Shopify Speed Score: What It Is and How to Improve It

Shopify computes the speed score by running Lighthouse against three pages on your store: the homepage, the most-visited collection page, and the most-visited product page (based on the last 30 days of pageviews). The Lighthouse Performance score for each page is taken, weighted by traffic share, and averaged into a single number. That number is then ranked against a comparison pool of similar Shopify stores (similar plan, similar industry, similar traffic volume) and converted to a percentile.

A few consequences follow from this.

The score is relative. A 50 means your store is at the median of the comparison pool. A 70 means you are at the 70th percentile. A 90 means you are faster than 90% of similar stores. The score can change without any change to your store, because other stores in the pool got faster or slower.

Only three pages are measured. Cart, checkout, account, search, and every other page type are not in the calculation. A store with a 90 score can still have a 30-score checkout page. This matters because the checkout is the highest-revenue page.

The pool is opaque. Shopify does not publish the comparison pool composition. Two stores at the same revenue can land in different pools and produce different scores from identical Lighthouse runs.

The data is field, not lab. Shopify uses Lighthouse running on real traffic, not synthetic tests. That means the score reflects actual visitor conditions (devices, networks, geography) and updates as those conditions change.

PageSpeed Insights (PSI) and the Shopify speed score look like they should agree. They almost never do. The differences are structural.

Test pool. PSI returns a Lighthouse run on the URL you submit plus the page's CrUX field data. Shopify computes a percentile against other Shopify stores. A 50 on PSI is an absolute Lighthouse number. A 50 on Shopify is a median ranking.

Pages tested. PSI tests the URL you submit. Shopify tests three specific page types weighted by traffic.

Update frequency. PSI runs Lighthouse on demand and reads a 28-day rolling CrUX window. The Shopify score updates every few days from its own data pipeline and can lag real changes by two weeks.

Device and throttling. PSI uses a fixed device (Moto G Power) and network (Slow 4G) profile. Shopify's field data reflects actual visitor devices, which on most stores skews higher because real visitors include desktop and high-end mobile.

Output. PSI returns separate mobile and desktop scores. Shopify returns one composite number.

The practical rule: PSI is what Google sees and is what you should optimize against for SEO. The Shopify score is a useful internal trend line for comparing your store to itself month over month, but it is not a ranking signal. If the two diverge significantly (Shopify score 75, PSI mobile 40), trust PSI for SEO decisions and treat the Shopify number as a relative ranking signal among Shopify peers.

Useful when: you are tracking your store's performance over time against a stable baseline. The score smooths out single-test noise and is a reasonable directional indicator of whether your changes are helping or hurting. It is also useful as a quick health check: any score under 30 is a sign that something is seriously wrong, and any score above 70 means you are in the top tier of Shopify stores.

Misleading when: you are comparing across stores. Two stores in different industries land in different pools and produce different scores from identical performance. It is also misleading when you have just made a major change. The lag means the score may not reflect your improvement for 14 to 28 days. Most importantly, it is misleading when used as an SEO target: passing CrUX thresholds matters for Google; the Shopify score does not.

Common misreading. Merchants see a 50 and conclude their store is slow. A 50 is the median. Half of all Shopify stores are slower. The store may still be passing CrUX and ranking well. Conversely, a store with a 60 Shopify score and a 35 PSI mobile score is failing CrUX even though Shopify shows it as average. Always check both.

The score improves when the three measured pages get faster in Lighthouse. The fixes are the same fixes that improve overall Shopify performance, just prioritized for those three page types.

1. Audit and remove app JavaScript. Apps are the single biggest lever. Each app adds 100 to 500ms of JavaScript execution, and the average store has 15 to 20 apps. Disable them one at a time in the admin, run PSI on the homepage and top product page, and remove or replace the worst offenders. Expect a 10 to 25 point Shopify score gain from app cleanup alone.

2. Migrate to a fast OS 2.0 theme. A theme migration from vintage to Dawn (or another modern theme) is often a 20 to 30 point gain. See the spoke guide on fastest Shopify themes for the ranked list and migration criteria.

3. Preload the LCP image. On the homepage and product page, the LCP image is usually the hero or main product photo. Preload it with `fetchpriority='high'` and serve it through Shopify's `image_tag` filter at the correct size. This single change can move LCP by 600 to 1200ms.

4. Fix CLS sources. Add explicit width and height attributes to every image, reserve space for review widgets and upsell popups, and use `font-display: swap` with size-adjusted fallback fonts. Layout shift contributes to the Lighthouse score even when LCP is fast.

5. Defer non-critical third-party scripts. Chat widgets, review widgets, social pixels, and marketing tools should load after first paint. Most apps offer a deferred-load option in their settings; for the ones that do not, use a custom defer wrapper in `theme.liquid`.

6. Wait for the score to update. After deploying changes, the Shopify score takes 7 to 14 days to reflect them in most cases, occasionally up to 28 days. Run PSI in the meantime to verify the lab and field metrics moved.

If the score has been stuck below 50 after standard fixes, the bottleneck is usually a single heavy app or a theme that cannot be optimized further. Our done-for-you Shopify speed optimization service profiles the exact source of the floor and handles the work to get past it.