Speed is not a technical detail. It is the first impression your website makes before anyone reads a single word you paid attention to.
Google PageSpeed Insights is the free tool most owners hear about, run once, panic about, and never open again. Usually because it hands back a number in a coloured circle and a wall of language that sounds like it was written for engineers, because it was.
Here is what it actually measures, what matters for a small business, and what you can safely ignore.
What PageSpeed Insights actually is
PageSpeed Insights is a free Google tool. You paste in a page URL, it loads that page the way a visitor would, and it reports how the page performed - plus a list of things that slowed it down.
It reports two kinds of data. Lab data is a simulated load on a controlled connection, which is what produces the score. Field data is real performance from real Chrome users over the previous 28 days, and it only appears once a page has enough traffic. Field data is the truth. Lab data is the diagnosis.
Why speed matters to the business, not just to Google
People leave slow websites. Not dramatically - they just tap back and try the next result, and you never know it happened. A slow site quietly reduces enquiries from traffic you already earned and already paid for.
- A slow first load costs you the visitors who were only mildly interested - which is most of them.
- Slowness is read as amateur. Fair or not, a laggy site makes a good business look small.
- Google uses page experience signals, including Core Web Vitals, as part of ranking. Helpful content still wins, but speed is a tiebreaker you can control.
- Speed compounds with ad spend. Every second of delay makes each click you bought worth less.
The four scores, in plain language
Performance
How quickly the page becomes visible, stable and usable. This is the score that swings the most between tests, because it is a simulation - run the same page three times and you will get three slightly different numbers.
Accessibility
Whether the page can be used by people relying on screen readers, keyboards or higher contrast. This one is worth taking seriously. It is also the score most likely to expose real, fixable problems: missing image descriptions, unlabelled buttons, text too pale to read.
Best Practices
General web hygiene: secure connections, no console errors, correctly sized images, nothing deprecated. Low scores here usually point at a plugin or an old build rather than your content.
SEO
A basic technical check only - is there a title, a description, is the page crawlable, is the text legible on mobile. A 100 here does not mean your SEO is good. It means nothing is obviously broken.
Core Web Vitals, translated
- Largest Contentful Paint (LCP): how long before the main thing - usually your hero image or headline - actually appears. Aim for under 2.5 seconds.
- Interaction to Next Paint (INP): when someone taps a button, how long before the page visibly responds. Aim for under 200 milliseconds.
- Cumulative Layout Shift (CLS): how much the page jumps around while loading. This is the one that makes people tap the wrong thing. Aim for under 0.1.
In owner language: show up fast, respond when tapped, and stop moving. Those three behaviours describe almost every genuinely good website you have used.
Mobile vs desktop: read the mobile tab first
PageSpeed Insights gives you both, and the mobile score is almost always the lower one because it simulates a mid-range phone on a slower connection. That is not the tool being unfair. That is a realistic customer standing outside your competitor's shop.
Most small business traffic is mobile. If you only have the attention for one tab, make it that one.
What to actually fix first
- Images. Oversized photos are the single most common cause of a slow small business site. Resize before uploading and use modern formats.
- Anything that moves. Reserve space for images, banners and embeds so the layout stops shifting.
- Third-party scripts. Every chat widget, pixel, review embed and font service has a cost. Keep the ones that earn their place.
- Fonts. Two weights of one or two families is plenty.
- Accessibility flags. Alt text, button labels and contrast are quick, permanent wins.
What not to obsess over
Do not chase 100. A perfect score on a simulated test is a hobby, not a business outcome. A real site with photography, a booking tool and analytics will not score 100, and it should not have to.
- Score wobble of a few points between runs is normal - it is a simulation, not a stopwatch.
- Warnings about third-party code you deliberately chose are information, not instructions.
- One slow page does not mean the whole site is broken. Test your homepage, your main service page and your contact page, and compare.
How we use it
We treat PageSpeed Insights as a diagnostic tool, not a scoreboard. We run it during a build, read the opportunities list, fix the things that genuinely affect a visitor - image weight, layout shift, script bloat, accessibility flags - and then leave the number alone.
The goal was never a green circle. The goal is a site that loads before someone gives up on you. Fix what a human would notice, ignore what only a test would notice. That's Built Different.