Buying a Website? How to Check It Isn't an AI Content Farm

Updated October 15, 2026. 8 min read

Buying a content website used to mean checking traffic, revenue and backlinks. Now there is a fourth question: how much of this site was produced by AI, and could a single Google update wipe out what you are paying for? Here is the checklist we would run before sending money.

Using AI is not automatically a deal-breaker. A well-edited site where AI helped with drafts can be a sound asset. The risk is a site whose traffic depends on mass-produced pages or on a repurposed domain, because those are exactly the patterns Google's 2024 spam policies target. This is general guidance, not financial or legal advice, so involve a professional for large deals.

1. Ask directly, and get the answer in writing

Start with the simplest step. Ask the seller how the content was produced: who wrote it, which tools were used, how it was edited and what the publishing schedule looked like. An honest seller will answer plainly. Then ask for the key statements to be included in the purchase agreement as representations, so a misleading answer has consequences.

Pay attention to vague answers. "We use a team of writers" with no names, invoices or style guide is worth probing.

2. Check how the site was built

Run the domain through Bot or Built or check the source yourself. Look for:

• Fingerprints of AI site builders such as Lovable or Durable.
• WordPress plugins that write or auto-publish posts, with paths like /wp-content/plugins/aiomatic or /plugins/ai-engine.

An AI builder isn't a problem for a small business site. For a site sold on the strength of its content, an auto-blogging plugin is a serious question to raise.

3. Look up the domain's history

Two quick checks:

• Registration date. Use any WHOIS or RDAP lookup. If the seller says the site is ten years old and the domain was registered last year, something doesn't add up.
• What the domain used to be. Open it in the Wayback Machine. If a "personal finance blog" was a dental practice or a local newspaper three years ago, the current site is built on an expired domain. That pattern matches Google's expired domain abuse policy, and it is one of the most common setups for modern AI content farms.

4. Read the traffic history, not just the total

Ask for read access to the site's Google Search Console, not screenshots. In the Performance report, set the range to the full history and look for:

• Sharp drops around Google updates. A cliff in early 2024 is worth asking about, since that is when Google rolled out its scaled content and expired domain policies.
• Very fast recent growth. Traffic that went from nothing to tens of thousands of visits in a few months often comes from publishing at a pace no human team sustains. It can also disappear just as fast.
• Manual actions. Check the Manual actions report in the same account. It should say no issues detected.

5. Measure the publishing pace

Count how many articles were published per week over the site's life, using the CMS dashboard if you have access, or the archive pages and sitemap if you don't. Then compare it with the team the seller describes. Two writers can't produce 300 thorough articles a month. Keep in mind that sitemap dates can be rewritten by some platforms, so confirm with the dates shown on the articles themselves.

6. Read ten random articles like an editor

Pick ten articles at random, not the ones the seller highlights, and check each for:

• First-hand detail. Photos the author took, prices, specific experiences, things only someone who did it would know.
• Accuracy. Verify three facts, figures or quotes per article. Invented sources and confident errors are a strong sign of unedited AI text.
• Originality. Search a distinctive sentence in quotes. If the same content appears on other sites, it may have been rewritten from them.
• Leftovers. Search the whole site for phrases such as "as an AI language model".
• Template sameness. If every article has an identical structure down to the same FAQ format, it was likely produced in bulk.

Don't rely on AI-detector scores for this step. As we explain in our article on detecting AI writing, popular tells and detectors have real error rates, and a detector score is not evidence you can negotiate with.

7. Check the authors

Look up every author listed on the site. Real writers usually have a footprint outside it: other bylines, profiles, social accounts with history. If the seller can't connect an author to a real person they paid, ask why the name exists.

8. Compare indexed pages with published pages

The Pages report in Search Console shows how many URLs Google has indexed and how many it has chosen not to. A site where most published pages are "crawled, currently not indexed" is a site Google doesn't find valuable, whatever the traffic is today.

9. Look at the backlinks in context

Using any backlink tool, check whether the links pointing to the site match its current topic. A cooking blog whose strongest links are from city council pages and university sites about a long-closed business is very likely sitting on a repurposed domain.

Red flags and how to weigh them

Walk away, or price in a large discount

• The domain belonged to an unrelated organization and the current content started recently.
• An auto-publishing AI plugin is installed and the seller didn't mention it.
• Hundreds of pages a month with no writers you can verify.
• A manual action in Search Console.

Ask more questions

• Traffic jumped sharply in the last six months.
• Most pages aren't indexed.
• Articles are accurate but generic, with no first-hand detail.

Usually fine

• The seller openly uses AI for drafts, and the articles you sampled are accurate, specific and clearly edited.
• The site was built with an AI builder but its content and history check out.

The short version

You are not buying articles, you are buying the trust search engines place in them. Content produced at scale with little oversight, or parked on a borrowed domain, is the kind of trust that can be withdrawn in one update. Spend an afternoon on these checks before you spend the money.