Shopify speed audit: measure the product page, not a homepage vanity score

Homepage PSI is the wrong trophy

Shoppers land on ads, collections and PDPs. A 95 on the homepage with a 4.5s LCP on the money page is a conversion bug dressed as performance. A Shopify speed audit should lead with sampled product URLs on mobile.

We use lab signals as a proxy, not as a field-CrUX guarantee. The methodology page states that limit. Do not ship a report that pretends one Lighthouse run is “Core Web Vitals passed”.

Where Shopify actually gets slow

Theme JS and sectioned OS 2.0 apps that hydrate the whole PDP. Review and personalization widgets. Unoptimized hero video. Third-party fonts. Chat that loads before the buy box. None of this is mysterious; it is visible in the network panel and in screenshots of a late LCP element.

• Largest image on the PDP (gallery vs app overlay)
• JS from review / upsell / popup apps
• Render-blocking theme CSS on mobile
• Third-party pixels competing with first paint

What “fix speed” should mean this week

Remove or defer the app that owns the LCP element. Compress the first gallery image. Stop auto-playing video above the fold. Then re-run the same URLs. A theme swap without an app diet repeats the score.

Run the Shopify speed audit, then the store audit if conversion and SEO need to sit next to the perf axis. We will not invent a panel median TTFB until n ≥ 30.