Speed guide · November 2026

Why Magento stores are slow on phones

Short answer: we looked at 11 US stores on Magento (Adobe Commerce). In Google's numbers for their real phone shoppers, 6 pass all three of Google's speed checks and 5 fail. 4 of the 5 that fail share one thing: a slow server. The first byte takes 1.4 to 2.7 s to arrive, against 0.4 to 1.0 s on the stores that pass. The front end matters too: 5 of the 7 stores on the lighter Hyvä theme pass, and neither of the 2 with a headless front end does.

What we measured

The home page of 11 US stores running Magento 2 or Adobe Commerce, with two kinds of numbers. Real shoppers: Google's Chrome UX Report for each home page, phone visitors over the 28 days to November 12, 2026 (3 in 4 visits get this or better). Our lab test: one load with Google's Lighthouse on a simulated mid-range phone with slow 4G, in November 2026. That is much slower than most shoppers' phones, but it shows every cause. We read each store's front end from the page itself.

Good marks: main content 2.5 s, taps 200 ms, page jumps 0.1. "Server reply": time to the first byte for real shoppers. "Our lab test": when the main content shows in our test on a mid-range phone on slow 4G. For comparison, see how 88 big US store sites do for real phone shoppers.

It's common across Magento

Across the whole web, real phone visitors get good speed on all three checks on 35% of Magento sites, against 76% of Shopify sites (HTTP Archive's Web Almanac 2025, e-commerce chapter, 50,983 Magento and 567,932 Shopify sites). Shopify tunes speed for all its stores; on Magento, each store's own team or agency has to.

What the slow ones have in common

1. A slow server reply. On 4 Wheel Parts, Sugarfina, Zumiez and DecksDirect, real shoppers wait 1.4 to 2.7 s for the first byte, so most of the 2.5 s allowed for the main content is gone before the page starts. Even the Hyvä stores that fail (4 Wheel Parts and DecksDirect) fail this way. On Magento a slow first byte usually means pages aren't coming from the full-page cache (Varnish or Fastly): cache misses, the cache cleared too often, or pages that can't be cached.
   Check first: "Time to First Byte" on pagespeed.web.dev for your home page and a category page, and the cache hit rate in Varnish or Fastly.
2. Too much JavaScript in the front end. The classic Luma theme loads dozens of small script files through RequireJS and builds parts of the page with Knockout after they arrive. Hyvä drops most of that, and 5 of our 7 Hyvä stores pass. Headless front ends ship their own large script bundles: Zumiez (ScandiPWA) loads its main content in 3.1 s for real shoppers, and Rural King (Next.js) answers taps in 690 ms. In our lab test the phone's processor stayed busy for seconds on 8 of 11 stores.
   Check first: how much JavaScript the home page runs before the first tap works (Lighthouse's "Reduce JavaScript execution time"), and what of it the first screen needs.
3. Page jumps from late boxes. Rural King's page jumps a lot (1.17, good is under 0.1) and Sugarfina's too (0.22). Banners, reviews, chat and consent boxes that arrive late push the page around while people tap.
   Check first: reserve space for banners and widgets; Lighthouse lists the elements that moved.
4. Images bigger than a phone screen. Banners and sliders sized for desktops, and product pictures below the fold loading before anyone scrolls. 5 of 11 home pages weighed 5 MB or more in our test.
   Check first: the size of the main banner on a phone, and whether pictures below the fold load late.

11 stores is a small sample: these are patterns we saw, not proof that one change fixes a store.

Why it matters

Shopify's 2026 study of its stores found conversion about 1.5% lower for every 32 ms slower response to taps (Shopify, "Store speed and conversion: what the data shows", April 2026). In Deloitte's 2020 study for Google (37 brands, 30 million phone sessions), when four speed measures, including how fast the page answers a tap, each improved by 0.1 s, retail sites' conversion rate was 8.4% higher (Deloitte, "Milliseconds Make Millions", 2020). Both are averages across many stores, not a forecast for yours. Google also uses real visitors' speed as a search ranking signal.

Your store