Blog

Does a slow website hurt me on Google?

Your site feels slow on your own phone, and you wonder whether Google holds it against you. Google measures speed in three specific ways, and each one has a published mark for good. This post explains the three, what each one says about your site, and how to find your own figures before you pay anyone to fix them.

What does Google mean by a slow page?

Three things, measured on real visits. Core Web Vitals measure loading with Largest Contentful Paint, responsiveness with Interaction to Next Paint and visual stability with Cumulative Layout Shift (Google Search Central).

Each measure answers a question your visitor has. Has the main thing on the page shown up yet? When I tap, does anything happen? Will the button stay where it is long enough for me to press it?

Does speed decide where you show up?

Here is what Google says: it recommends good Core Web Vitals and says they align with what its core ranking systems seek to reward (Google Search Central). Read that as a reason to get them right, not as a lever that moves you up a set number of places.

There is a plainer reason too. A visitor who taps your phone number and waits, or who taps the wrong link because the page jumped, is a visitor having a bad time on your site, whatever Google does with the figures.

What counts as fast enough?

The web.dev guidance sets a good mark for each measure, taken at the 75th percentile of page loads (web.dev):

  • Largest Contentful Paint within 2.5 seconds. It is the time until the largest image, text block or video in view has rendered, counted from when the visitor first asked for the page (web.dev).
  • Interaction to Next Paint of 200 milliseconds or less. Over 200 and up to 500 needs improvement, and over 500 is poor (web.dev).
  • Cumulative Layout Shift of 0.1 or less. It measures the largest burst of unexpected layout shifts during the life of the page (web.dev).

Why is your site slow to respond to a tap?

Interaction to Next Paint counts only clicks, taps and key presses, not scrolling or hovering, and runs from the moment the visitor acts until the browser next paints a frame (web.dev). Part of that time is a delay before the page's code even starts to respond.

Scripts are one thing that can lengthen that delay. Other work on the main thread, such as scripts loading, parsing and compiling, can lengthen the input delay, and script evaluation while a page loads can create long tasks that hold the browser back from responding (web.dev). Ask whoever built the site which scripts load on every page, and whether each one still earns its place.

Why does the page jump while it loads?

web.dev names these causes of Cumulative Layout Shift: images or videos with unknown dimensions, fonts that render larger or smaller than their fallback, and content added to the page dynamically (web.dev).

Each one has a plain question to put to your web designer. Do the images and videos carry their width and height in the page? Does the text change size when the font arrives? Does a banner or a chat box push the page down after it appears?

How do you find your own figures?

Start with real visitors, as web.dev advises (web.dev). Search Console has a Core Web Vitals report (Google Search Central).

Lab tests can leave out responsiveness. Some tools give no Interaction to Next Paint for a page, because they only watch it load without anyone tapping (web.dev). Our free scan reads real-visitor figures from Google's Chrome UX Report when your site has enough traffic to be in it. When it does not, the scan shows a single lab test, labeled as one, and E2 goes unchecked (how the scan works).

Each at the 75th percentile, E1 fires when Largest Contentful Paint is over 2.5 seconds, E2 when Interaction to Next Paint is over 200 milliseconds and E3 when Cumulative Layout Shift is over 0.1 (how the scan works). Two more checks: E4 fires when homepage time to first byte is over 800 milliseconds, and E6 when there are 3 or more render-blocking scripts in the head.

Fixing it or starting over

Start with fixes to the site you have. Fix Sprint is up to 12 hours of the fixes your scan found, shipped in 10 business days with before and after evidence. Every change waits for your approval before it goes live on your website.

If the site cannot be made fast without rebuilding it, that is web design work, and the same three measures are the test at the end.

Reading the 75th percentile, counted

Here is how the 75th percentile works for one page, using Largest Contentful Paint. The counts are made-up round numbers, so put in your own.

Page loads measured in the period
1,000
Loads where the main content showed within 2.5 seconds
700
Loads slower than 2.5 seconds
300
Share within 2.5 seconds
700 of 1,000: 70%
Loads needed within 2.5 seconds for the 75th percentile to pass
75% of 1,000: 750
Gap to close
750 minus 700: 50 loads

At 70%, the 75th percentile lands on a slow load, so the page does not meet the mark even though most loads were fine. Move 50 of the 300 slow loads under 2.5 seconds and it does.

This is why one fast test on your own phone settles nothing. The mark is set by the slower quarter of page loads, not by your best one.

Round numbers for the arithmetic, not a benchmark and not a forecast. Put in your own.

Questions

Is Interaction to Next Paint the same as First Input Delay?

No. Interaction to Next Paint replaced First Input Delay, which measured only the delay before the first interaction (web.dev). It observes the interactions across a visit and is worked out when the visitor leaves the page.

Why does my speed report show no responsiveness figure?

Some lab tools only watch the page load, with nobody tapping, so they give no Interaction to Next Paint for it (web.dev). You need real-visitor figures, such as the Core Web Vitals report in Search Console (Google Search Central).

Do I need a perfect score on every speed test?

No. The good marks are set at the 75th percentile of page loads (web.dev), so aim to meet them for real visitors. A lab score is one load in one setting, not the 75th percentile of real ones.

Related reading

Posts

Guides

Sources