Google judges how fast and stable a page feels with three Core Web Vitals: Largest Contentful Paint (LCP) should happen within 2.5 seconds, Interaction to Next Paint (INP) should stay at 200 milliseconds or less, and Cumulative Layout Shift (CLS) should stay at 0.1 or less. A page passes when at least 75% of real visits meet all three. For most small business websites the usual culprits are oversized photos, too many add-on scripts and slow hosting, and you can check your own site for free with Google’s PageSpeed Insights in about ten minutes.
What are Core Web Vitals, in plain words?
They are three measurements Google takes from real Chrome users visiting your site. Each one answers a question a customer feels but never says out loud.
| Metric | The customer’s question | Good | Poor |
|---|---|---|---|
| LCP, Largest Contentful Paint | How long until I can see the main thing on this page? | 2.5 seconds or less | Over 4 seconds |
| INP, Interaction to Next Paint | When I tap a button, does the page react straight away? | 200 milliseconds or less | Over 500 milliseconds |
| CLS, Cumulative Layout Shift | Does the page jump around while I am trying to read or tap? | 0.1 or less | Over 0.25 |
LCP is usually the big photo or headline at the top of your home page. INP covers every tap and click during a visit, such as opening the menu or pressing Call. CLS is a score, not a time: it adds up how far things move unexpectedly, like a banner that loads late and pushes your phone number down just as someone goes to tap it.
INP is the newest of the three. It replaced an older measure, First Input Delay, as a Core Web Vital in March 2024, so guides written before then may still talk about FID.
Do Core Web Vitals affect Google rankings?
Yes, but less than many sales pitches claim. Google says Core Web Vitals are used by its ranking systems, and it recommends good scores. It also says there is no single page experience signal, and that Search always tries to show the most relevant content even when the page experience is poor.
So speed works like a tie-breaker. If your plumbing page is the best answer for a search, a slow load will not wipe it out. If five local plumbers all have similar, useful pages, the fast one has an edge. Content and relevance come first; speed protects what the content earns.
💡 Speed will not rescue a page that says nothing useful. But on a phone, a slow page loses the customer before Google’s ranking even matters.
The bigger cost is human. A visitor on a phone, standing outside or between jobs, does not wait. If the page is blank for five seconds, they press back and call the next result. You never see that lost enquiry in any report.
How do I test my website speed in 10 minutes?
You need no tools other than a browser. Follow these steps with your home page first, then repeat for your most important service page.
- 1Minute 1: open pagespeed.web.dev, paste your full web address and press Analyze.
- 2Minute 2: stay on the Mobile tab. Most local customers arrive on a phone, and the mobile test is the stricter one.
- 3Minutes 3–4: read the top section, titled Discover what your real users are experiencing. If it says Passed, at least 75% of real visits over the last 28 days met all three targets.
- 4Minute 5: if that section says there is not enough data, that is normal for a small or new site. Skip to the lab section below it.
- 5Minutes 6–7: in the lab section, note the LCP figure and the Total Blocking Time figure. The lab test cannot measure INP, so Total Blocking Time is its stand-in for how sluggish taps may feel.
- 6Minute 8: scroll to Diagnostics. Look for lines about image sizes, unused JavaScript and server response time. These name the actual causes.
- 7Minute 9: open your site on your own phone using mobile data, not office Wi-Fi. Count the seconds until the main photo and phone number appear.
- 8Minute 10: write down three numbers (LCP, CLS, Total Blocking Time) and the date. Run the test again next month and compare.
Do not panic over the big coloured score out of 100 at the top of the lab section. It is a summary of one simulated visit on a mid-range phone and a throttled connection, and it moves a few points between runs. The real-user section is what Google’s ranking systems look at.
Why does PageSpeed Insights say there is no real-user data for my site?
Real-user figures come from the Chrome User Experience Report, which only includes pages with enough visits from Chrome users who share usage statistics. Google’s documentation explains that a page may have no data if it is new or has too few samples; the tool then tries the whole site instead, and if that also falls short, it shows nothing.
For a café with a few hundred visitors a month, this is common. It does not mean Google ignores your speed. It means you should rely on the lab test and your own phone test, and fix what they show.
What slows down a small business website?
Small sites rarely have complicated code problems. They have the same handful of habits, repeated. Here they are in the order we most often see them, with who can usually fix each one.
| Cause | Which metric it hurts | Who can fix it |
|---|---|---|
| Photos uploaded straight from a phone or camera, often 4–8 MB each | LCP | Owner, by resizing before upload, or the provider |
| A full-screen video or sliding banner at the top of the home page | LCP and INP | Owner decision; provider removes or replaces it |
| Chat widgets, pop-ups, review carousels and tracking tags all added over time | INP and LCP | Provider, after the owner says which ones are actually used |
| Images and adverts without a reserved space, so text jumps when they load | CLS | Provider |
| Late cookie banners or offer bars that push the page down | CLS | Provider |
| Custom fonts loaded from several places | LCP and CLS | Provider |
| Cheap shared hosting with a slow first response | LCP | Whoever holds the hosting account |
| Heavy drag-and-drop page code from some DIY website builders | LCP and INP | Hard to fix without changing platform |
The last row matters if you built the site yourself. On some DIY website builders the page code is produced by the platform and you cannot trim it, so there is a floor below which your scores will not go, however careful you are with photos.
