- Published
- Updated
- 10 min read

TL;DR: Google did not start grading whole sites by their worst page in March 2026. Site speed still counts page by page, measured from real visits. On phones, loading (LCP) is the metric most sites fail, not INP. INP trouble comes from JavaScript: plugins, tag managers and widgets. Fix the shared template and its scripts first.
Correction, October 2026: an earlier version of this note said the March 2026 core update introduced site-wide Core Web Vitals aggregation and averaged every page into one domain score. Google's documentation says no such thing. It also said 43% of sites fail INP and that INP is the most failed metric; the published data says otherwise. We have rewritten the note around what Google and the HTTP Archive actually publish.
Site speed is one of the few parts of search that a small business can measure and fix on its own schedule. It also attracts more rumors than almost anything else. This spring the rumor was that one slow page now drags your whole domain down. In this update I want to do something more useful than repeat it: show which site speed problems small business sites actually have, why so many trace back to JavaScript, and where to spend the next few hours.
What did the March 2026 core update change about site speed?
The dates are real. Google's Search Status Dashboard logged the March 2026 core update starting on March 27, 2026, and later posted "The rollout was complete as of April 8, 2026." (Google Search Status Dashboard (opens in a new tab))
That entry says nothing about Core Web Vitals or a new way to score site speed. Google describes core updates as changes that "are broad in nature, and don't target specific sites or individual web pages." (Google Search Central, core updates (opens in a new tab)) The thresholds did not move either: Google still asks for LCP within 2.5 seconds, INP under 200 milliseconds and CLS under 0.1 (Google Search Central (opens in a new tab)).
Where the "worst page" story went wrong
Google answers the page or site question directly: "Our core ranking systems generally evaluate content on a page-specific basis, including when understanding aspects related to page experience. However, we do have some site-wide assessments." (Google Search Central, page experience (opens in a new tab)) For site speed, page by page is the rule. Some site-wide assessment exists, and Google does not describe it. Nothing says your fastest and slowest pages are averaged, or that a share of failing URLs sinks the domain.
We went through that question in detail in Does Google grade your whole site by its slowest page? This note takes the next step: if site speed is measured per page, which pages and which metrics should you look at first?
Which site speed metric do most sites actually fail?
Loading, by a wide margin. For site speed on phones, the 2025 Web Almanac, built on Chrome UX Report data from July 2025, found that on phones "62% of websites have a good LCP," while "77% of websites have good INP" and 81% of mobile pages achieve good CLS (HTTP Archive Web Almanac 2025 (opens in a new tab)). All three together pass on 48% of mobile sites and 56% of desktop sites.
So the old claim that 43% of sites fail INP, and that INP is the most failed metric, does not hold. On a typical phone visit, the hero image and the first screen are where site speed is lost most often.
The three metrics in plain terms
- Largest Contentful Paint (LCP): how long until the biggest thing on the first screen appears. Good is 2.5 seconds or less.
- Interaction to Next Paint (INP): how long the page takes to show a response after a tap, click or key press. Good is 200 milliseconds or less; poor is over 500 milliseconds (web.dev, INP (opens in a new tab)).
- Cumulative Layout Shift (CLS): how much the layout jumps while it loads. Good is 0.1 or less.
Together they are Google's site speed yardstick. Each is judged at the 75th percentile of real visits, separately for mobile and desktop. Our guide to Core Web Vitals and performance optimization covers how those numbers are collected.
Why INP still deserves attention
INP became a Core Web Vital in March 2024, replacing First Input Delay (Google Search Central blog (opens in a new tab)). The difference matters. web.dev explains that "FID only measured the input delay of the first interaction on a page," while INP observes all interactions (web.dev, INP (opens in a new tab)). A page that answers the first tap quickly and then freezes on the booking form now shows up as slow.
INP also differs by page. The Web Almanac found mobile home pages "achieved 80% good INP scores," while secondary pages "declined to 69%" (HTTP Archive Web Almanac 2025 (opens in a new tab)). Its authors point out that secondary pages "often contain more complex interactions like filters, carousels, and form validation." That is the true part of the worst-page story: owners tune the home page and forget the service pages, booking forms and product filters underneath. If your site has a quote builder, a booking calendar or a product filter, INP is the site speed number to watch on those pages.
Why does JavaScript slow down site speed on a phone?
Because the browser has one main thread and your scripts share it. web.dev puts it plainly: "The main thread can only process one task at a time. Any task that takes longer than 50 milliseconds is a long task." (web.dev, optimize long tasks (opens in a new tab)) While a long task runs, a tap waits.
Image compression will not fix that kind of site speed problem. INP is made of the input delay, the time your event code takes to run, and the time to paint the next frame (web.dev, optimize INP (opens in a new tab)). Every part of that is script and rendering work. The original version of this note had that part right, and it is the most useful thing in it.
Plugins, builders and tag managers
Most sites run code they did not write. The Web Almanac found "Around 90%-92% pages use third-parties across different rank groups." (HTTP Archive Web Almanac 2025, third parties (opens in a new tab)) web.dev's guide to third-party JavaScript starts with a warning: "If you've optimized your code, but your site still loads too slowly, it's probably the fault of third-party scripts." (web.dev, third-party JavaScript (opens in a new tab))
The platform matters too, mostly through what gets added on top of it. On mobile, the Web Almanac's CMS chapter reports WordPress sites passing Core Web Vitals at 45%, while more tightly managed hosted platforms score higher. Its explanation: page weight on extensible platforms is driven "less by core and more by the accumulated impact of themes, plugins, page builders, and third-party tools." (HTTP Archive Web Almanac 2025, CMS (opens in a new tab)) The same report estimates around 60% of WordPress sites use a page builder.
