Website monitoring · No. 03
Website speed check
In short
A website speed check shows how fast a page loads and responds to visitors. The main tool is Google PageSpeed Insights: it gives a 0–100 score and the Core Web Vitals — LCP, INP and CLS. Google uses Core Web Vitals in ranking, and according to web.dev, slow websites keep fewer visitors and lose conversions.
Contents
What are website speed and Core Web Vitals?
Website speed isn’t a single number but several measurements: how quickly the main content appears, how quickly the page responds to taps and clicks, and whether it “jumps around” while loading. Google boils them down to three Core Web Vitals and describes them as a set of metrics “that measure real-world user experience” (Google documentation).
| Metric | What it measures | Good | Poor |
|---|---|---|---|
| LCP (Largest Contentful Paint) | When the largest visible element appears — the hero image, a banner or a block of text | 2.5 s or less | over 4.0 s |
| INP (Interaction to Next Paint) | How quickly the page responds to clicks, taps and key presses | 200 ms or less | over 500 ms |
| CLS (Cumulative Layout Shift) | How much the content shifts around while loading | 0.1 or less | over 0.25 |
Anything between “good” and “poor” means the metric needs improvement. The thresholds come from web.dev: LCP, INP, CLS. They’re meant to be judged not by a single visit but by most of your visitors:
“To ensure you’re hitting the recommended target for these metrics for most of your users, a good threshold to measure is the 75th percentile of page loads, segmented across mobile and desktop devices.”
PageSpeed Insights (PSI) is a free Google tool that shows both kinds of data. Lab data is the result of a test page load run by Lighthouse under controlled conditions. Field data is what real visitors experienced, taken from the Chrome User Experience Report (CrUX) over the previous 28 days (PSI documentation).
What does a slow website cost you?
A slow website loses visitors before they even see your offer, and converts worse among those who stick around:
“Slow sites have a negative impact on revenue, and fast sites are shown to increase conversion rates and improve business outcomes […]”
The second risk is search. Google states plainly that it takes these metrics into account:
“Core Web Vitals are used by our ranking systems.”
That said, don’t overestimate their weight. According to the same page, good Core Web Vitals don’t guarantee the top spot, and “Google Search always seeks to show the most relevant content, even if the page experience is sub-par.” Speed is an edge when the content is comparable, not a substitute for content.
The third risk is ads. You pay for every click, and a visitor who closes the page before it loads is a paid click with zero chance of a lead. In Google Ads, speed affects both your conversions and your cost per click. On conversions, Google gives an example: “in retail, we’ve seen that a 1-second delay in mobile can impact mobile conversions by up to 20 percent” (Google Ads Help).
Speed is part of the visitor’s landing page experience, and landing page experience is one of the three components of Quality Score, along with expected clickthrough rate (CTR) and ad relevance (Google Ads Help). Quality Score itself is a diagnostic tool, but what it reflects does show up in what you pay:
“Higher quality ads typically cost less per click than lower quality ads.”
So a slow page costs you twice: you pay more per click and get fewer leads from each one.
Visibility in AI answers
How does a speed check relate to AI agents?
Lighthouse — the tool that powers PageSpeed Insights — now has a separate Agentic Browsing category. It doesn’t measure speed. Instead, it rates how easy your website is for AI agents to work with — programs that open pages, fill out forms and take actions on a person’s behalf:
“The Agentic Browsing category evaluates how well your site is constructed for machine interaction through a set of deterministic audits.”
According to the documentation, the category covers several groups of audits:
- llms.txt — a machine-readable description of your website at the root of the domain. Lighthouse flags a problem if a server error occurs when it tries to fetch the
llms.txtfile; a missing file (a 404 response) is currently marked as not applicable, because the format is optional for now (Lighthouse documentation). - Accessibility for agents — a subset of accessibility audits: labels on buttons and form fields, and the integrity of the accessibility tree. According to the documentation, agents review exactly this tree to identify interactive elements (Lighthouse).
- Layout stability — the same CLS metric: if a button shifts while the page loads, an agent can click the wrong thing, just like a person.
- WebMCP — audits of the tools a website can declare for agents under the proposed WebMCP standard.
In the PageSpeed Insights report, a note under the Agentic Browsing score explains that these checks help make sure a website works well for AI agents and integrates WebMCP correctly, and that the category is still under development and may change (PageSpeed Insights).
