Does AI Content Rank? What Google and AI Engines Reward From Tech Company Blogs

Does AI content rank? It can. Google says AI use earns no special gain and appropriate use isn’t against its guidelines, so a useful, original page has a chance. A scaled page that adds nothing doesn’t.

Here’s a hypothetical that will sound familiar. A 30-person IT services firm connects an AI writer to its blog in January and publishes 20 posts a month, and by June it has 120 articles, one product claim that was never true, and a traffic line sitting exactly where it started.

Nobody checked. Nobody owned it.

The tool didn’t cause that. The system around the tool did, because AI scales whatever you hand it, good or bad, and a blog with no owner, no review and no point of view is a bad system running at 20 posts a month. Owners keep asking whether AI content ranks, and the argument around that question runs louder than the answer deserves.

Does AI Content Rank? What Google Actually Says

Google answered this in February 2023, and the answer hasn’t moved. Its guidance on AI-generated content says “Using AI doesn’t give content any special gains. It’s just content.” If the content is useful, helpful, original and shows E-E-A-T, Google says it might do well. If not, it might not.

The same page says appropriate use of AI or automation isn’t against Google’s guidelines, and the line it draws is purpose, because content generated primarily to manipulate search rankings breaks the spam policies and did so long before anyone had heard of a language model.

Then there’s accuracy. Google’s page on using generative AI content points out that these models don’t retrieve facts. They predict a likely sequence of words, so the output can contain inaccuracies. Google tells publishers to manually fact-check and review all of it before it goes live, titles, meta descriptions, structured data and image alt text included.

Read that twice. Review isn’t a nice-to-have. It’s the instruction.

What the Ranking Data Shows

In July 2025, Ahrefs ran the numbers on 600,000 ranking pages. The setup was 100,000 random keywords, the top 20 URLs for each, and its own AI content detector pointed at the lot. Results below.

Then it checked whether more AI text meant a lower position. It didn’t. The correlation was 0.011, which the study calls effectively zero, and pages at position one skewed slightly less AI, though Ahrefs itself calls that link very weak.

Before you quote that number, know what it can’t tell you. The detector is Ahrefs’ own, and Ahrefs admits no detector is perfect. Only pages that ranked were counted, so every AI page that sank without a trace is missing from the picture. A near-zero correlation isn’t proof that AI helps, either. That 81.9% could hold an expert’s draft with light AI help or a raw, unreviewed output, and the detector can’t separate the two.

Here’s what survives. AI-assisted pages do rank. Why they rank is something the study can’t answer.

Where AI Content Gets Cut

Google’s spam policies define scaled content abuse as many pages generated for the primary purpose of manipulating search rankings and not helping users, however they’re made, and the first example on the list is using generative AI tools to produce many pages without adding value for users.

It reaches into AI search, too. Google’s guide to generative AI features warns against building separate content for every variation of a query, fan-out queries included, primarily to manipulate rankings or AI responses. It adds that “a high quantity of pages doesn’t make a website higher quality or more relevant to users.”

Tech companies get here fast. Forty versions of one answer. The same service page cloned for every city and every industry. A post for each keyword in a spreadsheet.

I’ve said it about leads: more leads won’t fix an MSP’s growth. Pages work the same way. More pages won’t fix a blog that nobody owns, and a faster writer just buries the problem under more pages.

Does AI Content Rank in ChatGPT, Perplexity and Claude?

Nobody publishes a ranking formula for these tools, and nobody publishes a citation rate that splits human-written pages from AI-written ones either, so I won’t invent one. What does exist is documentation about access and eligibility, and it’s worth reading before you worry about anything else.

Google says optimizing for generative AI search is still SEO, and its guide to generative AI features says a page must be indexed and eligible to be shown in Search with a snippet. Same plumbing as always.

OpenAI documents three crawlers, and they do different jobs. OpenAI’s crawler page says each setting is independent of the others.

