The usual argument for a fast website is that Google prefers one. That's true, and it's the least interesting reason. Here's the better one: on a phone, on a patchy signal, in a car park, a slow site doesn't read as slow. It reads as broken, or as nobody being home. People don't wait and then judge you. They leave and never know they left.
The good news is that speed is diagnosable. Unlike most things a web studio will tell you about your site, you can measure this one yourself in two minutes and get a number that doesn't care whose opinion it is.
Measure it properly first#
Almost everyone measures wrong, in the same two ways: they test on a desktop, and they test on their own connection. Your site feels fast to you because you're on office broadband, twenty miles from the server, with every image already cached in your browser from the last time you looked. Your customer is none of those things.
How to get a number you can trust
- Use Google's own PageSpeed Insights — it's free and it needs nothing but your address
- Read the mobile tab, not desktop. Most small-business traffic is phones, and the phone score is usually far worse
- Test the page people actually land on, which is often a service page rather than your home page
- If you want to feel it rather than read it, load your site on your phone with wifi off, somewhere with two bars
The three numbers that matter#
Google publishes thresholds for what counts as good. You don't need to understand how they're calculated; you need to know which side of the line you're on.
- Largest Contentful Paint — when the main thing appears
- 2.5s
- Interaction to Next Paint — how fast a tap responds
- 200ms
- Cumulative Layout Shift — how much the page jumps about
- 0.1
The first is the one most people mean by "slow." The second is what makes a site feel cheap — you tap a menu and nothing happens for a beat, so you tap it again. The third is the one that makes people leave angry: the page shifts as an advert or an image loads late, and the thing they were reaching for moves out from under their thumb.
A slow site doesn't read as slow. It reads as broken, or as nobody being home.
The six usual culprits, in the order they hurt#
Slow is almost never one thing. It's six things arriving in the wrong order, and on a typical small-business site they're the same six. Roughly in order of how much weight they add:
Images straight off a camera
The single most common cause, and the easiest to fix. A photo from a phone is often 3–6 MB and gets displayed 600 pixels wide. Resized and saved in a modern format, the same picture is frequently under 100 KB with no visible difference. On a site with a dozen photos, this alone can halve the load.
The page builder's framework
If the site was assembled in a visual builder, every page ships that builder's scripts and styles whether the page uses them or not. It's the price of the tool, it's paid by every visitor, and there's no setting that removes it — which is why speed work on builder sites tends to plateau.
Third-party scripts
The chat widget, the review badge, the booking embed, the pixel from that campaign in 2023 that nobody turned off. Each is a request to somebody else's server, and your page waits on their bad day. This is the category where sites accumulate weight without anyone deciding to.
Fonts
Four weights of two families, fetched from another domain, each blocking text from appearing. Two weights, served from your own site, is usually indistinguishable to a visitor and much faster.
Plugin sprawl
Every plugin that touches the front end adds its own CSS and JavaScript to every page. The slider you stopped using still loads. So does the contact form plugin on pages with no form.
Hosting that's genuinely too cheap
Last on the list on purpose. Hosting is the first thing people blame and rarely the main problem — but under about $10 a month you are sharing a machine with hundreds of other sites, and it does show up on the first byte.
What you can fix this week#
Three of those six are fixable on almost any site, by almost anyone, without touching how it was built. Do them in this order:
| Do this | Effort | What it typically buys |
|---|---|---|
| Resize and compress every image over 200 KB | An afternoon | Often the largest single improvement available |
| Remove third-party scripts nobody uses | An hour, plus the arguing | Faster response to taps, fewer things that can fail |
| Cut the font list to two weights | An hour | Text appears sooner, especially on slow connections |
| Turn off plugins you don't use | An hour, carefully, with a backup | Less weight on every page, less to patch |
When it isn't the images#
If you've done all four and mobile is still poor, you've hit the floor of what tuning can do, and the honest answer is that the problem is structural. That's not a moral failing of the site — it's what happens when speed is something you buy back at the end rather than something the construction gives you for free.
It's also the point at which people get sold a speed package, so it's worth knowing how to read one. Tuning an existing site is honest work and often the right purchase — what it can't do is change what the site is made of, so on a builder site it buys you a step down and then a floor. Ask what the number was before, what it'll be after, and what happens to it the next time a plugin is added. A tuner who can answer all three is worth hiring. One who quotes a target score without seeing the site is quoting the average.
If speed is something you buy at the end, it was never in the build.
A page written from scratch starts fast for an uninteresting reason: there's nothing extra in it. No builder framework, no plugin styles, no theme carrying features you'll never use. It stays fast for a better one — nobody can install anything into it. That's the whole trick, and it's why we don't sell speed as a line item. There'd be nothing to sell you.
How fast is fast enough?#
Not as fast as possible. Past a point, the returns collapse and the work gets expensive — chasing a perfect score is a hobby, not a business decision. What's worth insisting on is that the three numbers above are in the green on a phone, on the pages people actually land on, and that they stay there as the site grows.
That last part is where most sites lose it. A site launches quick and gets slower one decision at a time: an uncompressed hero image here, a new tracking script there, a plugin somebody needed once. Speed isn't a state you reach. It's a thing that decays unless somebody is watching it.
Two related pieces: why your website isn't getting you leads, where slow is one of five causes and rarely the first one to fix, and WordPress vs custom code, which is the longer version of why the second culprit on that list is so hard to shift.