Sinton Agency Logo
Sinton Agency Logo

Published

·

6 min

read

AI slop cleanup: when to rescue a website and when to rebuild

An AI-built site that is not winning work leaves you with one decision: patch it or start again. Here is an honest way to tell which one you need, what each route really costs, and how to avoid paying twice.

Line drawing of a browser window split in two, one half a tidy interface on a grid, the other coming apart into broken blocks.

TL;DR

  • AI-built sites usually underperform for structural reasons, not cosmetic ones.

  • A cleanup fixes what you can see. It cannot fix foundations.

  • Five checks show you which route you are on.

  • The expensive mistake is patching now and rebuilding six months later.

  • Decide before you spend, not after.

You have a live site, and it is not bringing in clients

The site went up fast. An AI tool did most of the work, and at a glance it looks like a website. Months later, the enquiries are not there.

If you are not yet sure whether AI output is the problem, start with our guide on how to spot a vibe coded website. This article picks up after that, when you know something is wrong and have to decide what to do. There are two routes: clean up what you have, or rebuild it. Both options are legitimate. Picking the wrong one is what costs money.

What the AI build already cost you

Before you spend anything on a fix, it helps to see what is already on the invoice.

  • Indexed pages you did not plan. Thin content, duplicate titles and a muddled heading structure are already in Google's index. In its guidance on AI-generated content, Google says it judges content on quality and usefulness rather than how it was produced. The AI is not the problem. The thin pages it shipped are.

  • No design system. Without reusable components, type scale, or spacing rules, every new page is improvised from scratch. Small changes take hours, and the site drifts further out of line with itself.

  • Code nobody wants to inherit. Generated markup tends to be heavier than it needs to be, which hurts performance. Google's Core Web Vitals are the public measure of that, and heavy code is difficult to trim without rewriting it.

  • Buyers who have already gone. Anyone who visited, hesitated, and left isn't coming back to check whether you tidied up.

None of this means the site is worthless. It means you have to choose the fix with the real starting point in mind.

Five checks that decide rescue or rebuild

Run these five questions over your own site. Each one points in a direction.

  1. Are the foundations sound, or is the structure itself the problem? If the pages exist in a sensible order and each one has a clear job, you are looking at a rescue. If the site is one long homepage with no route to an enquiry, no cleanup will fix that.

  2. Can you maintain the platform? A site built in Framer or Webflow, or on a standard CMS, can be worked on by almost anyone (even without technical skills). A generated codebase on a hosting setup nobody documented usually cannot.

  3. Is there a design system, or a pile of one-off styles? Check three pages for the same button, the same heading size and the same spacing. If they differ, there is no system to build on.

  4. Did you write the content for your buyers? Read your service page and ask whether it answers the questions clients actually ask you on a call. You can rewrite generic or hyperbolic copy. A page structure that ignores the sales conversation needs replanning.

  5. Can you change sections by yourself? Try editing one section yourself. If you cannot do it without risking a layout break, you don't own a website. You own a snapshot.

Rule of thumb: if two or more of these point towards a rebuild, a cleanup is deferred spending rather than a saving.

What a rescue covers, and where it stops

A rescue is a focused round of design and technical work on a site that is fundamentally sound. It typically covers spacing and layout consistency, type and contrast, trimming what slows pages down, the SEO/AEO basics, and a copy pass on the pages that matter most.

It is quicker and cheaper, and for some sites it is the right answer.

It won't change what the site is for. It cannot invent positioning that was never decided, add a conversion path the page structure does not allow, or make a fragile build stable. If those are your problems, a rescue buys you a tidier version of the same result.

What a rebuild involves

A rebuild starts before the design. First the decisions: who the site is for, what each page has to do, and what a visitor needs to see before they get in touch. Then the structure, then the design and build.

We build in Framer and Webflow, so a rebuild takes weeks rather than months, and you get a site you can update yourself without breaking it. You also get a design system, which means the next ten pages cost a fraction of the first one.

The part people worry about is search. Rankings can dip during a move, and the way to protect them is the boring way: keep the URLs you can, redirect the ones you change, and follow Google's documentation on site moves with URL changes. Plan that before launch, not after.

Rescue or rebuild?

A short call and you will know which one.

How to avoid paying twice

The most common outcome we see is a cheap patch in spring and a rebuild by autumn. The patch was not wrong because it was cheap. It was wrong because nobody checked whether the foundations could carry it.

Before you commit to either route, do three things:

  1. Run the five checks above and write down the answers.

  2. Ask whoever quotes you what the work will not fix. A straight answer tells you more than the price does.

  3. Agree what success looks like in numbers, whether that is enquiries per month or a page that finally loads quickly on a phone.

If a cleanup gets you there, take it. If it does not, paying for it first makes the rebuild more expensive, not less.

Frequently asked questions

Can I just fix my AI-built site instead of rebuilding it?

Sometimes, yes. If the structure is sensible, the platform is maintainable and the problems are visual, a focused round of work is usually enough. If two or more of the five checks point the other way, a fix will not hold.

How much of the content can I keep if I rebuild?

Usually more than you expect. Your services, prices, photography and any real client proof carry over. What tends to be replaced is the generic AI copy and the page structure around it.

Will a rebuild damage my Google rankings?

There can be movement while the new pages are crawled and indexed, and nobody can promise otherwise. Keeping your URL structure where possible and redirecting the rest, as set out in Google's site move documentation, is what keeps that movement small.

Do you offer AI slop cleanup as a service?

Not as a standalone service. We design and build websites, and cleaning up someone else's generated code is rarely the best use of your budget. If we look at your site and think a cleanup is genuinely the right call, we will tell you that, even when it means no project for us.

How long does a rebuild take, and what do you need from me?

Weeks rather than months for a typical service business site. What we need is time from you at the start, on your goals, your buyers and your proof, plus your logo files and any content you want to keep.

Profile Picture of Thomas Sinton

Thomas Sinton

Author

Share this post via

We will create a website that attracts customers.

We will design a website or branding for conversion that will help your business increase sales and gain more customers.

Or write to us via the contact form.

Sinton Agency Logo

We will create a website that attracts customers.

We will design a website or branding for conversion that will help your business increase sales and gain more customers.

Or write to us via the contact form.

Trusted by

Trusted by

Trusted by