Short answer: Redesign your website only when data shows it fails your customers and small fixes no longer help. NN/g advises incremental changes in most cases, because drastic changes are jarring for users and risky for the business. When a full redesign is the answer, set measurable goals first, test with real users and keep what already works.
Redesign or small changes?
NN/g’s advice is to base any redesign on user data, with clear goals and measures of success. Make incremental changes to reach them, and only redesign completely if the data tells you to.
Ask why you want to redesign. NN/g notes that the unspoken reason behind many projects is that the owners are bored with the site, even though customers are not.
Customers see far less of your site than you do. Most companies are lucky if customers visit once a month. NN/g says even loyal customers spend less than five hours a year on a site.
Familiar designs help them: people like a design when they can find the features they need right away. Customers balk at change even when the new design is better.
When a full redesign makes sense
NN/g lists a few good reasons for a major overhaul. They go beyond looks: the problems are glaring and can no longer be fixed one by one.
| Reason | What you see | How to confirm it |
|---|---|---|
| Small fixes stopped helping | Many rounds of changes bring little or no gain | Compare results before and after your last changes |
| The technology blocks key changes | The site can’t support what customers need, such as working well on phones | List the changes you couldn’t make, and why |
| The structure is a tangle | Years of patches, and people can’t find what they need | Watch a few customers try your main tasks |
| Very low conversions across the site | High exit and bounce rates and few page views that no single fix improves | Check your analytics, calls and form entries |
| Competitors serve customers much better | You lose customers to sites that support their needs better | Try the same tasks on your site and theirs |
Reasons from NN/g’s article on radical redesign or incremental change, read November 4, 2026; the last column is our advice.
How to plan a website redesign
- Write down the problems to solve and the results you want, in numbers.
- Test your current site with a few customers before you change anything.
- Keep what works: the pages, words and paths people already use.
- Fix content, structure and how things work before you change the style.
- Test the new design with users, fix what they trip on and test again.
- Keep your page addresses, or redirect each old one to its new page.
- Launch, then measure against the goals you wrote in the first step.
NN/g’s early research on iterative design found that, in four case studies, each round of testing and redesign improved usability by 38% at the median. Our website migration guide covers the redirects, and our article on what a website costs covers the budget. If someone else will build it, our guide to hiring a web designer lists what to ask and what to keep in your name.
If you test two versions of a page live, Google asks you not to cloak them. If the test uses several URLs, send visitors to each one with a temporary 302 redirect, and give the variations a rel="canonical" link to the original. Remove the test once you have your answer.
How we change our own site
We rebuilt our site in October 2026, and since then we have changed it one step at a time instead of redesigning it again. Each change is measured or checked before and after.
On October 5, 2026, we turned on a page cache. Across seven rounds of tests, the median server response of our English home page fell from 987 to 37 milliseconds.
The visible changes came the same way. Case studies reached the home page on October 5 and a blog and guides section on October 17. A new tagline followed on October 18, and Miami went into the home page title on November 1.
Before you start a redesign
- You can name the problems and the results you want.
- You have tested the current site with customers.
- Small fixes have been tried and measured.
- You know which pages and paths already work.
- Every old page address has a place in the new site.
- You know how you will measure the result after launch.
RememberRedesign for your customers, not out of boredom. Change what the data says is broken, keep what works and measure the result.
If your site needs more than small fixes, our web design service can rebuild it with you.
Quick answers about website redesigns
How often should you redesign a website?
There is no set schedule. NN/g advises redesigning only when data shows a need, not because the site feels old. Customers see far less of your site than you do: even loyal ones spend less than five hours on it a year.
Will a website redesign hurt my SEO?
It can if page addresses change without redirects or pages that rank disappear. Keep your URLs where you can, redirect the rest one by one, and if you test versions live, follow Google’s advice: temporary 302 redirects and rel="canonical" on the variations.
Why do customers complain after a redesign?
NN/g says users dislike change and will complain whenever things move. That doesn’t mean the new design is worse: if its usability is better, people eventually grow to like it.
Should you test a new design before launching it?
Yes. In four early case studies NN/g analyzed, each round of testing and fixing improved usability by 38% at the median. Testing the current site first, even for a day or two, shows what to fix.
Is a website refresh the same as a redesign?
Not quite. NN/g describes some redesigns as a purely visual reskinning, with new styles and layouts. Others go deeper and fix the structure, content or usability, which is where most problems usually start.
Sources
Read on November 4, 2026. NN/g: Radical Redesign or Incremental Change?, Fresh vs. Familiar: How Aggressively to Redesign and Iterative Design of User Interfaces. Google Search Central: Minimize A/B testing impact in Google Search. Our own case: our page cache test of October 5, 2026, and the dates of our home page changes in our project history.