Do you need a website redesign, or just a fix?

Most sites redesigned to fix conversions did not have a design problem. They had two or three specific problems, and the rebuild buried them under a hundred other changes.

You are about to spend somewhere between eight and forty thousand euros.

The site feels dated, sales are flat, and a rebuild feels like the grown-up decision.

Here's the kicker. Conversion.com reviewed more than 60 client redesigns and found that fewer than one in four produced a measurable improvement in conversion rate. That is one agency's book, so weigh it accordingly. The same pattern appears where nobody is selling anything: Kohavi and Longbotham report in the Springer Encyclopedia of Machine Learning that only a third of changes tested at Microsoft improved the metric they were built to improve.

So before you sign anything, work out which problem you have.

Do you need a website redesign, or do you need a fix?

You need a redesign when the site cannot do the job at all: the platform blocks what you sell, the brand is genuinely wrong, or the structure cannot hold your current products. You need a fix when the site works and specific pages leak. Most businesses booking a redesign are in the second group.

Both feel identical from the inside, and the wrong choice costs about thirty grand.

How do you tell which one you have?

Run five checks. Each takes about two minutes and needs nothing but your analytics and your own site open on a phone. Count the yes answers when you are done. Four or five of them points to a rebuild. Zero to two points at a fix instead.

1. Can you name the pages that leak? If one of your top five landing pages converts at a third of the others, that is a page problem, and a page problem is a fix. If all five are equally poor, it is structural.
2. Does the platform block something you sell? Not "it is annoying to edit." Something you genuinely cannot do: subscriptions, multi-currency, a configurator, proper variants.
3. Has the business changed since the site was built? Different customer, price point or product line. A site built for one business rarely holds a different one.
4. Would a stranger describe what you sell correctly? Show the homepage to someone outside your industry for five seconds, then ask what you sell and who for. If they cannot say it, that is copy and hierarchy, both fixes.
5. Is it slow on a phone? Largest Contentful Paint over 2.5 seconds is worth fixing, and it is a fix. Plenty of new sites launch slower than the ones they replaced.

Rather not do it by hand? Revslip runs the checks on your URL and prices each finding.

Why do so many redesigns fail to move the needle?

A redesign changes everything at once, which makes the result impossible to read. CXL's argument is that when a hundred things move together, wins and losses cancel and nobody can tell which change did what. You get a new site, a flat conversion rate, and no information.

Say one old page converted well for a reason nobody wrote down. The rebuild replaces it, and conversion drops four percent. Was that the good page, the navigation, or the extra 300KB of JavaScript?

Nobody knows. That is the real cost, and it is worse than the invoice.

A redesign does not only risk your conversion rate. It destroys the evidence you would need to work out what went wrong.

What does the research say about big-bang rebuilds?

Treat a rebuild as a high-risk event, not a routine upgrade. CXL argues for changing one thing at a time so each change reports its own result, and Conversion.com's review points the same way, naming speed regression and removed content as the usual causes of a drop.

Two things get lost in most rebuilds:

1. Speed. New builds run heavier. If the old site loaded in 1.8 seconds and the new one takes 3.4, you paid to lose money.
2. Content that was quietly working. The dull specifications table. The testimonial with a real name and a real number in it. These get cut for looking untidy, and some were selling.

Worth saying plainly: the redesign-specific evidence is thin. Sixty clients is one agency's book, not a controlled study. It is credible only because it points the same way as the Microsoft figure.

How do you fix a site without rebuilding it?

Work in this order, and change one thing at a time. Every step reports its own result, which is the whole point and the thing a rebuild takes away from you. Most of these cost days rather than months, and none of them commit you to anything you cannot reverse next week.

1. Find the leak first. Rank pages by traffic multiplied by the gap between their rate and your best page's. The arithmetic is in how to calculate conversion loss.
2. Fix the message before the layout. Headline, subhead, the first thing a stranger reads. It costs nothing and it is the most common finding in any audit.
3. Cut friction on one page. One form, one checkout, one signup. Remove what nobody uses, then read the number two weeks later.
4. Then consider design. By now you know which pages matter and what the message is. A designer given that brief produces something different from one given "make it modern."

Most sites get most of the gain in step two.

Find the leak first. Run a free audit before you commission anything.

How will you know it worked?

Watch conversion rate on the page you changed, not the site-wide number. Compare two full weeks against the two before, matched day for day. Under roughly 400 conversions a fortnight, treat the result as directional rather than proven, and change one thing at a time.

If your site does 60 sales a month you will never get a clean A/B result on a headline change. That does not make the change wrong. Your evidence is the mechanism and the before-and-after, not a significance calculator.

When is a redesign genuinely the right call?

When the site cannot do the job. Not when it is ugly, not when you are bored of it, and not because a competitor launched something shiny last month. Four situations genuinely justify a rebuild, and none of them are about taste.

1. The platform blocks revenue. You cannot sell subscriptions, take the currencies you need, or run the configurator customers keep asking for.
2. The business changed. Different buyer, price or category. The old site was built for a company that no longer exists.
3. Mobile is broken beyond patching. Layout that fails at 360px because of how the thing was built. Check the gap is real first, because mobile converting below desktop is normal and smaller than most people think.
4. The brand works against you. Rare, and claimed more often than true. If you sell to enterprise and the site looks like a 2013 hobby project, it counts.

Even then, diagnose first and rebuild with a list of known problems in hand.

Questions people ask before commissioning a rebuild

These come up once someone has run the five checks and is deciding whether to spend. The pattern underneath them is the same: a rebuild is the most expensive way to find out what was wrong, and almost always the least precise.

How much does a website redesign cost?

Agency quotes for a small business site commonly land between eight and forty thousand euros. The number that matters more is the opportunity cost: three to six months building instead of fixing.

Will a redesign improve my SEO?

Usually not, and it frequently hurts. URL changes, lost internal links and slower pages are the three common ways a rebuild loses traffic. Map every old URL to a new one.

How do I find out what is actually broken?

Split your analytics first. The traffic but no sales diagnostic ranks the usual causes by how often each is the culprit, and why is my checkout abandoning does the same for the final step.

Can I just ask ChatGPT whether my site needs rebuilding?

You can ask, but the answer rests on a partial copy of your page and no measurements, which is a poor basis for a five-figure decision. The gap is explained in pasting your URL into ChatGPT is not a website audit.