What actually slows a Shopify store

Measure per origin, not per page

A page-weight number tells you there is a problem and nothing about what to do. The useful measurement is per third party: how much each domain sends, and how much main-thread time each one takes.

That list is usually shorter and more damning than expected, and it turns a vague performance project into a specific decision about a few apps.

Look for scripts from apps you removed, and apps that do nothing

Uninstalling a Shopify app can leave its snippet in the theme. The app is gone, the script still loads, and nothing in Shopify reports it.

Its cousin is the installed app whose output appears nowhere on the storefront, or two apps doing the same job. Each costs a fee and a script on every page. Removing them is pure gain with no functionality anyone is using at stake.

Fix the image the shopper is waiting for

On a product page the largest paint is almost always the product image. Serving it at the size the screen needs, in a modern format, and not lazy-loading it, is most of the available win.

Lazy-loading the above-the-fold image is a common own goal: a blanket setting applied everywhere, delaying the one asset that defines how fast the page feels.

Google’s guidance is that Largest Contentful Paint should happen within 2.5 seconds of the page starting to load, measured at the 75th percentile of real page loads, for a page to count as good.

Fonts that hold the text back

Several web font files, or fonts loaded without a swap setting, can leave text blank until they arrive. On a slow connection that is a page that looks broken. A font-display setting usually fixes it in one line.

Judge it on field data

A lab test on a fast connection reports a number no customer experiences. What matters is the distribution across real visits, which is dominated by mid-range phones on variable networks. Search Console’s Core Web Vitals report and the Chrome UX Report show that field data for most stores.

If a change does not move the field measurement, it did not happen as far as your shoppers are concerned.

What not to bother with

Chasing a perfect lab score is not a commercial activity. The commercial question is whether the page got faster on real phones and whether conversion followed, and the second is only answerable with a test.

What Liftable’s scan reports

The free scan measures from outside, on desktop and mobile: app scripts that delay the first paint of product pages, apps that load scripts with no visible function, apps that overlap, total JavaScript and the number of third-party domains, web fonts, oversized images, console errors and layout shift from upsell apps. Each finding comes with what loaded and how long it took.

It is one crawl, not your shoppers, so it cannot give you field data. And Liftable never uninstalls an app: it tells you which ones appear to do nothing and leaves the decision to you.

Sources

1. Google: Largest Contentful Paint (web.dev). Defines a good Largest Contentful Paint as 2.5 seconds or less, assessed at the 75th percentile of page loads, segmented across mobile and desktop.

Questions