So the first check on whether AI content ranks in ChatGPT is boring. Can the search bot reach the page? A blanket block written for training can shut the door on search too. If you want the longer version, we wrote up how to rank in ChatGPT answers and the difference between AEO versus GEO.

After access, it’s the same question Google asks. Is the page worth quoting? A model that needs one clean, accurate, specific answer will find a generic AI draft no more useful than the ten others it already has.

What Rewards a Tech Company Blog

Google’s guide draws a line worth borrowing. Commodity content, its example is “7 Tips for First-Time Homebuyers,” rests on common knowledge that could come from anyone. Non-commodity content gives an experienced take that goes past the ordinary. The guide says not to recycle what others have already said “or could easily be produced by a generative AI model.”

That last clause is the whole test. If a model can write your post with no input from you, so can your competitor’s model.

Tech companies sit on non-commodity material and never use it. Your engineers’ runbook for a client migration. The reason you stopped selling a product. How you price a project against a managed service, and what each does to margin. The question that kept showing up in tickets all quarter. A model wasn’t in any of those rooms, so it has nothing to say about them.

• One decision your team made, with the reason and the trade-off.
• A number you can back from your own books or tickets.
• A mistake, told straight.
• The name and title of the person who knows.

That last item is E-E-A-T in plain clothes, and we cover the author side in E-E-A-T for MSP websites.

The Review Step Before Anything Publishes

AI works like any lever: good thinking gets bigger, and so does the lack of it. Here’s the order I’d run the review in.

1. A named expert owns every claim. Not “marketing.” A person.
2. Check every number, product name, version and link against its source. A model will state a wrong fact in a confident voice.
3. Add something only your company knows, a decision, a figure or a lesson.
4. Put a real byline on it. Google says to consider accurate author bylines where readers would expect them, and notes that giving AI the byline is probably not the best way to make clear how AI was used.
5. Say how it was made where a reader would wonder. Google says AI disclosures are useful where someone might ask how the content was created.
6. Review the metadata too. Titles, descriptions, schema and alt text, because Google’s review instruction covers them.

Ideas are cheap. Execution wins. Unless somebody owns the review, it won’t happen, and the policy stays a wish.

Where Tech Company Blogs Get It Wrong

Volume as a plan is the usual one. Somebody sets a target of 20 posts a month because the tool makes it cheap, and a quarter on, nobody can tell you what any single post was meant to do.

Product claims no engineer read come next. When a security or infrastructure firm gets a protocol or a configuration wrong in print, the damage isn’t a ranking. A technical buyer who spots one error stops believing the rest of the site.

Then you find the keyword spreadsheet turned into forty posts that all answer one question, under a byline that says “Admin” on a page a buyer is meant to trust before a six-figure decision.

And the favorite: chasing a detector score. A pass tells you the text reads as human. Whether it’s true, or worth a reader’s time, is a separate matter, and Ahrefs itself says its detector deals in probabilities, not certainty.

What I’d Do With a 30-Person Firm’s Blog

Back to the hypothetical firm with 120 posts and a flat line. I wouldn’t start by writing anything. I’d sort.

Three piles. Keep what’s accurate and useful, rework what has a good question and a thin answer with a named expert in the room, and remove what repeats another post or states something nobody can support, then redirect it to the best page on that topic.

Then the rule going forward. Four posts a month that someone reviewed beat twenty that nobody did, and if the review step can’t keep up with the publishing schedule, the schedule is wrong, not the review.

Then measure what pays. Qualified conversations, not impressions. If a post can’t be tied to a buyer question your sales team hears, ask why it exists.

So, does AI content rank? Yes, when it’s useful and original, and when the company behind it stands on the claims. The better question is whether anyone on your team owns what goes out. If you want a second set of eyes on yours, see how we approach SEO, AEO and GEO for tech companies, or book a growth strategy call.

Book a Growth Strategy Call