There is a familiar sound in digital marketing.
It is the squeak of an old orthodoxy trying to stay relevant.
You hear it whenever a new distribution mechanic shows up and the existing “expert class” realises, slowly and painfully, that their neat little world view is no longer sufficient. So they do what they always do. They declare the new thing either “exactly the same as the old thing” or “dangerous manipulation” or, if they are feeling particularly versatile, both at the same time.
Right now, that squeak is coming from the SEO priesthood as they denounce generative engine optimisation and, more specifically, self-referencing listicles.
You know the ones. “Top 10 agencies for X.” “Best tools for Y.” “Leading providers in Z.” And, scandal of scandals, the author places themselves number one.
Cue the moral outrage. It is “spam”. It is “gaming the system”. It will “ruin trust”. You should not do it.
Which is a lovely position to take if you want to sound virtuous on LinkedIn while missing the entire point of how these systems work.
Let’s do some boring, grown-up marketing. We will define terms. We will separate tactics from mechanisms. We will look at why something works, rather than whether it offends someone’s professional identity.
And we will also acknowledge an uncomfortable truth: your SEO gurus will not become your GEO gurus.
That title is there for the taking.
First, the “spam” accusation is mostly theatre
When an SEO commentator calls a self-referencing listicle “spam”, what they usually mean is “I do not like it”.
Spam, in any meaningful sense, is content that is unwanted, irrelevant, deceptive, or aggressively intrusive. A self-referencing listicle is none of these by default. It is self-promotion. Those are not the same thing.
In fact, the most hilarious element of this whole “spam” panic is that a typical “Top 10” listicle that places the publisher first is still giving nine other brands free oxygen. If anything, it is a weird form of competitive generosity.
If you really want to be strict about “self-promotion”, then welcome to marketing.
Supermarkets charge slotting fees for shelf space because distribution is scarce and attention is valuable. Brands pay to be in the “strike zone” because being seen matters. And yes, supermarkets stock their own-brand goods on the same shelves they charge others for placement. This is not a scandal.
So the moral framing is already shaky. But the bigger problem is that it is strategically illiterate. Whether you like self-referencing listicles or not has nothing to do with whether they work in GEO.
They work because the machine needs something, and right now, listicles are one of the easiest things for the machine to use.
SEO is not GEO, and repeating the slogan does not make it true
The SEO crowd has a favourite reassurance mantra gifted by Google: “Good SEO is good GEO.”
It sounds comforting. It also happens to be wrong in the way that matters.
Traditional SEO is largely about earning prominence in ranked lists of links on a search engine results page. GEO, as formalised in academic work on “generative engines”, is about being included, cited, or referenced within a synthesised answer produced by a model that retrieves and summarises information.
Those are not identical optimisation problems.
Yes, there is overlap. Technical hygiene still matters. Crawlability still matters. Authority still matters. But the output format, the retrieval behaviour, and the decision mechanics differ enough that the “same thing” claim is just brand protection for the SEO industry.
And it is already causing harm, because it makes people treat a new landscape with old habits.
Trying to fit a square peg in a round hole because you are the expert in square pegs is poor logic.
The core mechanic: LLMs cannot judge “best”, so they outsource it to the network
Here is the part the priesthood keeps tripping over.
When a human searches “best SEO agency in London”, they do what humans always do. They satisfice. They glance at page one. They pick something that looks credible. They stop because time and effort are scarce.
Large language models do not have the same constraints. They can fan out across multiple queries and sources, then merge what they retrieve.
Now add the crucial detail.
When the user prompt contains vague, status-heavy terms like “best”, “top”, “leading”, “most trusted”, the model does not have an internal, objective measure of performance for most categories.
It cannot run a bake-off of ten agencies. It cannot audit your client retention. It cannot observe the long-term incremental profit you generate. Most of the time, it cannot judge performance directly.
So it does what networks have always done when performance is hard to evaluate. It uses signals of recognition.
