Age is not a reason

“The site is three years old” describes nothing about whether it works. Sites are replaced on that basis constantly, at real cost and real risk to visibility, without anyone stating what problem is being solved.

The tell is a project whose brief is a mood — dated, tired, not how we want to look. Those may be true, and they are not yet a specification. A rebuild justified that way tends to reproduce the same problems in a newer typeface, because nobody identified what the problems were.

Reason 1: the business changed

New services, a different buyer, a repositioning. When the site describes a company you no longer are, that is structural — and it shows up as inconsistent descriptions that models repeat.

Check whether it is genuinely structural first. If the positioning changed but the shape of the site still fits, this may be a copy and navigation project rather than a rebuild — which is weeks rather than months and carries almost none of the launch risk.

Reason 2: it can’t be changed by the people who need to

If a copy update requires a developer, the site will go stale regardless of how it looks. That is a tooling problem worth fixing — who can edit what is a design decision as much as a technical one.

A site you can’t update is already out of date, whatever it looks like.

Reason 3: it doesn’t convert

Measured, not felt: form starts versus completions, which pages precede an enquiry, where people leave. If those numbers are bad, targeted fixes usually beat a rebuild — the numbers worth watching tell you which.

Be careful with this one, because a rebuild is the most expensive way to test a hypothesis about conversion. Fixing the form, rewriting the headline, or adding the pricing information buyers wanted are each a fraction of the cost and produce evidence about what was actually wrong.

Reason 4: the foundations genuinely can’t carry it

A platform that cannot be made fast, cannot be made accessible, or cannot be secured is a real case. So is one whose vendor has stopped supporting it. These are verifiable technical facts rather than preferences, and they justify replacement on their own.

Prefer continuous change

Sites maintained steadily rarely need replacing. The big-bang redesign is largely a consequence of neglect: three years of deferred changes arriving as one expensive project with a launch-day risk.

The alternative is unglamorous and cheap — a quarterly review of what is inaccurate, what is underperforming, and what the business has stopped doing. That habit is what keeps a site from reaching the state where replacing it feels like the only option.

If you do rebuild, protect what worked

Keep the URLs, keep the specific factual paragraphs that were being cited, and run the pre-launch checklist properly. Most post-redesign losses are self-inflicted and preventable.

Before anything is designed, record which pages currently produce enquiries and which of your passages are being quoted in AI answers. Those are the assets — and a redesign that tidies them away is the most common way a site emerges better-looking and commercially worse.