar experience and gives the whole group one status, so one slow template can flag many pages |
Failure rates from the Web Almanac 2025, Performance. URL grouping and the 28-day validation session are described in Google’s Core Web Vitals report help page. Speed is one layer of a site audit; for where it sits in the order of operations, see the technical SEO checklist for founders.
How to benchmark your site’s speed against industry leaders
You do not need a paid tool to compare yourself with competitors, because CrUX field data is public for any origin with enough traffic. Google’s CrUX documentation says pages must be “publicly discoverable and there must be a large enough number of visitors” to appear.
- Pick 3 to 5 competitors that outrank you for your money keywords, plus one category leader.
- Run each home page and one comparable inner page through PageSpeed Insights. Record the field LCP, INP and CLS at p75 for mobile from the “Discover what your real users are experiencing” section, not the Lighthouse lab score. That section aggregates the previous 28 days.
- Note whether PageSpeed Insights shows URL-level or origin-level field data. When a URL has too little traffic, it falls back to origin data for the whole domain, and the “This URL” button is disabled.
- Use CrUX Vis or the CrUX History API to chart the trend over recent months, so you can see who is improving.
- Compare against the web-wide baseline above: 48% of mobile sites pass all three. If your competitors fail and you pass, speed is a tiebreaker in your favor; if they all pass and you fail, it is a gap to close.
If neither your URL nor your origin has CrUX data, PageSpeed Insights shows “No Data” for the field section and only Lighthouse lab data. In that case, the 320-site small business website speed study is a closer peer benchmark than the global Web Almanac numbers.
Frequently asked questions
What are the latest statistics on website loading speeds?
The newest large-scale field data is the HTTP Archive Web Almanac 2025, based on Chrome UX Report data from July 2025. 48% of mobile websites and 56% of desktop websites passed all three Core Web Vitals. Good LCP was 62% on mobile and 74% on desktop, good INP 77% and 97%, and good CLS 81% and 72%. Only 44% of mobile sites had a good server response time.
What were the website speed stats in 2022?
The 2022 Web Almanac reported 39% of mobile and 44% of desktop websites passing Core Web Vitals, measured with FID. Good LCP was 51% mobile and 63% desktop, and good CLS 74% and 65%. Recalculated with INP, which replaced FID in 2024, the 2022 mobile pass rate is 31%. WordPress sites passed at 30% on mobile, Duda at 67%.
When did INP replace FID in Core Web Vitals?
Interaction to Next Paint replaced First Input Delay as a stable Core Web Vital on 12 March 2024, according to Google's web.dev launch post. PageSpeed Insights switched its assessment to INP at the same time. Developers reading FID from the CrUX or PageSpeed Insights APIs had until 9 September 2024 to migrate. A good INP is 200 milliseconds or less at the 75th percentile.
How much does Core Web Vitals consulting cost?
Published prices range widely by scope. SpeedVitals lists optimization from $499, WP Speed Fix charges $795 for an audit, $995 for audit plus implementation and $2,495 for a 4 to 6 week sprint, and GoGoChimp sells a 14-day sprint for £1,500 plus VAT. Embedded consultancy at CoreWebVitals.io costs €6,000 to €18,000 per month for one to three days a week.
Do Core Web Vitals affect Google rankings in 2025 and 2026?
Google's Search Central page, last updated 10 December 2025, says it highly recommends good Core Web Vitals and that page experience aligns with what its core ranking systems seek to reward. It does not describe them as a dominant factor. The documented metrics are still LCP, INP and CLS, and we found no primary Google announcement of a new Core Web Vital in 2026.
How can I benchmark my site's performance against industry leaders?
Run your pages and your competitors' pages through PageSpeed Insights and compare the field data section, which shows real Chrome user data for the previous 28 days at the 75th percentile. Use CrUX Vis or the CrUX History API for trends over time. Compare home pages with home pages and inner pages with inner pages, and benchmark against the 48% mobile pass rate web-wide.
Is it true that 53% of mobile users leave a site that takes over 3 seconds to load?
It is a real Google figure, but an old estimate. Google's September 2016 post said 53% of visits are likely to be abandoned if pages take longer than 3 seconds, based on Google Analytics data from 3,700 opted-in mobile sites in March 2016, when the average mobile page took 19 seconds to load on 3G. Quote it with the year and the word likely.
More CO Consulting marketing research
- FTC Advertising and Endorsement Enforcement: 12 Compliance Statistics for 2026
- Marketing Cost & CAC Benchmark by Industry 2026: The True Cost Per Acquired Client
- The AI Search Readiness Study 2026: 15 Data Points on How Small-Business Websites Score for AI Citation
- US Digital Advertising Spend: 42 Statistics, Trends, and Data Points for 2026
- Social Media Marketing Statistics 2026: Adoption, Demographics, Time Spent, and Ad Spend
- CMO and Marketing Budget Statistics 2026: Verified Benchmarks From Gartner and The CMO Survey
- Marketing Employment and Salary Statistics 2026: 45 Verified Data Points From the U.S. Bureau of Labor Statistics
More Marketing Research guides
About the author
Christoph Olivier Christoph Olivier is the founder of CO Consulting and a fractional CMO who has managed millions of dollars in ad spend and built a combined audience of over a million followers across social platforms. He publishes original marketing data reports here on CO Consulting.