Performance Reference
Core Web Vitals: the lab score is a debugger, not a verdict
Your lab score is 98 and Search Console still says the page is failing.
The short answer
Core Web Vitals are three measurements of what a page felt like to real visitors: LCP for how quickly the main content appears, INP for how promptly it responds to interaction, and CLS for whether the layout stays still.
The number that counts is field data — what actual Chrome users experienced — read at the 75th percentile. A lab score from a single simulated run is a debugging aid, not a verdict, which is exactly why a page can score brilliantly in a test tool and still be marked as failing.
By the HostingFast team · Reviewed 12 August 2026
0
Hand-waving in this entry
100+
Entries wired to each other
Real
Numbers you can reproduce
Free
To read, no gate
The published thresholds are LCP under 2.5 seconds, INP under 200 milliseconds and CLS under 0.1, all judged at the 75th percentile of visits. That percentile is the part people miss: it means three quarters of your visitors must be inside the threshold, so a fast experience for most and a dreadful one for the slowest quarter still fails.
Responsibility splits along a clean seam. Hosting owns the floor, because time to first byte sits underneath LCP and nothing can render before the first byte arrives. Everything above that floor belongs to the page: image discipline, script weight, and space reserved so nothing jumps around while it loads.
Field and lab measure different things
A lab run uses one simulated device on one connection, which makes it repeatable and therefore useful for finding a cause. Field data aggregates real visits over a rolling window across every device and network your audience actually uses, which makes it slow to move and impossible to argue with.
Use them in that order. The field report tells you which pages have a problem and how bad it is; the lab tool tells you why. Trying to chase a lab score on its own is how people spend a fortnight optimising something no visitor was waiting for.
Which of the three hosting can move
LCP, partly. A cached response cuts the server share of it dramatically, and NVMe and OPcache shorten what remains on the uncached path. Beyond that, the largest element on the page — usually a hero image — belongs entirely to the page's own build, and no host can size it for you.
INP and CLS belong to the front end almost completely. INP is a JavaScript problem: long tasks blocking the main thread while somebody taps. CLS is a layout problem: images without dimensions, fonts swapping late, banners inserted above content. A faster server will not touch either of them.
Finding your own numbers
Two places. The Core Web Vitals report in Search Console groups your URLs by status and shows the trend over time. The field data panel in PageSpeed Insights gives you the same measurements for a single URL, provided the page carries enough traffic to register.
Low-traffic pages have no field data at all, which is not a failure — it just means you are working from lab measurements and reasoning, and you should say so when you report on it.
What the score is worth
It is one ranking input among many and it decides close calls rather than results. Relevance still leads. What good vitals genuinely remove is a handicap, and bad ones drag on everything you have, so the sensible ambition is to clear the thresholds and then go back to the content.
Neighbouring entries: TTFB, Cache, Lazy Loading and OPcache. Built-in WebP conversion runs at no extra cost here, which quietly removes a large share of the usual LCP problem before you touch a plugin.
Field data over feelings
Every speed claim in this reference points at a measurement someone can repeat. Where a metric comes from real visitors rather than a test tool, the entry says so, because the two disagree far more often than people expect.
WebP conversion is built in at no extra cost, and LiteSpeed caching sits in the server on every tier.
- Field and lab kept apart
- Published thresholds quoted
- Ownership split by layer
- Written by our support engineers
Why HostingFast
Standard on every plan
Three metrics, three owners
LCP is shared with hosting. INP and CLS are almost entirely the front end's problem.
The percentile explained
Three quarters of visits must clear the threshold, which is why averages flatter and field data does not.
Thresholds quoted
LCP under 2.5 s, INP under 200 ms, CLS under 0.1, all at the 75th percentile.
Lab as a debugger
Field data says which pages fail; the lab run says why. Using them the other way round wastes fortnights.
Where the data lives
The Search Console report for trends, the PageSpeed field panel for one URL at a time.
Honest about ranking weight
It removes a handicap and settles close calls. It does not outrank relevance.
Quick Start
From order to online
- 1
Start in Search Console
The vitals report groups your URLs by status. That tells you which pages have a real problem before you open any test tool.
- 2
Diagnose in the lab
Run the failing URL through a lab tool to find the cause: the LCP element, the long task, the shifting block.
- 3
Change one thing, then wait
Field data moves over a rolling window, so a fix does not appear immediately. Ship it, note the date, and check back.
Built In
Loaded onto every plan
- SSH, Git and Composer on the developer-focused plans
- LiteSpeed caching built into the server rather than patched in by plugin
- NVMe SSD storage on every tier, not just the expensive ones
- Built-in WebP image optimisation at no extra cost
- cPanel — the control panel most of the industry already runs
- Softaculous on board for one-click application installs
- 99.9% uptime as the target, monitored around the clock
- Human support on duty every hour of every day
- Your existing site moved over by our engineers, free of charge
- Spam and virus screening on every mailbox as standard
Frequently Asked
What people ask us most often
Why does my lab score disagree with Search Console?
Because they measure different populations. A lab run is one simulated device on one connection; field data is every real visit over a rolling window, judged at the 75th percentile. Old phones on poor mobile connections drag the field figure down while your test run on a fast desktop connection reports 98. When the two disagree, the field number is the one that counts.
Which vital can better hosting actually improve?
LCP, and only the server-side share of it. Caching removes most of the thinking time, and NVMe with OPcache shortens what is left on an uncached request. The largest element itself — usually a hero image — is sized and delivered by your page, so past a certain point the remaining work is yours. INP and CLS are JavaScript and layout problems that no server touches.
What if a page has no field data at all?
It has too little traffic to register, which is common on deep pages and not a fault. Work from lab measurements and reason about the cause, and be explicit that is what you are doing when you report it. The pages with enough traffic to be measured are also the ones where the ranking effect matters, so prioritise accordingly.
How long after a fix should the scores move?
Field data covers a rolling window of real visits, so expect weeks rather than days, and expect the curve to move gradually rather than step. Note the date you shipped the change, keep the lab measurement as evidence the fix works, and resist the urge to ship three more changes while you wait — you will not be able to tell which one did it.
Keep reading
TTFB (Time To First Byte)
Splitting a response time into its four parts and fixing the one that is costing you.
Lazy Loading
Deferring off-screen images without damaging the one measurement that matters most.
How to Test Your Website Speed
The measuring routine in order, with the mistake most people make on the first run.
Web Hosting
cPanel hosting on NVMe with SSL, migration and a first-year domain included.
PHP Hosting
Pick the PHP version per site, on hardware that is quick where it counts.
Changing hosts? Run through our checklist first.
A straightforward sequence for a switch your visitors never feel: which files move first, how to shift email across without losing a single message, the right moment to repoint DNS, and the two mistakes behind almost all the downtime we get asked to rescue.
Clear the thresholds, then move on.
LiteSpeed caching, NVMe storage and built-in WebP conversion on every tier, with no setup fee at any point.
View Web Hosting plans