Why Automating Your Brand's Blog With AI Could Blow Up Your SEO (And Your Brand Along With It)

A few weeks ago a friend of mine tells me, all proud:

"Dude, this weekend I'm launching my company's blog. I fed Claude the topics I want to rank for and I already have thirty posts ready. This is gonna explode my SEO."

"don't do it" I told him (I'm not anti AI, I use it every day, my whole agency runs on it), but what he was about to do is exactly what Google has been hunting down since March 2024.

His plan was to drop a dozen serialized pages onto his domain, no experience behind them, no point of view, the same calculated structure as millions of other articles published this week. And he was about to eat a penalty that takes months (sometimes years) to undo.

This article is basically the long conversation we had after that. If you're thinking the same thing he was, or you've got an in-house marketing team that is "tripping" and ready to publish more articles than Gutenberg testing the press, read this first.

The problem isn't AI. The problem is we all sound the same now

There's a stat that we can't ignore: by May 2025, 51.7% of new articles published online were already AI-generated. The number comes from a Graphite study that analyzed 65,000 articles from the Common Crawl between 2020 and 2025. In January 2020 AI text was 2.2% of the total. ChatGPT came out in November 2022 and from there the curve goes almost vertical. By the end of 2024 there were already more articles published by machines than by humans.

At this point you might tell me: "OK Andy, there's more AI content out there, so what?" The "so what" is this: the same study shows that, despite that flood, 86% of the content that actually shows up in Google search results is still human. Not me saying it. Graphite saying it in their AI Content in Search analysis.

Read it again. Half of the new internet is published by AI, but barely 14% manages to rank. That's not Google hating AI. That's Google weighing what actually adds value... separating wheat from chaff.

So the question is no longer "can I use AI to write my blog?" The question is "how likely is my content to land in the 86% human bucket or in the 14% AI that still ranks?" And to answer that you need to understand what happened in March 2024.

What Google did in March 2024 (on top of its usual Core updates)

On March 5, 2024, Google rolled out a core update and three new spam policies at the same time. The one we care about most is called scaled content abuse and it basically says: if you produce a bunch of pages with the primary goal of manipulating search rankings, I don't care how you make them, it's spam. I'll quote it because it matters: "this scaled content abuse is spam, whether produced by automation, by humans, or by some mystical shaman (this last part isn't quite textual, lol)".

So... Google doesn't penalize AI. It penalizes the production pattern. The difference between someone who writes one article a week digging deep into a topic, and someone who generates thirty posts in an afternoon with templated prompts.

Along with the update, Google said it expected to cut "unhelpful" content in its results by 40%. That number sounds like marketing fluff until you see the data on the impact side. Originality.ai analyzed 79,000 sites after the update and found that 1,446 received manual actions (around 2%). Of those 1,446 penalized sites, 100% had some kind of AI-generated content. 50% had between 90% and 100% of their posts made with AI.

The combined hit is estimated at 20 million monthly visits lost and around $446,000 a month in ad revenue that just stopped coming in. For small sites, that's the difference between keeping the lights on or shutting down. For big sites, it's a whole quarter blown.

And here's the part nobody really talks about: once you get hit with a manual action, you're not out in a week. You have to clean the content, wait for recrawls, file for reconsideration, prove you changed the pattern, swear you'll do better as a human being, etc... It can take more than a year to recover traffic... sometimes it's actually healthier to just abandon the domain.

When I showed all this to my friend he went quiet. It was exactly what he was planning. (you're welcome bro)

There's a much sneakier pattern that matters: content that just quietly dies on its own

The thing with these strategies is that at first they seem to work. Analytics is on your side. You launch a push of AI content, you get an initial traffic spike (because Google gives new sites the benefit of the doubt in the sandbox), and then a month or two in you crash to almost zero impressions and you stay there.

That's what happens with most AI content: there's no loud penalty, no Search Console email, nothing to report. Just a ramp that climbs for three months and then a flat plateau at zero that lasts forever. Google never tells you anything. It just stops showing you.

And that, for a brand, is worse. Because when there's a manual penalty you at least find out. When it's a silent algorithmic demotion, you could spend a whole year convinced that "SEO is growing" looking at the first-three-months chart, and really you're investing in an empty room.

EEAT: the filter AI can't pass on its own

In December 2022, Google added an extra E to its famous E-A-T framework (Expertise, Authoritativeness, Trustworthiness) and turned it into E-E-A-T. The first E stands for Experience. First-hand experience.

Why does that matter? Because for YMYL topics (Your Money or Your Life: health, finance, legal, big decisions) Google now prioritizes content where you can tell someone actually lived the thing they're writing about. Used the product. Visited the place. Treated the patient. Tried the case.

That's the one thing an LLM can't have. A model can synthesize what other people wrote about drinking mate in Mar del Plata, but it can't have drunk mate. It can generate an analysis of what it feels like to migrate to Stripe, but it never migrated to Stripe. And when your whole blog is about experiences nobody on your team actually had, Google notices. Not because it has some magic AI detector. It notices because the content is generic, interchangeable, with no concrete anecdotes, no numbers of your own, no photos of the process.

