New sites usually look better than the ones they replace. That's the easy part, and it's why redesign projects are rarely judged honestly until the numbers come in three months later.
The failure modes below are common, expensive and almost entirely preventable — but only before launch.
1. Throwing away the URLs
This is the big one. Every page that ranked, every link another site pointed at you, every bookmark — all of it hangs off a URL. Change the structure without a redirect map and you don't just lose the ranking, you lose the accumulated credibility behind it.
The fix is unglamorous: export every existing URL before the rebuild, decide where each one points afterwards, and implement permanent redirects at launch. Not the week after.
2. Deleting the pages that quietly worked
Redesigns are usually driven by how the site looks, so the decision about what to keep gets made on aesthetic grounds. An old, ugly page that brings in steady search traffic gets cut because it doesn't fit the new design.
Before deciding anything, pull twelve months of analytics and sort pages by entries and conversions. Anything in the top twenty gets rebuilt, not removed — however dated it looks.
3. Making it slower
New sites often arrive heavier: large hero images, a video background, animation libraries, three tracking scripts. On a good office connection nobody notices. On a phone on mobile data, the visitor leaves before the page renders.
Set a performance budget before design starts and treat it as a requirement, not an optimisation to attempt later.
4. Designing for the homepage
Everyone reviews the homepage. Meanwhile, most visitors from search land on a service or product page, and a good share never see the homepage at all.
Design the pages people actually arrive on with the same care — including a clear next step on each one, because a visitor who lands deep in the site shouldn't have to navigate back to the top to find out how to get in touch.
5. Launching without measurement in place
If tracking is rebuilt at the same time as the site, and events are named differently, you lose the ability to compare before and after. Six months later nobody can say whether the redesign helped.
Keep the old measurement running until the new one is verified, and write down what you expect to improve — enquiries, not pageviews — before launch day.
A redesign should be judged on enquiries three months later, not on how it looks in the launch email. If you're planning one, the audit is the cheapest part and the one that decides everything else.
Have a question this didn't answer?
Ask it directly. We answer questions about scope and budget without a sales call attached.
Start a project