A website redesign is justified when the site can no longer explain what you built, cannot change at the speed you are learning, or is actively stopping people finding you. Everything else on the usual checklist is a repair, and repairs are cheaper, faster and less risky than a rebuild.
Most redesign advice does not make that distinction, which is why so many founders spend three months and a large invoice solving a problem that was four days of work.
Most of what gets called a redesign is a repair.
The standard list runs: the site looks dated, it is slow, it is not mobile friendly, the branding has moved on. Every one of those is real. None of them is a reason to start again.
A slow site is a performance problem with specific causes, usually images and scripts, and it is fixed by addressing those causes. A site that looks dated is a design refresh, which is a fraction of a rebuild. Neither requires touching your information architecture, your URLs or your content.
The distinction matters because a rebuild carries risks a repair does not. New URLs lose whatever search equity the old ones held. A new stack means new bugs. And the three months it takes are three months you are not writing anything, which for an early company is usually the more expensive loss.
The signs that genuinely justify a rebuild
01 The site cannot explain the product any more.
You changed what you sell, or who you sell it to, and the structure fights you every time you try to say the new thing. Not the copy, the structure. If the navigation itself is wrong, editing pages will not fix it.
02 You cannot change it without help.
Every copy change is a ticket for somebody else. At an early stage the site should change weekly, because your understanding of the market changes weekly. A site only a developer can edit will simply stop being edited, and a stale site is worse than an ugly one.
03 The platform is blocking discovery.
If pages render only after JavaScript runs, search engines that execute JavaScript will cope and the crawlers behind AI assistants will not. That is a platform problem rather than a content one, and no amount of writing gets around it. We have written up how that failure actually looks.
04 The site cannot hold the proof you now have.
Early on you have no case studies and the site is built accordingly. Once you have real numbers and named customers, a structure with nowhere to put them is costing you the most persuasive thing you own.
The signs that do not justify one
It looks like a template. Most early sites should, and a template is the correct choice for longer than founders expect. Buyers do not reject software because the marketing site used a common layout.
A competitor relaunched. Their constraints are not yours and you cannot see their numbers.
Traffic is flat. That is almost always a content and distribution problem wearing a design costume. Rebuilding a site nobody visits produces a nicer site nobody visits.
Someone senior does not like the colours. Write down what the site is failing to do, in outcomes. If nothing can be written down, there is no brief.
Fix these first, in this order
Before commissioning anything, work through the repairs. Most sites need three or four of them and none needs a rebuild.
Performance. Google's Core Web Vitals set the thresholds: Largest Contentful Paint at 2.5 seconds or less, Interaction to Next Paint at 200 milliseconds or less, and Cumulative Layout Shift at 0.1 or less, measured at the 75th percentile of real loads. Oversized images are the usual cause and converting them is an afternoon.
The first screen. If a visitor cannot tell what the product does and who it is for within one screen, that is a copy problem. It is also the highest-leverage thing on the site and it costs nothing to change.
Proof placement. Move the numbers and the named customers up. Most sites bury them below three sections of feature description.
The pages that convert. A landing page built for a company with no proof yet is a different exercise from a general homepage refresh, and usually a better use of the same week.
Before you brief anyone
Write down what the site is failing to do, in outcomes rather than adjectives. "Buyers cannot tell we serve enterprise" is a brief. "It feels old" is not.
Then decide what must survive. Your URLs, your content, your analytics history and your search positions all have value, and a redesign that discards them starts you from zero on the one asset that took longest to build.
If you do rebuild, keep the URLs. If a URL genuinely has to change, that is a redirect plan, and the plan is part of the project rather than something to sort out afterwards.