Website speed, in plain English
Why a slow page wastes ad money, the four things that make a site slow, what fast feels like to a visitor, and what we do about it on every build.
Every brand we meet has been told its website is slow and almost none of them has been told what that means. The reports are written in developer language, the scores change every time you run them, and the agency that built the site says it is fine. Here is what speed is, in plain words, why it costs you money, and what to do about it.
Why does a slow page waste ad spend?
You pay for the click the moment it happens. The platform does not refund you if the visitor leaves before the page appears, and on a phone, on mobile data, a surprising share of them do. They tapped an ad that promised something, got a white screen, and went back to the feed. The ad looks like it failed. The page failed.
The waste compounds. The platform sees a click that did not become anything and charges you more for the next one. The search engine measures the same slowness and ranks the page lower, so the free traffic shrinks as well. And the visitors who waited arrive irritated, which is not the state you want a buyer in on the first screen.
It is the same bucket we describe in every ad account review: if the ads are doing their job and the page is not, more budget just buys more visits to a leaking bucket.
What makes a site slow?
Four things, and the order below is roughly the order we find them in.
| The cause | How it feels | The fix |
|---|---|---|
| Images too big | A blank frame, then the picture appears in bands | Size every image to the frame it sits in, in a modern format, and load the ones further down the page later |
| Too many scripts | The page is visible but nothing responds to a tap | Remove every tracker, chat widget and plug-in that does not earn its place; load the rest after the page |
| A slow or distant server | A long pause before anything at all appears | Serve the pages from a network near the visitor, cached, so the server is not asked twice for the same thing |
| A layout that moves | You go to tap and the button jumps | Reserve the space for images, adverts and fonts before they arrive, so nothing shifts |
Most slow sites have all four, and the first two account for most of the delay. A page builder with a dozen plug-ins and full-size photographs straight from the camera is the usual starting point, and it can be made quick without a rebuild. It is a day of work, not a project.
The scripts deserve a sentence of their own, because they are the cause owners least expect. Every tool ever added to the site — the chat bubble, the pop-up, the heat map, the tracking tags from three different agencies — left a script behind, and each one runs before the visitor can tap anything. Most sites we audit carry tags that nobody on the current team can account for. Removing them is free, and it is usually the single biggest gain on the page.
What does fast feel like to a visitor?
Nobody measures speed with a stopwatch. They feel it.
- The page is there when the thumb stops. No white screen, no spinner, no logo pulsing while something loads.
- The words come first. A visitor can start reading before the pictures land, and the pictures land where the words said they would.
- Nothing jumps. The button you were about to tap is still there when your finger arrives.
- A tap does something straight away. The menu opens, the image enlarges, the form takes the letter.
- The second page is instant. Once a visitor is in, moving around the site should feel like turning a page, not loading one.
That is the whole standard. A site that does those five things on a mid-range phone on mobile data is fast, whatever a scoring tool says, and a site that fails them is slow, whatever the agency says.
How do you check your own site?
Open it on your phone, on mobile data, not on the office wireless. Open an ad, tap it, and count in your head. Do it on the oldest phone in the building, because that is closer to your customers’ phone than yours is.
Then run the page through Google’s free PageSpeed tool and ignore the score. Read the list underneath it instead: it names the heavy images, the scripts that blocked the page and the things that moved. Most of what it lists maps onto the four causes above. If you have Search Console, open the Core Web Vitals report and look at whether the pages marked poor are growing or shrinking over the months. Growing means somebody keeps adding weight.
What do we do on a build?
Speed is a decision made at the start, not a fix applied at the end, so every build we ship starts with a budget: a maximum weight for a page, enforced by a test that fails when a change goes over it. A site cannot quietly get heavier if the build refuses to ship when it does.
Seyr is an ecommerce storefront we built around the purchase rather than the portfolio shot. The product images are sized to the frames they sit in and served in modern formats. The scripts are the ones a checkout needs and nothing else — no chat widget, no tag that nobody can name the purpose of. The pages are served from near the visitor, and the client runs it without us.
zollerhistory.com is the site of a militaria auction house, with an archive of past lots that a collector will search through for an hour. An archive is the hardest kind of site to keep fast, because every page carries photographs and the pages run into the thousands. We serve them as static pages, built in advance rather than assembled on each visit, so the server has nothing to do but hand them over, and we load the images as the visitor scrolls to them rather than all at once.
On both, the fonts load once and the layout reserves its space, so nothing moves. None of that is clever. It is the four causes, decided against before the first page was designed.
When is speed not the problem?
When the page is fast and still does not sell, the page is saying the wrong thing. A visitor who arrives from an ad promised one thing and finds another leaves at the same speed whether the page took a moment or a second. Speed is the floor: it is what makes the rest of the page possible to judge. It is not the reason anyone buys.
So fix it first, because it is cheap and it is measurable, and then stop talking about it and look at what the page actually says. If the headline does not repeat the promise in the ad, no amount of speed will hold the visitor long enough to find out what you meant.