Speed gets filed under “technical polish” — something to look at after the real work is done. The data says it is the real work.
The cost is measurable
Google’s research with SOASTA found that as a page’s load time goes from one second to three, the probability that a visitor leaves without interacting rises by 32%. From one second to five, it rises by 90%. That visitor never sees your offer, your case studies or your contact form. They are simply gone.
The other direction is just as blunt. In Milliseconds Make Millions, a study Google ran with Deloitte across 37 brands, a 0.1-second improvement in mobile load time was associated with an 8.4% lift in retail conversions and a 9.2% lift in average order value. A tenth of a second.
For a small or mid-sized business that is not an abstraction. If your site brings in 40 enquiries a month, the people who bounce on a slow page were prospects you already paid to get there — in ads, in time, in referrals.
What “fast” means now
“Fast” stopped being a stopwatch feeling and became three specific measurements Google calls Core Web Vitals:
- LCP (Largest Contentful Paint) — time until the main content is visible. Good is under 2.5 seconds.
- INP (Interaction to Next Paint) — how quickly the page reacts when someone taps or clicks. Good is under 200 milliseconds. INP replaced the older FID metric in March 2024, and it is stricter.
- CLS (Cumulative Layout Shift) — how much the page jumps around while it loads. Good is under 0.1.
Google measures these on real visits to your site and uses them as a ranking signal. So a slow site is penalised twice: once by the people who leave, and again by the search engine that shows it lower.
Where the seconds hide
Rebuilding sites, we find slow almost always comes from the same short list:
- Images shipped at full size — a 4000-pixel photo poured into a 600-pixel column. The browser downloads all of it.
- A page builder’s overhead — Elementor, Divi and WPBakery add nested markup and their own scripts to every page, whether the page uses them or not.
- Third-party scripts — chat widgets, tag managers, embedded fonts, A/B testing tools. Each is a request to someone else’s server that has to finish before the page settles.
- No caching — every visitor rebuilds the page from scratch instead of being handed a ready-made copy.
- A server far from the visitor — shared hosting in one country serving customers in another.
None of these are exotic. They are the default state of a site that was never tuned.
What we do about it
When we build or rebuild a site, performance is part of the structure, not a pass at the end:
- Images go through a pipeline that generates the right sizes and modern formats automatically, with width and height set so nothing shifts.
- Pages are static wherever they can be — plain HTML served from a CDN close to the visitor.
- Third-party scripts are counted, questioned, and loaded so they never block the page.
- We measure against real Core Web Vitals field data, not a lab score, and we check it on the deployed site rather than the build.
The goal is plain: the page is useful the moment it appears, and it responds the moment it is touched. That is what the numbers above reward.
Want to know what your current site scores? Get in touch and we’ll run it for you.