The contrast gets sharper when you put Bankrate next to CNET. Both are huge financial outlets that admitted using AI content. CNET had to correct 41 of 77 AI-generated articles because of factual errors and ate a full-on reputation crisis. Bankrate, on the other hand, still ranks well for competitive financial keywords with AI-assisted content. The difference: Bankrate has human experts review and edit before publishing. CNET didn't.

The conclusion Google has been saying for three years is the same: the problem isn't the tool, it's the lack of oversight. But "oversight" for real isn't doing a quick pass and adding a couple of jokes. Oversight means a human with real experience built the angle, brought the examples, made the editorial calls. AI can help you write faster. It can't replace the person who knows what to say.

The paradox nobody saw coming: big tech is hiring storytellers (or as I like to put it: they don't drink their own kool-aid)

While a good chunk of the market is celebrating that "we don't need writers anymore", the companies that understand AI the most are doing the exact opposite.

The Wall Street Journal published an analysis a few months ago on the explosion of storytelling roles in North American companies. The data:

• The share of LinkedIn job postings in the US that include the word "storyteller" doubled in the last year. There are around 50,000 active listings under marketing and another 20,000 under media and communications.
• On earnings calls and investor days, executives mentioned "storyteller" or "storytelling" 469 times in 2025, vs. 359 in 2024 and just 147 in 2015. The number comes from FactSet.
• Vanta, a compliance company, opened a Head of Storytelling role with a salary of up to $274,000.
• Google, Microsoft, Notion, Chime and Anthropic (yes, the company that makes Claude) are building dedicated narrative teams. Anthropic recently posted a Copy Lead role for Enterprise.

The company that sells the most sophisticated LLMs on the market is hiring humans to write their brand. Think about that for two seconds. It's not because Anthropic doesn't know how to use its own model. It's because they get something the companies automating blindly still don't: when everyone has the same knife, the one who knows how to cook wins.

The part that hurts in-house marketers the most: your brand isn't measured by backlinks anymore

Here's another layer that's going to hurt even more in the next twelve months. AI is changing how brand authority is measured, and the direction is the opposite of what most people assume.

Ahrefs studied 75,000 brands and measured what factors correlate with visibility in AI Overviews, ChatGPT, AI Mode and other generative search surfaces. The results:

• Brand web mentions: correlation of 0.664.
• Brand anchors: 0.527.
• Brand search volume: 0.392.
• Backlinks: 0.218.
• YouTube mentions: 0.737 (the highest of all).

Translation: what moves the needle on whether ChatGPT and Google AI Overviews cite you isn't how many backlinks you have. It's how many people talk about your brand outside your site. In contexts where a real person chose to mention you. On YouTube, in podcasts, in articles, in signed posts.

That only gets built with human content, opinionated, with a point of view. Thirty generic posts on your blog don't generate brand mentions. No podcast is going to quote you. No article is going to feature you. They're noise in an ocean of noise.

Classic SEO (keywords, structure, backlinks) is moving toward something that looks more like PR. And PR is done with a voice of your own, not with templates.

So what do you do with AI

Here's where an in-house marketer might tell me: "OK Andy, what's your solution? That we hand-write forty posts a month?" No. My position is pretty specific and I'll lay it out here, because it's what we do at Ander.Agency and it's what I tell anyone who asks.

First: Less is More (so original). Less noise is needed for something that's actually yours to form. One article a month with a real point of view beats four a week with nothing to say. This goes against everything the content marketing industry tells you, I know. But the numbers from the last two years are clear and they point in the opposite direction of "publish at scale".

Second: use AI where it actually helps, not where it doesn't. It helps for research, organizing drafts, translating, generating structures, brainstorming headlines, editing. It doesn't help to replace the experience of the person who actually has something to say. When I tell a client "this article needs your CEO's voice", it's not purism: it's because the CEO saw things the model didn't.

Third: cut the editorial calendar. Raise the investment per piece. A piece that takes two weeks and is signed by someone with real authority on the topic is worth more than thirty anonymous pieces written in one afternoon. The metric that matters isn't "how many posts I published this quarter" but "how many times my brand was mentioned this quarter".

Fourth: make your content impossible for a machine to generate. How? With proprietary data. With your customer cases. With numbers from your own operation. With opinions that require having been in the room. If an AI could have written your article without privileged access to your company, then your article is disposable.

So, in short

Delete the thirty posts. Sit down with two people from your team who know more about the product than you do. Pull five topics where they actually have something to say that no one else in the market can say. Write them together, one a month, for six months. Use Claude to speed up the edit, not to replace the voice. And in a year, check what happened with traffic.

It's boring advice. But SEO in 2026 is won by productive boredom, not by empty scale. The brands that figured this out early are building something that's hard to replicate. The ones still churning out content in bulk are going to be fighting, six months from now, to get off page six of Google.

The one who already gets where this is going is Anthropic, which while selling the most advanced AI model on the market, hires humans to write their brand.

If you need a team that leans on AI without stepping out of the loop, bringing real experience, judgment and that "anti slop" taste, Better Call Ander.Agency