Why AI-Built Websites Struggle in Google, and How to Fix It

It is mostly a rendering problem

Google does not penalise a site because an AI tool wrote the code. What it evaluates is what it can fetch and the usefulness of what it finds. Many AI builders default to React single-page apps, so the server sends a nearly empty HTML shell and JavaScript fills in the page afterwards.

Google can handle that, within limits. Its JavaScript SEO documentation describes three phases: crawling, rendering and indexing. Pages may wait in a render queue, and the same page recommends server-side rendering or pre-rendering, partly because not all bots can run JavaScript. That second point matters more now, because you also want AI search crawlers to read your pages. For the details, see whether Google renders JavaScript.

<title>My App</title>
<body>
  <div id="root"></div>
  <script src="/app.js"></script>
</body>

<title>Pricing | Marlo</title>
<body>
  <h1>Pricing</h1>
  <p>Plans from €9 a month.</p>
</body>

The five-minute test

Run these on your homepage and one inner page:

1. View the raw source. Right-click and choose View Page Source, which shows the HTML before JavaScript runs. Search for a sentence from the page. If it is missing, and you mostly see an empty div and script tags, your content depends on JavaScript.
2. Check the title. Is the title in the source specific to that page, or the same on every route?
3. Request a page that does not exist. Visit /this-page-does-not-exist and check the status in your browser's network panel. It should be 404. A 200 with a “not found” message is a soft 404.
4. Inspect a link. Internal links should be real a elements with an href attribute.
5. Use Search Console. Run URL Inspection on the page, then look at the rendered HTML Google captured and whether the page is indexed.

The fixes, in order of impact

Serve real HTML. This is the big one. Your options, roughly from simplest to most involved:

The last option is often underrated. Pages behind a login do not need to rank, so a logged-in app can remain a single-page app while your public pages are pre-rendered. Our guide to SSR, CSR and SSG goes further into the choice.

Give each route its own metadata. Every indexable page needs a unique title, description and canonical URL, present in the HTML rather than added later. AI-built projects often keep one template title everywhere, which makes pages look like duplicates.

Return honest status codes. Google notes that client-side routing can make status codes difficult and can cause soft 404s. It suggests redirecting to a URL that returns a real 404, or adding a noindex tag to error views. Be careful with the second approach: if a page starts with noindex, Google may skip rendering, so JavaScript cannot reliably remove the tag later.

Use real URLs and links. Google only discovers links that are a elements with an href, and it cannot reliably resolve fragment URLs like #/pricing. Clickable div elements that call a router function are invisible as links. Ask your tool to change them; this tends to be a quick fix.

Publish a sitemap and check your robots file. Generated projects often lack a sitemap or carry leftovers from a template. Create an accurate sitemap, submit it in Search Console, and check that your robots.txt does not block the scripts or pages Google needs.

Move critical structured data into the HTML. If it only appears after JavaScript runs, you are relying on the same queue. Add it at build or server time where possible.

Then fix the content

Technical fixes only let search engines see what is there. Sites generated from one prompt often have the opposite problem: pages that are thin, interchangeable and light on detail. A services page saying you deliver “innovative solutions tailored to your needs” does not answer anything.

Add what a generator cannot know: your prices or how they are set, your location, real photos, named people, examples of past work, and answers to the questions your customers ask. Those details help readers decide, and they are hard for competitors to copy. We have seen this tension on our own site: when we audited our guides, templated pages with near-identical text were the problem, not the tooling behind them.

What to expect

After shipping fixes, resubmit your sitemap and request indexing for your most important pages. Then give it some time, because a site with little authority is crawled less often, and Google does not promise to index every page. If pages remain out of the index, work through why a site is not showing up on Google and why pages get crawled but not indexed.

If you would rather have someone check the build and fix it properly, book a call.

Common questions