One second is worth more than one redesign

Site speed gets discussed as a technical matter, which is why it keeps losing budget arguments to things that look like marketing. It is easier to approve a redesign you can see than a set of optimisations you cannot.

So here is the commercial version. Speed is not a technical metric. It is the rate at which you lose people before they have seen anything you paid for.

What the delay actually costs

The mechanism is simple and slightly brutal: every additional second of load time, some proportion of your visitors leave. They are not making a considered judgement about your business. They are on a phone, on mobile data, with three other tabs open, and nothing has happened yet.

The people you lose this way are disproportionately the ones you paid for. Someone who arrived from an ad, or from a search result you worked to rank for, costs you money whether or not the page loads in time. A slow site means paying full price for traffic and receiving a fraction of it.

There is a second cost that shows up later. Google measures real-world loading experience through Core Web Vitals and uses it as a ranking signal. It is not the strongest signal — good content on a slow site still outranks thin content on a fast one — but it is a tiebreaker, and most competitive queries are decided by tiebreakers.

The handful of fixes that recover most of it

The useful thing about performance work is how lopsided it is. A small number of fixes account for most of the gain, and almost none of them require redesigning anything.

Images, which are almost always the problem

On a typical small-business site, images are the majority of the page weight, and most of that is avoidable. Two failures dominate.

Photographs saved as PNG. PNG is for graphics with flat colour and sharp edges — logos, icons, screenshots. Used on a photograph it produces a file many times larger than necessary. We recently found a hero image that was 885 KB as a PNG and 37 KB as a WebP. Identical to look at. That one file was most of the page.

Images served far larger than they display. A 4000-pixel-wide photograph rendered in a 600-pixel column downloads at full size and is then shrunk by the browser. The visitor pays for every pixel they never see.

Converting to WebP and resizing to the largest size actually used typically cuts total image weight by 70 to 90 per cent. It is the single highest-return change available on most sites.

Fonts

Every custom font is a separate download, and if it is coming from a third-party host it also costs a DNS lookup and a new connection before the download even starts. Self-hosting your fonts and preloading the ones used above the fold removes that entirely.

While you are there: two weights of one family is usually enough. Loading six is common and rarely deliberate.

Scripts you forgot you added

This is the one that surprises people. Analytics, a tag manager, a chat widget, a social sharing plugin, a popup tool, an A/B testing script, a heatmap recorder. Each was added for a reason, each reason was sound at the time, and nobody has ever removed one.

Third-party scripts are the worst kind of weight because you do not control them. They can change size without telling you and they load from servers you cannot tune.

Audit what is on your site. Anything you are not actively using the data from should come off.

Page-builder overhead

If your site is built with a page builder, it is probably loading that builder’s entire CSS and JavaScript library on every page — including pages that do not use it. We measured one site where non-builder pages were downloading 98 KB of builder CSS that styled nothing at all.

The one that is not about weight

Layout shift — content jumping around as things load — is measured separately and is mostly a single fix: give every image explicit width and height attributes. The browser then reserves the right amount of space before the image arrives, and nothing moves.

It takes minutes and it fixes the most irritating thing about slow pages, which is tapping a button that moves as you tap it.

How to know if this applies to you

Run your site through PageSpeed Insights and look at the mobile score, not the desktop one. Desktop scores flatter almost every site.

Then look at the specific opportunities it lists rather than the number. The score is a summary; the opportunities are the work. If “properly size images” and “serve images in next-gen formats” are near the top, you have the common problem and the common fix.

Why this beats a redesign

A redesign changes what your site looks like to the people who see it — and the page most people want to rebuild is rarely the one losing the money. Performance work changes how many people see it at all. If you are choosing between the two on a limited budget, the second has a floor under it — a faster site is better regardless of whether the new design lands.

And unlike a redesign, you can measure it honestly. The number was X, the work happened, the number is Y. Few marketing investments give you that.