A shop that is quick on a phone

Shop speed is discussed as a search ranking issue, which undersells it. On a phone, on a mediocre connection, a slow shop loses people before they have seen a product, and they do not report it as a complaint.

The thresholds, as Google publishes them

Core Web Vitals are three measurements with published good thresholds, assessed at the 75th percentile of real page loads, measured separately for mobile and desktop.

• Largest Contentful Paint, how long until the main content appears: good is 2.5 seconds or less, poor is over 4 seconds.
• Interaction to Next Paint, how quickly the page responds when touched: good is 200 milliseconds or less, poor is over 500 milliseconds.
• Cumulative Layout Shift, how much the page moves about while loading: good is 0.1 or less, poor is 0.25 or more.

Two things about the measurement matter as much as the numbers. It is the 75th percentile, so the slowest quarter of your visitors decide your score, not the average. And the field data comes from real Chrome users over a rolling 28-day window, which is why a fix does not show up immediately and why lab tools and field data disagree.

What is actually slow in a shop

1. 01Images. Almost always the largest thing on the page and almost always larger than they need to be. Serve them at the size they display, in a modern format, and let the browser pick a size for the screen.
2. 02Apps and plugins. Each one adds script that runs on every page. A shop with a dozen installed is carrying a dozen tax collectors, and the ones nobody uses are the easiest win available.
3. 03Themes with carousels, sliders and video headers. They ship slow by design, and the hero video is the single heaviest thing most shops serve.
4. 04Fonts, when several weights load before any text appears.
5. 05Third-party tags: chat widgets, analytics, pixels, review platforms. These run on the main thread and are the usual cause of poor interaction scores.

Fixing layout shift

• Give every image and video explicit width and height, or a CSS aspect ratio, so the space is reserved before the file arrives.
• Reserve space for anything that appears late: banners, cookie notices, promotional bars, advertising slots.
• Do not insert content above something a reader is already looking at.
• Load fonts so that text does not reflow dramatically when the real font arrives.

Measure the way Google does

Lab tools are useful for finding causes and misleading as a score, because a test on a fast machine does not resemble your slowest quarter of visitors. Use field data for whether you have a problem and lab data for what it is. Interaction cannot be measured in a lab run at all, since there is no real interaction to measure, which is why lab tools report a proxy for it instead.

Test on a real mid-range phone on mobile data at least once. Everything on your own site is in your browser's cache, so your experience of your own shop is the one experience nobody else has.

Keep it fast with a budget

Shops do not become slow in one decision. They become slow one app, one font and one tracking tag at a time, each defensible on its own, and the page that was quick at launch is not quick a year later. A written budget is what turns this from drift into a decision.

1. 01Write down a limit for the page weight and the number of third-party scripts on your product and listing templates. The numbers matter less than having them.
2. 02Make adding anything a trade. A new app or tag means removing one, or an explicit decision that this one is worth the cost.
3. 03Measure once a month, on mobile field data, and keep the history. A single reading tells you nothing about direction.
4. 04Re-measure after installing anything, not at the end of the quarter when you can no longer tell which addition did it.
5. 05Check the slowest quarter rather than the average, because that is what the assessment uses and those are real customers.

The honest reason this works is social rather than technical. Nobody argues against a faster shop in the abstract and everybody argues for one more tool, so the limit has to be written down before the argument arrives.