Last Updated on September 28, 2026 by Becky Halls
Site speed is the SEO topic everyone nods along to and nobody actually fixes. It sits on the audit, goes amber, and stays amber for eight months while other work jumps the queue. Meanwhile a competitor with a faster, snappier site edges past you on all the results that were close. With Core Web Vitals in 2026 that gap is wider than people think, because only 55.7% of web origins pass all three Core Web Vitals on the January 2026 CrUX data. Nearly half the web is leaving a small ranking edge on the table.
I’m not going to pretend speed is the thing that saves a bad site. It isn’t. But if you and a rival are both relevant and both useful, this is often what decides who wins the click.
Is Core Web Vitals even a ranking factor?
Yes, and I wish the debate would die down. John Mueller put it about as plainly as Google ever does: “It is a ranking factor, and it’s more than a tie-breaker, but it also doesn’t replace relevance.” That last clause is the one to tattoo somewhere. A blazing fast page about the wrong thing still won’t rank. Between two relevant pages though, speed and stability can be the deciding vote.
The three metrics, in plain English
There are three scores and they measure three different flavours of “does this site feel good to use”.
LCP (Largest Contentful Paint) is how fast the main thing on the page shows up. Usually your hero image or headline.
INP (Interaction to Next Paint) is how fast the page responds when you actually do something. Tap a button, open a menu, and see how long before anything happens.
CLS (Cumulative Layout Shift) is whether the page jumps around while it loads, so you go to tap one thing and an ad shoves it sideways.
LCP gets all the attention because it’s the easy one to explain. INP is the one I’d fix first if you’re failing it, and here’s why.
Where most sites break, and where it hurts most
The pass rates tell an interesting story. Roughly 87.1% of origins hit a good INP score, versus about 68% for LCP. So more sites fail LCP than INP overall.
But when a site fails INP, it fails hard, and users feel every millisecond of it. INP replaced First Input Delay back in March 2024, and it’s a much harsher, fairer measure. The Chrome team described the shift like this: “Rather than only measuring the first interaction, INP takes all interactions into account, reporting one of the slowest over the entire lifetime of the page.” It reports one of your slowest moments across the whole visit. That’s the number a frustrated person actually remembers, right before they hit back and click the next result.
If you’re in the failing 13% on INP, it’s usually one or two heavy scripts choking the main thread. A chat widget. A tag manager stuffed with forgotten tags. Some ancient carousel plugin. Which is why I like starting there: the fix is often small, cheap, and dramatic.
How I’d fix Core Web Vitals without a full rebuild
Nobody has budget for a ground-up rebuild, and you rarely need one. Most Core Web Vitals wins come from unglamorous housekeeping.
For INP, audit your third-party scripts first. Open the Performance panel in Chrome DevTools, do the things a real user does, and watch what blocks the main thread. Defer anything non-essential, lazy-load the chat widget until someone scrolls, and question every tag in GTM. Half of them are firing for a tool you stopped using in 2023.
For LCP, serve your hero image in a modern format, size it properly, and preload it. Get off render-blocking CSS and fonts. A slow server response time (TTFB) drags LCP down too, so decent hosting and caching earn their keep.
For CLS, set explicit width and height on images and video, reserve space for ads and embeds before they load, and stop injecting banners above content that’s already on screen.
“When a client tells me their site is slow, nine times out of ten it isn’t the code we wrote. It’s the six tracking scripts marketing bolted on over three years and nobody ever removed. I run a field-data check in Search Console first, because lab scores lie and real users don’t. Then I strip the site back to what it actually needs and add things back one at a time. INP usually sorts itself out by the third thing I delete.” Peter Fox, Head of Dev Team at 3Way Social
Does faster actually help conversions?
This is the argument that gets speed work signed off, so lead with it internally. A 0.1 second improvement in load time lifted retail conversions by 8.4%, according to a Deloitte and Google study. A tenth of a second. The ranking bump is nice, but the revenue bump is what gets the developer time booked in.
So even if you don’t buy Core Web Vitals as a ranking factor, and Mueller’s already told you it is one, the business case stands on its own. Faster sites make more money. That’s been true since long before Google gave the metrics names.
Core Web Vitals in 2026 – FAQ
Are Core Web Vitals in 2026 a strong ranking factor?
They’re a real but modest one. Relevance and content quality matter far more. Where two pages are close, Core Web Vitals can tip the result, and they affect whether people stick around after they land.
What’s a good INP score?
200 milliseconds or under is good, 200 to 500 needs work, and over 500 is poor. Aim for the field data in Search Console, not the lab score in Lighthouse, because they often disagree.
Why did INP replace First Input Delay?
FID only measured the delay on your first interaction, which flattered a lot of sites. INP looks at interactions across the whole visit and reports one of the slowest, so it reflects real experience much better.
Can I fix Core Web Vitals without a developer?
Some of it, yes. Image sizing, a caching plugin and removing dead tracking scripts are all doable on most CMS setups. INP fixes usually need someone who can read the Performance panel, though.
How often should I check Core Web Vitals?
Monthly is plenty for most sites. Check after any big theme, plugin or template change, since those are what quietly break your scores between audits.
Peter Fox is Head of Dev at 3Way.Social, where he spends a suspicious amount of his week deleting other people’s tracking scripts.