None of that means a platform is bad. It means site speed is decided by the stack you assemble, not the logo on the login screen. When I look at a slow small business site, the list usually includes a chat widget, two analytics tags, a heat map tool nobody opens, a review badge, a social feed and a page builder's full script bundle on every page. Lighthouse flags "every JavaScript file with more than 20 kibibytes of unused code" for a reason (Chrome for Developers, Lighthouse (opens in a new tab)).
How to check INP on your own site
You can see a rough version in a few minutes. Open Chrome DevTools and go to the Performance panel. Chrome's documentation says that if you interact with the page, "the Performance panel also captures your local Interaction to Next Paint (INP) and its score" (Chrome for Developers, DevTools (opens in a new tab)). Turn on CPU throttling so your laptop behaves more like a mid-range phone, then open menus, submit forms and use filters.
Treat that local site speed test as a lead, not a verdict. web.dev notes that field data "is essential in identifying slow interactions" (web.dev, find slow interactions (opens in a new tab)). Your real customers, on their real phones, are the number Google uses.
Why fix site speed at the template, not the page?
Because that is how both your site and Google's site speed reporting are organized. Search Console's Core Web Vitals report groups "pages that have a similar user experience," and "The LCP, INP, and CLS status applies to the entire group." A group's status is "its most poorly performing metric." (Search Console Help (opens in a new tab))
That is the real kernel behind the worst-page rumor. A layout shift in your blog template shows up on every post. A heavy script in your header shows up on every page. Fix the template once and the site speed of a whole group can move. On a site with 300 posts on one template, one fix covers 300 URLs. That is a property of templates, not a new Google rule.
A site speed fix order for a small business
- Open Search Console's Core Web Vitals report, mobile first. Write down each poor or needs-improvement group and its worst metric.
- Match groups to templates. Home, service page, location page, blog post, product page. Two or three templates usually explain most of the list.
- If LCP is the problem, start with the hero. web.dev warns: "Never lazy-load your LCP image." (web.dev, optimize LCP (opens in a new tab)) Size it for the phone and make sure it is in the first HTML the browser receives.
- If CLS is the problem, reserve space. "Always include width and height size attributes on your images and video elements." (web.dev, optimize CLS (opens in a new tab)) Do the same for banners and embeds.
- If INP is the problem, list every script. Who added it, what it does, who reads its data. Remove one nobody can explain, then the next. Load the rest only on the pages that need them.
- Wait 28 days. The report covers the last 28 days of visits, so a site speed fix needs a month before you judge it.
How much does site speed matter for ranking?
Site speed counts, and it is not the whole picture. Google says "Core Web Vitals are used by our ranking systems." It also says "Google Search always seeks to show the most relevant content, even if the page experience is sub-par," and that a great page experience can contribute to success where lots of helpful content competes (Google Search Central, page experience (opens in a new tab)).
The old version of this note quoted a Google spokesperson saying Core Web Vitals are not giant ranking factors. We could not find that line on a Google page, so it is out. Google's own documentation makes the same point in its own words, quoted above.
Here is how I put it to owners. If your page is the best answer, poor site speed will not erase it. If several local businesses answer equally well, site speed is one of the things that can separate you. And whatever it does for rankings, a page that freezes while someone taps "Book now" costs you the booking.
The spam update and the core update, a few days apart
One part of the old note was accurate. Google's March 2026 spam update ran from March 24 to March 25, and the core update began two days later (Google Search Status Dashboard, spam update (opens in a new tab)). If rankings moved that week, either one could be the cause.
The two work differently. Google says sites hit by a spam update should review its spam policies, and that after a link spam update "any ranking benefit the links may have previously generated for your site is lost." (Google Search Central, spam updates (opens in a new tab)) No site speed fix brings that back, and no spam fix makes a slow page fast. We dropped the old note's figures on affiliate and finance sites because we could not trace them to a primary source. Our note on reading a Google core update in Search Console covers that side.
Frequently asked questions
Did Google start grading whole websites by their worst page in 2026?
No. Google's documentation says its systems generally evaluate page experience page by page, with some site-wide assessments it does not detail. The March 2026 core update notes do not mention Core Web Vitals.
Which Core Web Vital do most sites fail?
Largest Contentful Paint. In the 2025 Web Almanac, 62% of mobile sites had good LCP, compared with 77% for INP and 81% for CLS.
What is a good INP score?
200 milliseconds or less at the 75th percentile of real visits. Between 200 and 500 milliseconds needs improvement, and over 500 milliseconds is poor.
Do plugins slow down site speed?
They can. Each plugin or widget that loads JavaScript competes for the browser's main thread. The more scripts run on a page, the longer a tap can wait, which shows up in INP.
Will compressing images fix INP?
No. Images affect loading and LCP. INP measures how fast the page responds to input, which depends on JavaScript and rendering work.
How long until site speed fixes show in Search Console?
About a month. The Core Web Vitals report covers the last 28 days of real visits, so give each fix a full cycle before you compare.
Where we fit
Kief Studio is in Shrewsbury, and site speed is part of what our managed website operations covers: we host on dedicated machines, watch uptime, speed and security, and fix problems before visitors see them. When a site has outgrown a template and its pile of plugins, our custom web application work builds around how the business runs and keeps it fast, patched and yours after launch.
On the search side, our SEO, AEO and GEO work writes true pages, structures them and keeps the site fast and crawlable. If a number cannot be sourced, we leave it out, which is why this note now says what Google documents and nothing more.
First conversation is free. No commitment.
Send us your Core Web Vitals report and we will read it with you




