Essay · 30 August 2026

What hosting companies get wrong about design

A laptop displays a graph titled "TTR" showing a green line chart tracking performance metrics over time.

Open ten hosting websites in ten tabs and try to tell them apart. Dark interface, glowing server, a headline about speed, three plans in three columns. Close the tabs. The only name you remember is the one you already knew.

Much of our work is for hosting and infrastructure companies, so take this as affection rather than superiority. The people building these products are some of the most technically serious we have worked with. Their websites rarely show it.

The same three numbers

Almost every hosting site opens with the same three numbers. Uptime, response time, and a count of data centres on a map with glowing dots, presented as though the buyer has never seen them before. They saw them in the last tab.

Numbers only tell you something when they differ. Quote the same figure to the same decimal place as everyone else and you have not made a claim, you have paid the entry fee. It is a restaurant advertising that its kitchen is clean.

So the buyer reads around the number instead. Is the documentation written for adults, or for a persona? Do the interface screenshots look considered or cobbled together? Is the site itself quick? Those signals survive when the claims cancel out, which they always do.

The product is hidden until after the sale

Hosting has an awkward shape as a product. Nearly everything that separates a good host from a bad one sits behind the checkout: the control panel, the migration, how support answers at two in the morning, what actually happens on the worst day of your year. None of it is visible to someone deciding whether to hand over their infrastructure. They commit first and find out after, which is a lot to ask.

Which leaves the website carrying the whole job. It is the only part of the product a buyer can inspect. Treat it as a brochure, something marketing owns while engineering builds the real thing, and you have the economics backwards. At the point of decision it is the real thing.

Speed claims stopped working

Fastest. Blazing. Lightning. The category has spent years reaching for the same adjectives and buyers have spent the same years discounting them. Saying you are fast costs nothing, so it proves nothing. Being fast is expensive, which is exactly why it lands.

Technical buyers treat the site as a specimen. Sell performance on a sluggish homepage and the claim and the evidence argue in public. Evidence wins.

A site that is visibly quick does not need the word fastest anywhere on it.

What we would do instead

So show the product before the checkout. Real screenshots, real flows, the actual panel rather than a tasteful abstraction of one. If it is not ready to be seen, that is a product problem in a marketing costume.

Then write for the reader you have got. Hosting buyers are frequently engineers, and engineers reward plain description and punish adjectives. The bar for copy in this category is low enough that clear writing is a cheap advantage, which is a strange thing to be able to say about an industry this good at everything else.

And let the design carry what the copy cannot. Restraint, typography, pace, coherence. They read as competence before anyone has read a word.

None of this is secret, which is what makes the sameness odd. Our theory is that it survives because hosting companies benchmark against each other, and when everyone is looking sideways nobody moves. That is also the opening. In a category this uniform, an ordinary amount of nerve looks like a lot.

We work with hosting and infrastructure companies on exactly this problem. If that is you, here is how we think about it.