When should you redesign your website?

What a redesign is, and why the decision is hard

A redesign changes how a site looks, how it is organised, how it is built, or all three. It can be a new design on the same platform, a new platform with the same pages, or a rebuild of everything. The bigger the change, the more it costs and the more can go wrong.

The decision is hard because the pressure to redesign is rarely about evidence. Nielsen Norman Group’s article “Radical Redesign or Incremental Change?” (2015) says many redesigns start from boredom, a sense that the site looks dated, or a competitor’s new site. Visitors are task-focused and usually prefer familiar designs, so a fresh look alone seldom helps them.

So start with a different question: what specific problem, shown by what data, would a redesign fix? If you cannot answer in one sentence, you are not ready to decide.

Signal one: performance data

Speed and stability are measurable. Google’s web.dev defines three Core Web Vitals: Largest Contentful Paint (how fast the main content appears), Interaction to Next Paint (how quickly the page responds to a tap or click) and Cumulative Layout Shift (how much the page jumps around). The “good” thresholds are 2.5 seconds or less, 200 milliseconds or less, and 0.1 or less, measured at the 75th percentile of real page loads.

Check Search Console’s Core Web Vitals report. It uses field data, meaning measurements from real visitors, and labels each group of similar pages Good, Needs improvement or Poor. A group takes the status of its worst metric. If your main templates are Poor on mobile, you have a real problem.

Then ask whether it can be fixed without a redesign. Oversized images, heavy scripts and cheap hosting can often be fixed in weeks. If the slowness is built into the platform or theme, so that every fix fights the system, that points toward a rebuild.

Be honest about the stakes. Google’s page experience documentation says good scores do not guarantee top rankings and that relevance comes first. The business case is stronger on conversion: in a web.dev case study published in 2021, Vodafone ran an A/B test on a landing page and found that a 31 percent better LCP went with 8 percent more sales. That is one company’s test, so run your own measurements.

Signal two: conversion data

Conversion rate is the share of visits that become the action you want: an enquiry, a booking or a purchase. Divide the actions by the visits for the same period and the same segment.

Segments matter more than the average. Suppose, hypothetically, mobile brings 2,000 visits and 10 enquiries (0.5 percent) while desktop brings 1,500 visits and 30 enquiries (2 percent). The gap says the mobile experience is the problem, not the whole site. That points to a targeted fix.

Nielsen Norman Group lists “conversion problems that are site-wide and severe” as one reason to consider a major overhaul. The word site-wide is the test. If one form, one page or one traffic source underperforms, fix that. If every page and every device converts poorly and testing has not moved the numbers, the structure or message may be wrong.

Add qualitative evidence. Watch five visitors try to complete a task on the current site. Nielsen Norman Group’s research says about five users in a qualitative test surface most usability problems, and a day or two of testing costs little compared with a rebuild.

Signal three: security and maintenance

Some sites are risky to keep running. Warning signs are a platform or plugins that no longer receive updates, a developer who left no documentation, changes that routinely break something else, or a site that cannot meet current standards such as HTTPS. web.dev explains that HTTPS protects the integrity and privacy of what visitors send and receive, and that modern browser features require it.

Count the maintenance cost honestly. If every small edit needs a developer, or a plugin update threatens the site, the platform is costing you money every month. Compare that yearly cost with the cost of moving to a platform your team can edit safely.

An unsupported platform is the clearest case for a rebuild, because patching it only delays the move. It is also the case where a careful site migration plan matters most, since the old and new structures will differ.

Signal four and five: brand change and mobile

A real brand change is a legitimate reason. A new name, a merger, a new positioning or a move to a different audience makes the old messaging wrong. Updating a logo and colours can be done inside the existing structure. Changing who you serve and what you promise usually means new pages and a new structure.

Mobile is the other signal. Google indexes and ranks sites using the mobile version of the content, according to its mobile-first indexing documentation, and it recommends responsive design that serves the same page to all devices. If your phone version hides content, loads separately or breaks forms, you have a search problem and a conversion problem at once.

Test the five most important tasks on a real phone: find the service, read the price or process, find the phone number, fill in the form, and pay or book. Any task that fails is a candidate for a fix. If most fail, the layout needs rebuilding.

The case for iterating instead

Nielsen Norman Group’s default advice is incremental change, because user experience is a quality discipline and small steady improvements compound. Its advice is to make the smallest change that solves the problem. Iteration also tells you what worked. Change everything at once and you cannot tell which change helped and which hurt.

NN/g lists the conditions for a major overhaul: incremental fixes have reached diminishing returns, the technology cannot support critical needs, the architecture is tangled, conversion problems are site-wide and severe, or benchmarking shows the site is far behind competitors. Outside those, iterate.

NN/g’s own homepage redesign is an example. Its 2018 case study describes testing three rounds of prototypes and then rolling out gradually, applying shared header, footer and styles to the old pages in the meantime. Even a team that chose a redesign avoided a single big launch.

A decision framework you can use this week

This is our rule of thumb, not a research finding. Answer three questions with data from the last three to six months.

• Is there a measurable problem? Failing field data, low conversion in a key segment, or a platform you cannot update safely. If not, stop: do not redesign.
• Is the cause local or structural? A slow template, a weak page or a clumsy form is local. A platform limit, a tangled structure or a changed brand is structural.
• Would fixing the local causes cost more than the rebuild? Price the fixes honestly, including the time to coordinate them.
• Can you afford the migration risk? A redesign needs a 301 redirect plan, testing and monitoring, covered in redesigning without losing SEO.

What is changing

Responsive layouts, field-data metrics and AI-assisted search are changing what a site has to do. Google’s documentation on page experience (updated in 2026) names Core Web Vitals as the page experience signal its ranking systems use, and says chasing perfect scores purely for SEO is not advised. Treat the thresholds as a floor.

Assistants also read your pages. Clear structure, plain answers and accurate facts help both visitors and answer engine optimization. That is a reason to fix content and structure, which a redesign can do, and a reason not to treat design alone as the goal.