Article in English.
A mobile speed audit checks representative pages on constrained devices and connections. Investigate the loading trace and customer task rather than relying on a desktop score.
Run a practical mobile acceptance check
Use a phone to open a shared service URL, read the offer and complete a test enquiry. Record delays, layout movement and input problems separately. Compare the findings with the mobile and desktop LCP guide before assigning a technical cause.
Choose representative journeys
Select a service enquiry, a product purchase or another important task. Open the exact shared or campaign URL, including redirects. The homepage alone may not reveal a slow product template or a broken form.
Use audience evidence where available. When it is unavailable, document the devices and connection conditions you chose rather than describing them as typical of every customer.
Record a baseline
Keep the page URL, release, viewport, device, network settings and cache state with a trace or screenshot. Separate a controlled lab run from available real-user evidence.
| Review item | Record |
|---|---|
| Initial content | When the main information appears and which element is delayed |
| Interactions | Whether menus, filters and inputs react appropriately |
| Layout stability | Which elements move unexpectedly |
| Completion | Whether the task succeeds and reaches the intended confirmation |
Investigate resources and rendering
Check the image actually requested by the browser, the scripts required on the route and the work that delays content. Responsive layout does not automatically mean an appropriately sized image or lightweight code.
Avoid compressing every image or changing hosting without a finding that supports the work. Use the LCP investigation guide for the loading path.
Verify usability alongside speed
Check readable text, visible labels, appropriate input types, error recovery and comfortable touch controls. A quicker page that hides necessary information or makes payment harder is not a successful improvement.
See the mobile-first design guide for a design review.
Confirm the repair
Repeat the relevant test under comparable conditions and check the whole task. Record whether the suspected bottleneck changed and note variation across runs.
Use LCP monitoring to assign ongoing ownership rather than treating one favourable test as permanent proof.
