Speed audit: Shopify's Dawn demo store

Shopify store · theme Dawn 15.0.0 · 3 pages tested on a simulated phone

1. What real visitors experience

Google has no real-visitor data for this store yet. It needs enough visits from Chrome users over 28 days; small or new stores often have none. Until then the lab test below is the only measure.

2. The lab test (the PageSpeed score)

PageSpeed loads your page once at full speed, notes when the first frame was drawn, then simulates a slow phone. Anything that finished downloading before that first frame is counted as needed for it. On Google's test machine the first frame often comes a second or two late, for reasons that are not your store's fault (Shopify's own demo store does it too). Then everything downloaded meanwhile counts against you - which is why the same page can score 90 one minute and 65 the next. The fixes below are about what is downloading, and what blocks the screen, in those first two seconds.

* simulated slow-phone times, as on pagespeed.web.dev. "First frame" is the real, unsimulated moment the first frame was drawn; "late start" means it came more than 0.7 s after the page's HTML had arrived.

Home page · theme-dawn-demo.myshopify.com/

Biggest element on screen: a photo: A woman crouches while holding multiple blue handbags. HTML: 42 KB compressed (244 KB of code), built by the server in 29 ms. Requests: 274 (1758 KB). Finished by 2.3 s (what a late start would count): 1284 KB.

Google's own runs (PageSpeed Insights)

Our lab runs (Lighthouse 13.6.0, simulated slow phone)

What the first frame waits for

Product page · …yshopify.com/products/small-convertible-flex-bag-cappuccino

Biggest element on screen: a photo: div.product-media-container > modal-opener.product__modal-opener > div.product__media > im. HTML: 54 KB compressed (346 KB of code), built by the server in 27 ms. Requests: 270 (1599 KB). Finished by 2.3 s (what a late start would count): 1151 KB.

Google's own runs (PageSpeed Insights)

Our lab runs (Lighthouse 13.6.0, simulated slow phone)

What the first frame waits for

Collection page · theme-dawn-demo.myshopify.com/collections/bags

Biggest element on screen: a photo: Small Naomi. HTML: 49 KB compressed (346 KB of code), built by the server in 26 ms. Requests: 258 (1589 KB). Finished by 2.3 s (what a late start would count): 1189 KB.

Google's own runs (PageSpeed Insights)

Our lab runs (Lighthouse 13.6.0, simulated slow phone)

What the first frame waits for

3. Third-party code (apps, pixels, chat)

"Your call": holding it is possible but affects ad tracking, sign-up pop-ups or when review stars appear, so we only change it with your OK.

4. Redirects

One hop from an alternate address is normal. Test, link and advertise the final address: each hop costs about 1 s in Google's simulation.

5. Fixes, most valuable first

Checked and fine

• The address tested answers directly, with no redirect.
• Phone processor load (Total Blocking Time) is low on every page tested.
• Nothing jumps around while the pages load (layout shift under 0.1).
• No video players load up front.
• The server builds the pages quickly.

6. What we will not charge you for

Tested on a real store's live pages, on Google's own machine, and did not work:

• Marking the page 'render-blocking until the headline' with <link rel="expect" blocking="render"> - Worked in local tests, but made Google's own test start late 4 times out of 4.
• Loading Google Fonts only after the page has loaded - The late first frame (~2.3 s) still came after the page's load event, so nothing changed.
• Adding another 'preconnect' to Shopify's file server - 1 on-time run of 2, no better than without it.
• A tiny always-running animation to 'keep frames coming' - Looked good in one batch of runs, then 0 of 2 and 1 of 2 in fair same-batch comparisons.
• Hiding product and collection pages until the main photo appears - Works for a text headline, not for a photo: Lighthouse still counts every request finished before the photo (score stayed 86 / 81).
• Judging any change on ONE PageSpeed run - The same unchanged page scores ~65 or ~85 depending on whether Google's test starts late. Compare 4+ runs per version, in the same batch.

7. How this was measured

• Real-visitor data: Chrome UX Report (last 28 days), read through Google's PageSpeed Insights.
• Lab: 9 page loads with Lighthouse 13.6.0 (the engine behind pagespeed.web.dev), phone, simulated slow 4G, from Singapore on 2 Nov 2026, 08:45 SGT; plus 10 runs on Google's own machine. These loads may show up as visits in your analytics around that time.
• Plain HTTP checks of each page's HTML and of the redirect hops (no browser, not page views).
• Honest limit: Google's test sometimes starts late whatever the store does, so nobody can promise 90+ on every run. What fixes can promise: real visitors passing Core Web Vitals, and on-time runs in the 90s on a typical Shopify theme.