Because familiarity is part of what people are judging. When you change a site, returning visitors have to relearn where things are, and that effort feels like a flaw in the design. The work looks better to you and feels worse to them, and both reactions are real.
We see this on every redesign that involves an established brand. The new site is faster, clearer, and better organized. Two weeks after launch someone forwards an email from a long-time customer saying they preferred the old one. The instinct is to defend the work. The better instinct is to ask what they actually lost.
Most of the time they lost a landmark. A button that used to be blue and on the left. A page they reached by muscle memory. Recognition is built out of small things like that, and they are cheap to preserve if you know to look for them.
It is a handful of visual and structural signals people learn without noticing. The primary color, the logo and its placement, the typeface, the shape of the navigation, and the layout of the page they visit most. Change one or two and the site still feels like you. Change all five and it feels like a different company.
The important word is signals. Nobody consciously remembers your hex codes. They remember that your site is dark with an orange accent, that the menu sits across the top, and that pricing is the fourth item. That mental model is an asset you spent years building.
This is why we start redesigns by writing down what must survive. It is a short list, usually four or five items, and it is the most useful document in the whole project because it turns a vague fear into a constraint the design can work with.
That conventional layouts win on first impression. A study by Tuch, Presslaber, Stoecklin, Opwis, and Bargas-Avila, published in the International Journal of Human-Computer Studies in 2012, found that visual complexity and prototypicality affect aesthetic ratings within the first 50 milliseconds of exposure. Sites with low complexity and high prototypicality were rated most appealing.
Prototypicality is the technical word for looking like what people expect a site of that type to look like. The finding is not that originality is bad. It is that people form a judgment before they have consciously processed anything, and at that speed they are pattern matching against every similar site they have seen.
The same research found the effects persist at exposure times as short as 17 milliseconds, with visual complexity registering earlier than prototypicality. Whatever your redesign does, it is being judged before anyone reads a word of the copy.
Jakob Nielsen at Nielsen Norman Group put the practical version of this bluntly back in 2009, writing that "typically, a fresh design will be a worse design simply because it's new and thus breaks user expectations". His recommendation was to "evolve a UI with gentle changes rather than offer a totally fresh design". That advice has aged well.
The primary color, the logo, the navigation structure, and the URL of your most visited pages. Each of those is a landmark. Changing one at a time is a design decision. Changing all four in a single launch is a rebrand, and it should be planned and communicated as one.
Color is the strongest of the four and the most frequently sacrificed, usually because a new designer finds the existing palette dated. Often it is. The fix is almost always to refine the same color rather than replace it, adjusting saturation and lightness until it works in modern contexts while staying recognizably the same hue.
Navigation structure is the one people underestimate. If your pricing page has been the fourth item for five years, a returning visitor's hand goes there before their eye does. Reordering the menu for logical reasons is fine, but it is a real cost and should be worth paying.
A lot, if you keep the landmarks. In our experience you can rebuild the entire layout, replace the typography at body level, change the photography style, and rewrite most of the copy, and people will still say the site "looks like you" as long as the color, the logo, and the navigation shape survive.
That is a genuinely liberating constraint. Teams often assume that keeping recognition means keeping the design, so they end up with a timid refresh that solves nothing. The opposite is true. Protect four things properly and you have freedom everywhere else.
Nielsen named the exception worth knowing. He argued that a radical redesign is justified when incremental changes have already produced an overly convoluted design with no unified structure. That is a real condition, and if your site has reached it, small changes will not save you. We wrote about how sites get there in our piece on why website redesigns fail.
Change the execution, not the identity. Keep the color and update how you use it. Keep the logo and give it room. Keep the typeface family and fix the scale, spacing, and line height. Modern design is mostly better spacing and better hierarchy, not different ingredients.
Most sites that look dated do not have a color problem. They have a spacing problem, a type scale problem, and too many competing elements on screen at once. Fixing those three things produces a dramatic improvement while leaving every recognition signal intact, which is why it is where we start.
Building the new work as a system rather than a set of pages also protects you later. When the spacing, type scale, and color usage are defined once and reused, the site stays coherent as new pages appear. Our guide to design systems covers how we set that up.
Treat URL changes as the highest risk part of the project. A visual redesign that keeps every URL intact carries almost no search risk. The moment the URL structure changes, you are running a migration, and migrations lose traffic when the redirect mapping is incomplete.
Our default position is to keep existing URLs unless there is a strong reason to change them. A tidier structure is not a strong reason. Merging genuinely duplicated pages or fixing a structure that confuses users is. When URLs do change, every old URL needs a permanent redirect to its closest equivalent, mapped one by one rather than swept into a catch-all.
The full process is in our guide to website migration without losing SEO. It is worth reading before the design work starts, not after, because the URL decisions shape the sitemap.
Show it to people who already use the site, not just to people seeing it fresh. New visitors will always prefer the new design because they have nothing to compare it against. Existing customers are the ones carrying the recognition you are trying to protect, and they will tell you which landmark you moved.
A useful exercise is to show a returning customer the new homepage for a few seconds, take it away, and ask what they remember. If they describe your brand, the design is working. If they describe a generic modern website, something in the identity has been sanded off.
Also check the numbers you already have. The pages with the most traffic and the highest conversion rates in Google Analytics and Search Console are the ones where changes cost the most. Redesign them last and with the most care.
Expect them, log them, and wait before acting. Nielsen's observation that any new interface generates complaints simply because it is new means the first two weeks of feedback are not a reliable signal. Separate the "this is different" comments from the "I cannot find X" comments, because only the second kind describes a real problem.
The "I cannot find X" complaints are gold. They tell you exactly which landmark you removed, and they are usually fixable in an hour without touching the design. A returning menu item, a redirect, or a link added back to the footer resolves most of them.
Watch the behavior rather than the emails, too. Redesigns also tend to add weight, and the HTTP Archive Web Almanac reported the median mobile home page at 2.6 MB in 2025. A slower site will hurt you more than an unfamiliar one, and it is easier to measure.
Write down the four or five things that must survive the redesign before anyone opens Figma. Color, logo, navigation shape, top page URLs, and whatever else your customers would notice. Everything not on that list is fair game, and having the list makes the rest of the project faster rather than slower.
Redesigns fail when nobody defines what is being protected, so every decision becomes an argument about taste. Define it once, at the start, and the arguments turn into design problems with clear constraints. That is a much better place to work from.
If you are planning a redesign and want a second opinion on what to keep, let's talk. We're happy to look at your site and tell you honestly which parts are carrying the brand at phoenix.studio.
Tell us where you want to go. We'll tell you how we'd get you there.