Is Your "GEO" Just Another Name For Spam?

What if a lot of the GEO advice being shared right now is not actually a new marketing discipline?

What if it is just old-fashioned search spam with a new acronym?

That question is worth asking because some of the most popular tactics being promoted as Generative Engine Optimization should look very familiar to anyone who has worked in SEO for more than a few years. Manufactured listicles designed to place a brand next to competitors. Paid brand mentions inserted into third-party articles. Manufactured recommendations across Reddit, forums, and social platforms. Hundreds or thousands of AI-generated pages targeting every possible variation of a question. Content architectures built around increasingly elaborate query fanouts.

The target has changed. The behavior has not.

The goal is still to manufacture signals that make a system believe a brand or website is more relevant, authoritative, or popular than it otherwise would be.

And that should make in-house SEO teams uncomfortable.

Here is what makes this conversation difficult: some of these tactics probably work, but not for long.

If you can place your company on enough “best X” pages, manufacture enough conversations mentioning your brand, or publish enough content targeting the questions an AI system might retrieve, you may increase the probability that your brand appears in an AI-generated answer.

That is not an absurd hypothesis. Generative search systems still have to retrieve information before they can use it. Microsoft describes grounding as the layer connecting AI systems to current, authoritative information, with embeddings helping retrieve and organize the information eventually used to produce an answer.

Research into GEO also suggests that optimization can influence what gets cited once content enters that retrieval pipeline. The original GEO research reported substantial visibility improvements under its experimental conditions, while more recent work has found topical relevance to be one of the strongest factors affecting citation selection.

SEO has repeatedly gone through the same cycle.

Someone discovers a signal that appears to influence visibility. Marketers find a way to manufacture that signal. Tools make manufacturing it cheaper. Agencies package the tactic. Social media amplifies the success stories. Everyone scales it. Eventually, the signal becomes less useful because it is too easy to manipulate.

That history matters because Google has already made its position unusually clear. Its spam policies explicitly say they cover attempts to manipulate not only traditional rankings but also “generative AI responses in Google Search.” Google also defines scaled content abuse as producing large amounts of content primarily to manipulate rankings rather than help users, regardless of whether the content was created by AI, humans, or some combination of the two.

In May 2026, Google went even further, updating its documentation specifically to clarify that existing SEO best practices remain relevant to its generative AI features and that its spam policies apply to those experiences.

In other words, adding “GEO” to the strategy deck does not create an exemption from the rules we already understand.

Consider the increasingly popular advice to get your brand mentioned in listicles and third-party recommendations.

There is a legitimate version of this strategy. If respected publications, industry experts, customers, and communities genuinely talk about your company, those references create useful context around your brand. They help humans understand what your company does, and they create information that retrieval systems can potentially discover and use.

That is marketing.

The problem starts when the objective changes from earning those mentions to manufacturing them.

Paying to insert your company into dozens of “best software for X” pages is not a brand strategy simply because an LLM might retrieve those pages. Seeding fake recommendations into community discussions does not become sophisticated because someone calls it “LLM seeding.”

The distinction matters because AI search appears to make third-party context more important, not less. One large-scale study comparing traditional and AI search found AI systems heavily favored earned media and authoritative third-party sources over brand-owned and social content.

That creates a legitimate reason for SEO teams to care more about digital PR, analyst coverage, industry publications, customer advocacy, and other forms of earned media.

It also creates an enormous incentive to fake those signals. Those are not the same strategy.

Query fanout deserves similar scrutiny.

Understanding the related questions, concepts, entities, comparisons, and follow-up queries surrounding a topic can be extremely useful. It can expose gaps in your content and help teams understand how their subject connects to the broader information environment.

That is good research.

But there is a short distance between using query fanout to understand a topic and using it to manufacture hundreds of nearly identical pages for every conceivable variation of that topic. At that point, we have recreated scaled SEO content. Only now we have a better spreadsheet.

This distinction becomes even more important as generative systems use embeddings and semantic retrieval. The objective should not be to repeat every possible query variation across your site. It should be to build enough useful, coherent information that retrieval systems can correctly understand where your brand, products, expertise, and content belong within a topic.

That is a fundamentally different optimization problem.

You are not trying to say the same thing 500 different ways. You are trying to make your relationship to a subject unmistakable.

This is where the incentives for agencies, consultants, vendors, and in-house teams can diverge.

A short-term GEO experiment that produces a 30 percent increase in AI citations makes an excellent case study.

The cleanup six months later does not. In-house teams absorb that second part.

They inherit the bloated content libraries, questionable placements, manufactured profiles, polluted analytics, and technical debt created by tactics that were supposed to capture a temporary visibility opportunity.

More importantly, they own the domain if search engines eventually decide those tactics cross the line.

Google says policy violations can cause sites to rank lower or disappear from results entirely, and its policies specifically identify practices such as scaled content abuse, link spam, and attempts to exploit established site reputation.

That does not mean every paid mention or programmatically generated page will trigger a penalty. It means the risk calculation needs to include more than whether the tactic works today.

The right question is whether you would still be comfortable explaining the strategy after the loophole closes.

There is a useful version of GEO hiding underneath all of this. It just looks less exciting on social media.

Start with the same technical foundation SEO has always required. Make important information crawlable and indexable. Maintain a coherent site architecture. Publish genuinely useful content. Make entities and relationships understandable. Give retrieval systems enough context to determine what your organization knows, sells, does, and represents.

Then look beyond your own website.

Invest in digital PR and legitimate third-party coverage. Give journalists, publishers, customers, experts, and communities reasons to discuss your brand. Build a reputation in the places where your market actually exchanges information.

This matters because AI visibility increasingly depends on information outside the final answer itself. Bing, for example, now exposes AI citation activity and the grounding queries used to retrieve publisher content in Webmaster Tools. Microsoft describes these systems as retrieving and grounding answers against web sources rather than simply generating responses in isolation.

The opportunity is real. But the durable opportunity is not figuring out how to inject your brand into as many AI answers as possible before everyone else catches up.

It is making your brand a credible part of the information environment those systems retrieve from.

There is still a lot we do not know about GEO.

AI search systems are changing quickly. Different engines retrieve different sources. Results vary between prompts and even between repeated runs. While already-retrieved content can influence citation behavior, the evidence for stable, long-term, cross-platform improvements in organic discoverability remains much thinner.

That uncertainty should make teams more careful, not less.

Experiment with GEO. Measure AI citations. Study grounding queries. Learn how your brand appears across different systems. Improve the contextual information surrounding your products and expertise. Follow best practices. There is real work to do here.

But do not confuse exploiting an immature retrieval system with building a strategy.

The easiest GEO tactics to scale will also be the easiest tactics for platforms to identify, discount, or filter once abuse becomes widespread. We have decades of SEO history telling us how that cycle tends to end.

The better approach is less novel: integrate GEO into SEO, digital PR, content, brand, and technical infrastructure. Use what we are learning about AI retrieval to make those disciplines stronger rather than creating a parallel growth-hacking program designed to manufacture signals.

Because if your GEO strategy depends on an AI system being unable to tell the difference between earned authority and manufactured authority, you do not really have a GEO strategy.

You have a spam strategy with a new name.

Transformer: The neural-network architecture underlying most modern LLMs. Transformers use attention mechanisms to determine how different parts of an input relate to one another.

One more thing: AI is only as good as its operator, and if you are reading this newsletter, you’re better than most!

Till next time,

PS: Let me know what you think of this issue, or anything else here: [email protected]