The Agentic Browsing score is calculated differently from the other Lighthouse categories:
“Unlike other Lighthouse categories, the Agentic Browsing category does not have a weighted average score from 0 to 100.”
Instead, the report shows the share of readiness checks your website passes — for example, “3/3”: three out of three applicable audits passed. So every failed audit — a problem with llms.txt, button labels, layout shift or WebMCP tools — lowers your agentic readiness score.
A valid llms.txt file at the root of your website shows up in the list of passed audits and adds one more passed check to the score. If the file is missing, the audit lands in the “Not applicable” group, along with the WebMCP audits for tools your website doesn’t have, and doesn’t count toward the score. So if you want to show that your website is ready for AI agents, it’s worth adding the file.
It’s important to understand the limits of this score. Google notes that the Agentic Browsing category and WebMCP support “are experimental and based on proposed standards.” The documentation doesn’t tie this score to Google rankings or to appearing in AI Overviews or other AI answers. Treat it as a gauge of how ready your website is to be used not only by people but also by AI agents.
Why do websites get slow?
Speed rarely drops on its own — it usually gets eaten away by gradual changes:
- Heavy images. A banner or product photo was uploaded at full resolution without compression, and LCP climbs on mobile devices.
- New scripts and pixels. Chat widgets, ad tags and analytics tags load up the browser and hurt INP.
- A slow server. The website has outgrown its hosting plan, caching got switched off, or the database responds slowly — and the page waits a long time for the first byte.
- Theme or plugin updates. A new template pulls in extra fonts, libraries and stylesheets.
- Embeds without reserved space. Banners, ad slots and images without set dimensions push content around and hurt CLS.
How to spot a speed problem
Here are typical situations you can see in PageSpeed Insights or in how visitors behave:
| What you see | Likely cause | What to do |
|---|---|---|
| The mobile score dropped sharply after an update | A new plugin, theme or script | Compare the Diagnostics section before and after, and disable the suspicious script |
| Poor LCP: the main image shows up late | A large uncompressed image or a slow server | Compress the image, and check caching and hosting |
| Poor INP: the page freezes after a tap | Too much JavaScript, including third-party tags | Remove tags you don’t need and defer non-essential scripts |
| Poor CLS: buttons “jump” | Banners or images without set dimensions | Set dimensions and reserve space for ad slots |
| The lab score is good, but field data is poor | Real visitors have slower devices and networks than the test | Go by field data and test on mobile |
How to check website speed yourself
All the main tools are free:
- Open PageSpeed Insights, paste in the page URL and run the analysis. Look at both tabs — mobile and desktop: a score of 90 or above is considered good, 50 to 89 needs improvement, and below 50 is poor (PSI documentation).
- Compare the real-user field data at the top with the lab score below it. If there’s no field data, your website doesn’t get enough visitors for Chrome to collect statistics.
- In the Diagnostics section, find the biggest problems — usually images, third-party scripts and server response time.
- In Google Search Console, open the Core Web Vitals report: it groups your pages into “Good,” “Needs improvement” and “Poor.”
- Re-run the test after every website update: a one-off test won’t tell you when exactly speed started to slip.
- Or use the FullCheckApp “Website speed check” and get the results in a clear visual format.
The lab score can vary from run to run, because, as web.dev puts it, lab data and field data “actually measure very different things” (web.dev). What matters is the trend, not a single number.
If you chose the last option, you’ve already made the right call: the check runs automatically, and you’ll be the first to know about any warning or critical issue.
What FullCheckApp checks
Once a day, at a time you choose, FullCheckApp measures your website speed with Google PageSpeed Insights (Lighthouse) and saves the results in your report:
- Mobile and desktop speed scores — from 0 to 100.
- Other Lighthouse scores — Accessibility, Best Practices, SEO and Agentic Browsing (how ready your website is for AI agents).
- Page preview — what the page looked like on mobile and desktop at the time of the test.
- Lab metrics — LCP, CLS, TBT, FCP, Speed Index (SI) and Time to Interactive (TTI), separately for mobile and desktop.
- Real-user data — LCP, CLS and INP from the Chrome UX Report for the last 28 days, if your website gets enough visitors.
- Trends — a chart of your score over the last 30 days, so you can see when speed started to drop.
- On-demand testing — after a website update, you can click “Measure now” instead of waiting for the daily check.
