A redesign is the easiest project in this business to sell. A four-year-old site always looks dated to the person who sees it every day, and there will always be a provider willing to agree. Here is how to tell whether yours genuinely deserves one, and what costs less when it does not.
Key takeaways. A redesign is justified when the structure blocks progress, not when the design has aged. Most problems blamed on a site’s age, slowness, poor conversion, sagging rankings, can be fixed without rebuilding. And the place where redesigns really cost money is not development, it is migration: URLs changed without redirects erase years of search rankings overnight.
Three cases where you should not redesign
- You dislike the site but it converts. Your visitors do not share your fatigue: they are seeing it for the first time. Rebuilding what works is the fastest way to break what was working.
- The problem is the content. Vague copy and missing photos are not cured by a new layout. A redesign will simply display them more prominently.
- You have no measurement. Without knowing where visitors come from and where they drop off, a redesign is a bet. Install analytics, wait two months, then decide.
The seven signals, and what they actually call for
They always get listed together, as if they all justified the same spend. They do not cost the same to fix, and only two of them require starting over.
| Symptom | What it calls for | Order of magnitude |
|---|---|---|
| The site does not adapt to mobile | redesign | full rebuild |
| Abandoned or unmaintained technology | redesign | full rebuild |
| The site is slow | optimisation | a few days |
| The design looks dated | refresh | a few days |
| Content no longer matches the business | rewrite | a few days |
| You cannot edit it yourself | add a CMS | variable |
| Traffic but few enquiries | conversion work | a few days |
The first two rows are structural: a site that does not adapt to small screens, or that rests on an abandoned component, cannot be repaired, it gets replaced. The other five are jobs of a few days each, whose combined cost stays well below a redesign.
Slowness, with the right figure and its date
Everyone repeats that “each second of delay drives away 7% of visitors”, never with a source. The verifiable figure is Google’s: 53% of mobile visits are abandoned when a page takes longer than three seconds to load. Honest caveat: that data is from March 2016, it is old, and I cite it because it is sourced rather than because it is recent.
More to the point, slowness almost never justifies a redesign on its own. The usual causes are uncompressed images, render-blocking fonts and third-party scripts, and all three are fixed without touching the architecture. The detail is in the article on what actually makes a site slow.
Mobile: the global figure and the French one
Mobile leads global traffic, with slightly over half of page views. France splits differently: for August 2026, 52.98% of pages were viewed on desktop, 44.59% on mobile and 2.43% on tablet (StatCounter, France, August 2026, retrieved 1 September 2026).
In other words, a site that does not display properly on a phone loses close to one French visitor in two, not three quarters. That is already enough to decide, and it avoids quoting a figure the client can check and find wrong.
Redesign, refresh or optimisation
Three distinct jobs, often sold under the same word.
- Optimisation keeps everything and fixes what hurts: speed, journeys, copy, technical SEO. It returns the most per euro spent, and it is what gets proposed least.
- A refresh keeps the structure and the addresses, and changes the skin. No ranking risk, since no URL moves.
- A redesign changes the architecture, the technology and often the addresses. It is the only one of the three that exposes your rankings, which is why it needs preparing.
The “every four or five years” rule you read everywhere has no basis. A seven-year-old site on solid foundations, kept up to date, beats a two-year-old site built badly. Age is not a symptom, it is a sales argument.
Where redesigns actually fail
Not in development, which usually goes fine. In migration, and always in the same place: the addresses.
Every page of your current site has an address Google knows, that other sites cite, and that past customers may have saved. If the new version changes those addresses without permanent redirects, all those links land on an error page. Accumulated ranking does not move by itself: it disappears, and rebuilding it takes months.
The check is one question to ask before signing: “will you deliver a redirect map, old address by old address?” If the answer hesitates, the provider has not budgeted for it, and you will pay for that three months later without connecting the two.
The steps of a redesign that goes well
- Survey what exists: pages that receive traffic, queries that bring it, points where visitors leave.
- Set a measurable objective. “More modern” is not one. “Double the quote requests” is.
- Write the scope, separating the essential from the desirable.
- Choose the technology with whoever picks the site up in three years in mind.
- Draw up the redirect map before development, not after. This is the step everyone postpones.
- Approve mockups, build, then test on phone and on desktop, both of which matter in France.
- Watch rankings for three months after launch. A brief dip is normal, a lasting one signals a missing redirect.
The mistakes that cost the most
- Changing addresses without redirects. The most common, the most expensive, and the least visible at the time.
- Copying the old content across. You are paying for a redesign in order to carry over what was not converting.
- Taking the cheapest quote without checking what it excludes. Migration and redirects are the first items removed to bring a price down.
- Skipping the handover. A site you cannot edit will become outdated exactly like the last one.
Frequently asked questions
How often should a site be rebuilt?
There is no schedule. A well-built, maintained site lives seven to ten years without a redesign, through successive refreshes. A badly built one needs rebuilding after two. Structure decides, not the calendar.
Will I lose my search rankings?
Not if addresses are kept or redirected one by one. A dip of a few weeks after launch is normal while Google re-crawls. A dip that lasts beyond two months is not normal and can be diagnosed.
Can a site be rebuilt in stages?
Yes, and it is often the better option when traffic already exists. Rebuild the pages that receive visits and convert first, measure, then continue. Less spectacular than a launch, and far less risky.
How long does a redesign take?
Two to three weeks for a brochure site whose content is ready, four to six for a store. The real timeline almost always depends on copy and photos arriving, rarely on development.
In short
Two symptoms force a redesign: a site that does not adapt to mobile, and technology that is no longer maintained. Everything else is fixed for far less. And if you do rebuild, the one question that protects your investment is the redirect map, asked before signing rather than after launch.
If you are torn between rebuilding and keeping what exists, a website takeover starts from the opposite principle: keep what holds up, fix the rest, with an honest diagnosis before any commitment.
