“The website is slow” can describe several different problems. One page stays blank, another displays quickly but ignores a tap, and another moves its controls while the visitor tries to select them. Compressing images alone may leave the actual customer problem untouched.
Begin with pages people use to evaluate the business or make contact, on representative devices and connections. Establish what the visitor is waiting for at each stage.
Identify the symptom before choosing a remedy
Web Vitals guidance distinguishes main-content loading, interaction responsiveness and visual stability through LCP, INP and CLS respectively. A business owner does not need to memorise the calculations, but should understand that these measures describe different experiences.
| Observed experience | What to investigate | Avoid assuming |
|---|---|---|
| Main content takes a long time to appear | Response timing, loading order and important images | Large images are the only cause |
| The page is visible but taps respond slowly | Browser processing during interaction | A hosting change will fix everything |
| Text or controls move unexpectedly | Space for images, fonts and later content | It is merely a cosmetic problem |
| A form gives no clear outcome | Sending state and the receiving system | A fast page guarantees a fast process |
Record the affected page and the steps leading to the problem. If it appears only after opening a chat widget or filtering information, a test that stops immediately after page load may miss it.
Treat a score as one part of the evidence
Separate simulated testing from observations of real visits. A controlled test helps investigate causes, but cannot represent every customer's device, connection and behaviour.
When reading a report, check whether it describes the individual URL or the whole website, and whether it concerns mobile or desktop use. If there is insufficient real-user data, state that limitation and use representative trials rather than interpreting the absence as proof of a good experience.
Keep a baseline with its testing conditions. Comparing a fast device with a slower one, or different network conditions, can produce an apparent improvement unrelated to the change.
For delayed main content, trace the loading sequence
Establish when the browser receives the initial document, discovers the important image and becomes able to display it. The LCP optimisation guide recommends examining the complete loading path rather than assuming a smaller file alone resolves the delay.
Prepare images for their actual display sizes and inspect quality after compression. Images further down the page may wait until they approach the viewport, but the main image needed on arrival should not be unnecessarily deferred in the same way.
Review video, fonts and third-party embeds for their importance at the start of the visit. Work that is not yet needed can be reprioritised, provided essential content and functionality remain available.
For slow taps, inspect the work happening during interaction
Menus, filters, chat widgets and analytics can make the device process several things together. The INP optimisation guide explains how long-running browser work can delay responses even after the page is visible.
Investigate the interaction that is actually slow. Remove unnecessary work and divide processing that blocks responsiveness. Show an appropriate waiting state when an action depends on another system, but do not report success before its outcome is known.
After changes, try repeated taps, opening navigation while the page is still loading and completing forms on a less powerful device. The first visible frame is not the whole experience.
For moving layouts, reserve the space before content arrives
An image without allocated dimensions or content inserted above the current reading position can push the page around. The CLS optimisation guide describes reserving image space through dimensions or aspect ratios before loading.
Check fonts and embedded content too, particularly where their final appearance changes text height. Moving controls matter because they can cause incorrect selections, not simply because the page looks unsettled.
Prioritise the people and tasks affected
Address problems on important pages that are observed in realistic use. Check that an improvement helps visitors complete their task without making images illegible, breaking forms or removing useful information.
Agree acceptance evidence separately for each symptom and device class. Retain measurements alongside direct task testing, and compare the same conditions after the change. A single favourable run should not be the only evidence used to accept a broad improvement.
If the choice between improving and replacing the website remains uncertain, use the existing-website assessment guide. Review the business-enquiry journey so faster loading supports the work that matters to the business.
Are customers waiting for pages or wondering whether a tap worked?
Talvyn Labs can examine mobile journeys, identify the source of delays and prioritise improvements around both speed and practical usability.
Improve your mobile website experience