Worked example: a florist’s home page that loads in 6 seconds
Picture a single-shop florist. The home page opens with a sliding banner of five bouquet photos, each taken on a phone and uploaded at full size. Below it sits a review carousel, a chat bubble, an Instagram feed and a newsletter pop-up that appears after three seconds. The cookie banner loads last and shoves everything down.
A mobile lab test on a page like this would typically show LCP well past the 2.5-second line, because the browser must fetch a multi-megabyte photo before the first slide appears. The pop-up and banner would push CLS up. The feed, chat and carousel scripts would raise Total Blocking Time. These are illustrative, not measured results, but every cause listed is one you can find in Diagnostics.
The fixes, in order of effort:
- 1Replace the slider with one well-chosen photo, resized to the width it is shown at and saved in a modern format.
- 2Remove the Instagram feed and link to the profile instead. A feed loads dozens of images nobody asked for.
- 3Keep either the chat bubble or a WhatsApp button, not both, and load it after the page is visible.
- 4Drop the timed pop-up. Put the newsletter sign-up in the footer.
- 5Give the cookie banner a fixed place at the bottom of the screen so it overlays instead of pushing content.
- 6Check the hosting response time. If the server alone takes over a second, move host.
None of this changes what the florist sells or how the site looks in a screenshot. It changes how fast a customer on a bus sees the flowers and the Order button.
Is a PageSpeed score of 100 necessary?
No. Chasing 100 can waste money. What matters is passing the three thresholds for real users, and a page can pass with a lab score in the 80s. Be wary of anyone selling a speed package that promises a perfect score; ask instead which of the three metrics will improve, by how much, and how you will see it in the real-user data.
There is also a trade-off with useful features. A map, a booking widget or a video tour may cost a little speed and still be worth it if customers use them. Keep what earns its place. Load it lower on the page or on click, so it does not hold up the first view.
Does website speed matter for AI assistants and AI Overviews?
Google says there are no extra technical requirements for its AI features beyond a page being indexed and eligible to show a snippet, and it lists ordinary good practice such as keeping important content as text. Speed is not named as a special AI factor. The honest link is indirect: a page that loads properly, with its key facts in readable text rather than inside images or slow scripts, is easier for every kind of crawler and visitor to use.
If you want the structural side of that, our guide on how AI assistants read your website covers crawlers, robots.txt and structured data.
What should I ask a web provider about speed?
- 1Will you resize and compress the photos I send, or do I have to?
- 2Which third-party scripts will the site load on day one, and why each one?
- 3Where is the site hosted, and roughly how fast does the server respond?
- 4Will you run a mobile PageSpeed Insights test before launch and send me the result?
- 5If a later change, such as a new widget, slows the site down, who notices and who fixes it?
- 6Is the page code yours to trim, or generated by a platform you cannot change?
With Adyft Sites, the owner does not upload anything raw. You send photos and details on WhatsApp, we prepare the images and build the pages to be fast on phones, and hosting and security are part of the monthly fee. If a test shows a problem later, message us on WhatsApp and we will look at it. Starter is $1,499 plus $69 a month for five pages, live in under seven days.
If you enjoy tuning things yourself and already have a quick site, you may not need any of that; the self-test above is enough to keep an eye on it each month.
Questions about website speed and Core Web Vitals
What are good Core Web Vitals scores?
Largest Contentful Paint within 2.5 seconds, Interaction to Next Paint of 200 milliseconds or less, and Cumulative Layout Shift of 0.1 or less. Google assesses them at the 75th percentile of real visits, split by mobile and desktop.
How fast should a small business website load?
Aim for the main photo or headline to appear within 2.5 seconds on a phone using mobile data. That is Google’s good threshold for LCP and a fair test of whether a customer will wait.
Is the PageSpeed Insights score out of 100 a ranking factor?
Not directly. The score summarises one simulated lab visit. Google’s ranking systems use Core Web Vitals from real Chrome users, which appear in the top section of the report when there is enough data.
Why is my site fast on Wi-Fi but slow for customers?
Office Wi-Fi and a new phone hide problems that a mid-range phone on mobile data exposes. Test on your phone with Wi-Fi turned off, and use the Mobile tab in PageSpeed Insights.
What is the quickest speed fix for most small sites?
Resizing photos. Images taken on a phone are often several megabytes each, and the top image largely decides your LCP. Shrinking them to the size actually shown is usually the biggest single gain.
Will a faster website get me more customers?
It removes a reason to leave, which helps, but it does not create demand on its own. Clear services, prices, area and contact details still do most of the work; speed makes sure people stay long enough to read them.
Sources: Google Search Central, Understanding Core Web Vitals — https://developers.google.com/search/docs/appearance/core-web-vitals · Google Search Central, Page experience — https://developers.google.com/search/docs/appearance/page-experience · web.dev, Web Vitals — https://web.dev/articles/vitals · web.dev, Interaction to Next Paint becomes a Core Web Vital — https://web.dev/blog/inp-cwv-march-12 · Google, About PageSpeed Insights — https://developers.google.com/speed/docs/insights/v5/about · Google Search Central, AI features and your website — https://developers.google.com/search/docs/appearance/ai-features
A quick website, without the tuning
Send your details and photos on WhatsApp. We write, build, host and keep your site in shape. Starter is $1,499 plus $69 a month.