Store speed on the devices your customers actually use
Almost every store we are asked to speed up is fast in the office and slow in the field. The developer machine is a workstation on fibre; the customer is on a mid-range phone with a weak signal and eleven third-party scripts loading before the price appears.
Speed work that matters starts by measuring the second machine.
Where the time actually goes#
| Cause | Typical share | Fix |
|---|---|---|
| Third-party scripts | Largest in most stores | Remove, defer, or self-host what stays |
| Unoptimised images | Large | Modern formats, correct dimensions, lazy-load below fold |
| Render-blocking CSS and fonts | Medium | Inline critical CSS, subset and preload fonts |
| Uncached dynamic pages | Medium | Cache product and category pages properly |
| Server response time | Smaller than assumed | Fix queries only after the above |
The order of work that pays#
- Measure on a real mid-range phone on a throttled connection. Lab scores from a fast machine mislead.
- Inventory every third-party script and remove the ones nobody can justify.
- Fix images: correct size, modern format, explicit dimensions to stop layout shift.
- Cache category and product pages, including for logged-out visitors with a personalised header.
- Only then look at server-side query optimisation.
Layout shift is a conversion problem#
Content that moves after load causes mis-taps, and a mis-tap on a product page is a lost buyer, not a metric. Reserve space for images, banners and anything injected by an app.
Cookie banners and promotional bars are the most common source of shift, and both are entirely within your control.
What speed is worth#
Faster stores convert better, but the honest framing is smaller: speed removes a reason to leave. Do not expect it to fix a page that fails to answer the buyer's questions.
Frequently asked questions
Which metric should I optimise?
Largest contentful paint and layout shift on a mid-range phone. Those track what a buyer experiences.
Are apps really the main cause?
In most hosted stores, yes — front-end apps add scripts to every page.
Does a faster server help most?
Rarely. Server time is usually a small share compared with scripts and images.
ecommerce site speedcore web vitalsstore performanceimage optimisationthird party scripts