Report labels: green means OK, yellow means a warning, red means critical; you set the severity separately for each website. You get an email report after every check, and critical issues are also sent to Telegram.
| Other Lighthouse scores | Mobile | Desktop |
|---|---|---|
| Accessibility | 76 | 81 |
| Best Practices | 96 | 100 |
| SEO | 100 | 100 |
| Agentic Browsing | 50 | 50 |
| Metric | Mobile | Desktop |
|---|---|---|
| LCP — Largest Contentful Paint | 6.78 s | 2.25 s |
| CLS — Cumulative Layout Shift | 0.000 | 0.018 |
| TBT — Total Blocking Time | 282 ms | 239 ms |
| FCP — First Contentful Paint | 3.82 s | 1.14 s |
| SI — Speed Index | 8.23 s | 2.27 s |
| TTI — Time to Interactive | 15.6 s | 3.22 s |
| Real users (Chrome UX Report, 28 days) | ||
| LCP | 3.44 s | — |
| CLS | 0.040 | — |
| INP — Interaction to Next Paint | 155 ms | — |
Score over the last 30 daysMobileDesktop
Speed only matters if your website responds in the first place. Home page availability is covered in the article on the website availability check, and your search performance in the article on Google rankings and clicks monitoring.
FAQ
What is a good PageSpeed score?
According to the PageSpeed Insights documentation, a score of 90 or above is considered good, 50 to 89 needs improvement, and below 50 is poor. The mobile score is usually lower because the test simulates a slower device and network. Also look at the Core Web Vitals field data: it shows what real visitors experience.
What’s the difference between LCP, INP and CLS?
LCP shows how quickly the main content of the page appears; good is 2.5 seconds or less. INP shows how quickly the page responds to clicks and taps; good is 200 milliseconds or less. CLS shows how much the content shifts while loading; good is 0.1 or less. All three are judged at the 75th percentile of visits.
Why does my PageSpeed score change every time I test?
The Lighthouse lab test loads the page from scratch every time, and server response time, network conditions and third-party scripts vary from run to run. So the score can swing by a few points. Watch the trend over days and weeks rather than a single test, and compare results under the same conditions.
Does website speed affect Google rankings?
Yes, but it isn’t decisive. Google says Core Web Vitals are used by its ranking systems, but good scores don’t guarantee the top spot, and Search always seeks to show the most relevant content. Speed gives you an edge over pages with similar content, but it doesn’t replace quality content.
What is Agentic Browsing in Lighthouse?
It’s an experimental Lighthouse category that rates how well a website works for AI agents: whether the llms.txt file is available, whether buttons and fields have clear labels, whether the layout is stable, and whether WebMCP tools are declared correctly. The Lighthouse documentation doesn’t link it to Google rankings or AI Overviews.
Sources
- web.dev — Web Vitalsweb.dev/articles/vitals
- web.dev — Largest Contentful Paint (LCP)web.dev/articles/lcp
- web.dev — Interaction to Next Paint (INP)web.dev/articles/inp
- web.dev — Cumulative Layout Shift (CLS)web.dev/articles/cls
- web.dev — Why speed mattersweb.dev/learn/performance/why-speed-matters
- Google Search Central — Understanding page experience in Google Search resultsdevelopers.google.com/search/docs/appearance/page-experience
- PageSpeed Insights — About PageSpeed Insightsdevelopers.google.com/speed/docs/insights/v5/about
- Chrome for Developers — Lighthouse: Agentic Browsingdeveloper.chrome.com/docs/lighthouse/agentic-browsing
- Chrome for Developers — Agentic Browsing scoringdeveloper.chrome.com/docs/lighthouse/agentic-browsing/scoring
- PageSpeed Insightspagespeed.web.dev/
- Chrome for Developers — Lighthouse: llms.txtdeveloper.chrome.com/docs/lighthouse/agentic-browsing/llms-txt
- Google Ads Help — About ad qualitysupport.google.com/google-ads/answer/156066?hl=en
- Google Ads Help — About Quality Score for Search campaignssupport.google.com/google-ads/answer/6167118?hl=en
- Google Ads Help — Evaluate the performance of your landing pagessupport.google.com/google-ads/answer/7543502?hl=en
Check your website
Add your website address — the first check runs right away, and you’ll see the status of every item in this article.
Check for free