Measure First. Guessing Is Expensive.

Before changing anything, run your slowest page through a speed test on mobile, and run it three times. One test tells you almost nothing; three tell you whether the problem is consistent or the server occasionally stalling. What you are looking for is which of the three numbers is bad, because each points at a different cause and a different fix. Paying someone to "optimise the site" without this step is how people spend money and gain half a second.

The Three Numbers That Matter

• LCP — how long until the main content appears. Should be under 2.5 seconds. Bad LCP usually means images or a slow server
• INP — how quickly the page responds when someone taps. Should be under 200 milliseconds. Bad INP means too much JavaScript
• CLS — how much the layout jumps while loading. Should be under 0.1. Bad CLS means images and ads without reserved space
• Test on mobile, on a normal connection. Your office wi-fi and a desktop browser are not where your customers are
• Test a product or service page, not just the homepage. The homepage is usually the one page somebody already optimised

The Five Causes, In Order of How Often They Are It

• Images. A photo straight from a phone camera is four megabytes and gets displayed at 600 pixels wide. This is the number one cause and the cheapest to fix
• Plugins. Every plugin loads its own scripts and styles on every page, including the ones that do not use it. Twenty plugins is not twenty features — it is twenty tax collectors
• Hosting. Cheap shared hosting means your site waits in line behind hundreds of others. If the server takes over 600ms to respond before anything else happens, no front-end work will save you
• Third-party scripts. Chat widgets, analytics, pixels, review embeds, fonts from external services. Each one is a request to somebody else's server that you do not control
• The build itself. Render-blocking CSS, no caching, no compression, uncontrolled fonts. This is the part that needs a developer rather than a setting

The Fixes That Pay for Themselves First

Ordered by result per hour spent. The first three usually take a page from unbearable to acceptable, and none of them require rebuilding anything.

• Convert images to WebP and serve them at the size they are actually displayed. Frequently a 60-70% reduction in page weight on its own
• Lazy-load everything below the first screen, and set explicit width and height so nothing jumps
• Audit plugins. Deactivate, measure, repeat. Anything you cannot justify out loud comes off
• Turn on caching and compression properly. On decent hosting this is configuration, not development
• Self-host fonts and load only the weights you use. Two weights, not nine
• Defer third-party scripts, and load the chat widget after the page is interactive rather than before
• Move to better hosting if the server response time is still the bottleneck. Do this last — it is the fix people try first and it rarely is the cause on its own

What Slow Actually Costs

Speed is not a vanity score. A visitor who waits four seconds on a phone with one bar of signal simply leaves, and you never see them in your analytics as anything except a bounce. It compounds in two directions: search engines use page experience as a ranking signal, and paid traffic gets more expensive because you are paying for clicks that never reach a loaded page. A slow site quietly raises the price of every channel you use.

When Speed Is Not the Real Problem

Sometimes a site scores well and still feels slow, and that is a design problem wearing a performance costume. A page that loads in 1.8 seconds but requires four clicks to find a phone number feels slower than one that takes three seconds and answers the question immediately. If your numbers are green and people still complain, stop optimising and start watching someone use the site. What they are describing is friction, not latency, and no amount of caching fixes it.

Conclusion

Ask for a measurement before and after, on the same pages, on mobile. Any honest speed work comes with those numbers, and any offer that does not is selling you a feeling. Most business sites can go from poor to good in a day or two of focused work, because the causes are boring and repeatable. The ones that need a rebuild are the ones where the site was assembled from a heavy theme and never had a performance budget at all — and that is a different conversation, with a different price.