The numbers worth knowing
Google publishes three thresholds. Largest Contentful Paint under 2.5 seconds — how long until the main content appears. Interaction to Next Paint under 200 milliseconds — how quickly the page responds when someone taps something. Cumulative Layout Shift under 0.1 — how much the page jumps around while loading.
Those are the targets. They are measured on real visits from real devices, which is the part most people miss. Google's ranking signal uses field data collected from actual Chrome users on your site, not the score you get from running a test on a fast laptop on office fibre.
Your PageSpeed score is not the goal
The number out of 100 is a lab simulation. It is a useful diagnostic — it tells you which things are slow — but it is not what Google uses to rank you, and chasing it into the nineties has a strong tendency to produce work that does not change what visitors experience.
Look at the field data section instead: that is the real measurement, taken from your actual visitors. A site can score 70 in the lab and pass every Core Web Vital in the field, or score 95 and fail. The field data is the one connected to both rankings and revenue.
What actually makes sites slow
Almost always two things: images and third-party scripts. A hero image exported at full camera resolution and scaled down in the browser can be several megabytes doing the work of a couple hundred kilobytes. Multiply that across a gallery and the page is heavy before any code runs.
Third-party scripts are the quieter problem. An analytics tag, a tag manager, a chat widget, a heatmap tool, two ad pixels and a review embed — each one added by a different person for a good reason, none ever removed. Together they can cost seconds, and they run on every page load forever.
Audit what is loading before optimizing anything else. It is common to find tags for tools nobody has logged into in two years, still costing every visitor time.
Does speed really affect SEO?
Yes, but less than the industry implies. Core Web Vitals are a genuine ranking signal and a real tiebreaker between pages of comparable quality. They will not lift a thin page above a thorough one, and no amount of speed substitutes for actually answering the query.
The bigger effect is on conversion. Visitors leave slow pages before they read anything, which means content quality never gets a chance to matter. If you are paying for the click, you paid for a visitor who left before the page rendered. That is usually the argument that justifies the work, not the ranking one.
Know the ceiling before you start
Every platform has a floor you cannot get under. A page builder that ships a large framework, a theme with a dozen plugins, a stack with server rendering on every request — these set a limit on how fast the site can be, no matter how carefully you optimize around them.
Anyone selling speed work should measure first and tell you the realistic ceiling. Sometimes the honest answer is that a rebuild costs less than repeatedly optimizing around a platform that was never going to get there — that is when a website redesign beats another round of speed optimization.