A site can look fast on an office desktop and be genuinely slow on the mid-range phone your visitors actually hold. Google grades mobile field data; conversion happens mostly on mobile; and the differences between the two experiences are not subtle. Here is what makes mobile different and what to do about it.
Why mobile is a different sport
- CPU, not bandwidth: mid-range phones have a fraction of desktop processing power. The same JavaScript that parses in 200ms on desktop takes 1s+ on a throttled mobile chip — this is why INP dies on mobile first. Page weight matters less than script cost.
- Real network variance: office Wi-Fi flatters desktop tests; visitors ride cellular with 100–300ms RTT. Every request round trip costs multiples.
- Render path: smaller screens still parse the same CSS and DOM. Overloaded markup costs mobile relatively more.
Test like a phone
- Lab: WebPageTest with a Moto G4-class profile and 4G throttling — the standard mid-tier mobile simulation. PageSpeed Insights” mobile tab approximates it.
- Field: Search Console’s mobile CWV report is the grading data. If desktop passes and mobile fails, believe mobile.
- Never test on your own flagship phone — it lies the way office Wi-Fi does.
The mobile-first fixes (they transfer, that is the point)
- Cut script cost: the third-party inventory in the scripts guide matters 3–5× more on mobile CPUs. Delay everything that informs nothing above the fold.
- Cache aggressively: network round trips amplify TTFB — cached responses at double-digit TTFB feel instant on cellular, uncached ones crawl. The layers in the caching guide.
- Right-size images per viewport: srcset serving 400px files to phones saves real transfer on slow networks — the pipeline in the image guide.
- Fonts: two families, swap display — every font request is a cellular round trip, per the font guide.
- Trim the DOM: builder pages with 2,500+ nodes parse painfully on mobile chips — the bloat audit in the coverage tools guide.
Mobile-only traps
- Sticky banners and interstitials eat a small screen and add CLS; Google penalizes intrusive interstitials on mobile specifically.
- Touch handlers blocking paint: carousel libraries initializing on load burn mobile CPU before first scroll.
- Desktop-only testing of checkout: mobile checkout abandonment is its own discipline — test the full flow on a throttled device.
The honest loop
Measure mobile field data → fix the top script/weight offender → re-measure monthly. Most sites can reach passing mobile CWV with caching, image discipline, and script diet alone — no redesign. The free audit reports your mobile numbers with the offending assets named.
Mobile CPUs parse JavaScript 3–5× slower and cellular networks add round trips. Script-heavy sites fail on mobile while desktop scores pass — cut script cost, not just page weight. A mid-tier device (Moto G4 class) with 4G throttling — WebPageTest offers it as a standard profile. Your flagship phone is not representative of visitor experience. Yes — mobile field data is the grading source for Core Web Vitals and indexing is mobile-first. Desktop scores are effectively informational. Most transfer: caching, image pipeline, and script diet help both. Mobile amplifies script cost specifically, so prioritize removing JavaScript over shaving bytes.Why is my site fast on desktop but slow on mobile?
What phone profile should I test with?
Does Google rank my site on mobile performance?
Do desktop optimizations help mobile automatically?
