When a website redesign is worth it — and when it isn't

Every few years a website gets rebuilt because it is a few years old. The reasoning is circular and almost everybody uses it, including people who would never accept that logic anywhere else in the business.

The rule of thumb is that a company redesigns every two to three years. It is repeated in pitch decks, in agency blog posts and in board papers. It has no traceable origin. The clearest account comes from a marketing agency that went looking for the source and published the result: "This is a commonly repeated fact in web design circles, but it's tough to find the source for this data."

That is worth sitting with. The most-cited number in the redesign conversation is one that the people repeating it cannot source. It is not wrong, exactly — it is unfounded, which is a different and more awkward thing.

So there is no schedule in this article. What follows instead is what actually changes the answer.

These are the four I see in the projects that turn out to have been worth doing. None of them is about the site's age.

• The business has changed and the site has not. New services, a new market, a different kind of client. The site describes a company that no longer exists, and every meeting starts with a correction.
• Someone checks you before they meet you. A partner, a procurement team, a payments underwriter. If a serious buyer's first impression is a site the business is embarrassed by, that is a cost being paid quietly, every week.
• You cannot change it. The builder cannot export, the agency has stopped replying, or a copy change needs a purchase order. This is the one that compounds: the longer it holds, the more the site drifts from the truth.
• It fails on a phone. Not "looks dated on a phone" — fails. Text that needs pinching, a form that cannot be completed, a menu that does not open.

Two or three of these together is a real case. One of them, on its own, is usually a smaller job than a rebuild.

This is the part that costs people money, because a rebuild is a satisfying thing to buy when the real problem is uncomfortable to name.

• Nobody knows you exist. A redesign does not create demand. If the enquiries are not coming because nothing points at the site, new typography changes nothing at all.
• The offer is wrong. If the pricing, the positioning or the proposition is the problem, a better-looking version of the same words is a more expensive way to make the same argument.
• The site is fine. Sometimes it genuinely is, and the disappointment is coming from somewhere else. That is worth establishing before spending, not after.

If I look at a site and find one of these three, I say so. It is a short conversation and it costs nothing, which is a better outcome than a rebuild that was never going to work.

The genuine danger in a rebuild is not the design. It is that a website is a set of addresses other people have written down, and a rebuild can change all of them at once.

Google keeps two separate documents for this, and the difference between them is the whole point. For a move that changes URLs, its guidance states that "the visibility of your content in Search may fluctuate temporarily during the move", and recommends keeping redirects in place for at least a year. For a move that keeps URLs identical, the guidance is far calmer: the expected effect is a temporary dip in crawl rate, not a ranking event.

The quantified evidence follows the same split, and it is thinner than the confident numbers suggest. The largest published dataset I could find covers domain migrations — not visual redesigns — and reports that on average "it took 523 days for Domain B to show the same level of organic traffic as Domain A", with 17% of migrations not recovering after 1,000 days. Its author is open about the sample being self-selected, which means it describes the migrations people chose to report, not migrations in general.

No study exists on visual redesigns that preserve their URLs. I looked. That absence is the most useful thing in this section: the risk everyone worries about is concentrated in a part of the job that is entirely under your control, and the part everyone actually buys — the design — has no measured downside at all.

Speed is worth doing on its own terms, and the thresholds are public: Google assesses Core Web Vitals at the 75th percentile of real page views, and roughly half the mobile web currently passes all three. What Google will not do is quantify the benefit. Its own page-experience documentation says good scores do not guarantee ranking. Anyone attaching a percentage to that is inventing it.

Everything above points at the same short list, and none of it is exotic. It is just work that has to happen before the new design is signed off rather than after it is live: every existing address inventoried, a redirect written for each one, titles and descriptions carried across with the page they belong to, the sitemap rebuilt from what actually shipped, and someone watching the fortnight after launch, when a missed redirect shows itself.

That is the method rather than the pitch, and it is written up in full on the redesign service page. The two client rebuilds on this site — CityRoamer and Developer Genome — were both done that way.

What this article deliberately does not do is tell you what it costs. That question deserved its own page, and it has one: what a website redesign actually costs.