Page speed isn't just a Lighthouse score. It's a conversion lever. We've run the numbers across dozens of Shopify stores: when LCP goes from 4 seconds to under 1.5, mobile conversion often moves 15–25%. When it stays above 3, you're leaving money on the table no matter how good the offer is.
What we're actually measuring
Core Web Vitals are the baseline. LCP (Largest Contentful Paint) is the one that correlates most with "did this feel fast?" — usually your hero image or a big block of content. We aim for under 2.5s on mobile, under 1.8s on desktop. FID (First Input Delay) matters for tap responsiveness; under 100ms is the target. CLS (Cumulative Layout Shift) kills trust when the page jumps around; keep it under 0.1.
Shopify throws a few curveballs. Apps inject scripts. Themes load a lot of CSS. Images are often huge. So we don't chase a perfect 100. We chase "good enough that speed isn't the reason people leave." That's usually a mobile LCP under 2.8s and a Lighthouse performance score in the high 70s or better.
Where the gains actually come from
Image optimisation is the biggest win. WebP, correct sizing, lazy load below the fold. One client was serving 2MB hero images; we got that down to 180KB without a visible quality hit. LCP dropped by 1.2 seconds. Fonts are next — subset them, load them async, and avoid more than two weights. Then scripts: defer or async everything that isn't critical, and audit apps. We've seen a single app add 800ms. Sometimes the right move is to replace it or build a minimal custom solution.
What "good" looks like in 2025
Stores we work with that convert well on mobile typically sit in this range: LCP < 2.5s, FID < 100ms, CLS < 0.1. Lighthouse performance 75+. That's not vanity. It's the band where we stop seeing speed as the main objection. After that, the work is offer, copy, and flow. If you're outside that band, fix speed first. Then optimise the rest.