ENATIV Logo
ENATIV
Back
© 2026 ENATIV
ProjectsServicesKnowledge hubContact
contact@enativ.pl
  1. Knowledge hub
  2. ›
  3. SEO & analytics
  4. ›
  5. Core Web Vitals

Core Web Vitals

Three Google metrics judged on field data: loading, responsiveness and layout stability of a page.

SEO & analytics

Category
SEO & analytics

In practice

Core Web Vitals are three metrics judged on real visits: LCP, INP and CLS. I start from field data, not from a single Lighthouse score, because field data tells me whether the page fails people. Only then do I look for the cause and fix it in code. I check them on every client site, as standard.

What is Core Web Vitals?

Core Web Vitals are a set of metrics Google uses to describe how a user experiences a page. Each measures a different aspect: LCP the loading of the largest element, INP the response to interactions, CLS layout stability. The set can change over time, so today's thresholds are not fixed for good. I treat them as a direction of work, not a goal in themselves.

The judged data is field data, collected from real users, at the 75th percentile of visits and separately for phones and desktops. A page passes when it meets the thresholds of all three metrics. The same data feeds the Search Console report and PageSpeed Insights, so it shows up in Google tools with no instrumentation of your own.

LCP
2.5 s or less
INP
200 ms or less
CLS
0.1 or less
Percentile
75th of visits

How I approach measuring

I work in a loop: measure, fix the cause, measure again. I do not promise a specific result, because it depends on hosting, traffic and the devices of your users, and I control none of them.

  1. Field data first

    I start from real-visit data, for example the CrUX report, because a lab test shows one device, one network and one location. I use the lab result to look for a cause, not to judge.

  2. Cause, not symptom

    A metric that fails I break into parts: what the largest element is, which script blocks the response, what shifts the layout. I fix one cause at a time, so I know what worked.

  3. Measuring after the change

    Once a fix is in production I wait for field data, because the CrUX report covers a 28-day window and the full effect shows up late. Meanwhile I compare lab results from before the change and after it.

When it fits, and when it doesn't

Any page can be measured, but fixing pays off where the result really changes something.

A good fit when

  • A store or sales page where slow loading and a jumping layout show up in the cart.
  • A site with heavy mobile traffic, where weaker devices and networks make the biggest difference.
  • A page after a redesign whose field data got worse than the previous version.
  • A page with traffic that already has field data, so it is clear where it really hurts.

Not a fit when

  • A new page with no traffic: there is no field data yet, so I care about sound architecture first, not numbers.
  • A page that already passes all three thresholds and has no complaints from users.
  • A page whose problem is the offer or the content, not speed.
  • A goal of "100 points in Lighthouse" with no link to what the user sees.

When the issue is how a search engine reads the content, I start with structured data, that is Schema.org, and with technical SEO. When the app itself is slow, with performance optimization.

Frequently asked questions

A lab test such as Lighthouse runs on one device and one network, from one location, so it gives a repeatable result for hunting causes. Field data comes from real visits and is a distribution, judged at its 75th percentile. They can differ and both are true, because they measure different things.
The one that fails its threshold in field data. If more than one fails, I start with the one whose fix is cheapest relative to what the user gains. I do not chase green where everything already passes.
No. The threshold applies at the 75th percentile, so as many as one visit in four can still be worse than the threshold. A green result also says nothing about the content or about how the page runs on a very weak phone. It signals that there is no big problem here, not that all is well.
It does not decide it, but it helps or hurts. The metrics depend on what the browser receives: the size of images and scripts. Next.js splits code by route, and React does not settle how many scripts you add. Measurement guards the result, not the name of the tool.

On this page

  • In practice
  • What is Core Web Vitals?
  • How I approach measuring
  • When it fits, and when it doesn't
  • Frequently asked questions

Related topics

  • Core Web Vitals
  • Next.js
  • React
  • Structured data (Schema.org)
  • WordPress

Related services

  • SEO / PositioningTechnical SEO audits, on-page optimization, and monthly reporting to grow organic search traffic sustainably.
  • Performance OptimizationMeasured performance optimization - component-mount profiling and Core Web Vitals audits with a documented before/after result, not a vague promise.

Is your page missing the thresholds?

Tell me which page you have and whether there is field data yet. I will say whether the problem is speed or something else, and where I would start. The scope is described on the [[service:performance|performance optimization]] and [[service:seo|SEO]] pages.

Describe your project