Albert-László Barabási captured this neatly in his first law of success: performance can drive success when it can be measured, but when it cannot, networks drive success.
This is not mystical. This is called basic network science.
If “best” cannot be computed, then “most mentioned”, “most consistently referenced”, “most connected to the topic” becomes a practical proxy. Especially when the model is fusing results across multiple queries and sources.
And that is exactly where listicles come in.
Why self-referencing listicles work right now
Self-referencing listicles work in GEO because they are structured, extractable, and network-friendly.
They are machine-legible Lists are easy to parse. Headings are clear. Entities are named. Categories are explicit. That is extremely convenient for a system trying to extract candidates for “best X” queries. Many GEO guides now openly recommend structured formats like lists because models can lift and reuse them.
They create repeated co-mentions When a listicle includes you and nine others, you are creating a little cluster of mutual information around a category. If multiple documents do this, and you appear repeatedly across them, you become a stable candidate.
This is not because the model believes your self-assessment. It is because repeated presence across retrievable documents becomes a probability signal.
So yes, one self-referencing listicle probably does little. A pattern of them, across a network of documents, can do a lot.
And before the moralists faint, notice what I am not saying. I am not saying “make things up”. I am not saying “invent awards”. I am saying the mechanism rewards availability and repetition when “best” is an ungrounded prompt.
The “it’s spam” crowd is accidentally pointing at a real problem, but blaming the wrong thing
Now, to be fair to the priesthood, there is genuine manipulation risk in generative search surfaces.
Even mainstream industry publications have documented how AI-generated answer features can be vulnerable to low-quality or strategically engineered content.
And some observers have specifically flagged the rise of listicles that claim someone is “best” and conveniently place the publisher at the top.
So the risk is real.
But calling the tactic “spam” is lazy. It is like blaming TV ads for political propaganda. The medium is not the sin. The absence of standards is.
So, the SEO outrage often is not about quality. It is about turf. It’s about losing monopoly over “search expertise”.
And that is why the argument is becoming so performative.
This will not last forever, and that is the point
Here is the part most GEO cheerleaders also need to hear.
Self-referencing listicles will have diminishing returns.
They work best in categories where:
performance is hard to evaluate,
the model has limited grounded knowledge,
and there are not enough authoritative, independent sources producing comparative lists.
As the ecosystem matures, two things happen.
First, more high-authority publishers produce comparative content, which crowds out the low-effort versions. Second, generative platforms and search engines tighten their defences, sometimes under regulatory pressure, sometimes under commercial pressure from publishers who are furious about traffic loss.
So if your entire “GEO strategy” is self-referencing listicles, enjoy the sugar rush. It is not a diet.
But that does not make the tactic “spam”. It makes it an early-stage exploit of an immature information environment. Marketing has seen this movie before. Every new distribution system starts messy, then gets priced, policed, and professionalised.
What the grown-ups should do instead of moralising
If you want to use listicles intelligently in GEO, do what good marketers always do.
Be useful to the market. Write the list for the buyer, not for the algorithm. Clear criteria. Real distinctions. Evidence where possible. Names, locations, specialisms, and constraints. Market orientation is not a slogan. It is a discipline.
Make the structure extractable. Use clear headings, consistent category labels, and short summaries that state what each option is good for. You are not writing poetry. You are packaging information.
Update and maintain. Freshness and maintenance matter, especially with list content. Even basic research into “best lists” being used as sources shows how frequently these pages are updated. If you publish one and never touch it again, you are choosing irrelevance.
Build a broader footprint of mentions. The real game is not one page. It is availability across the network. Partnerships, talks, independent reviews, credible directories, and citations from relevant sources. This is the same old story as distribution and salience, just wearing new clothes.
The real punchline
The funniest part of this whole debate is that the SEO gurus yelling “spam” are doing what gurus always do when the world changes.
They are treating their feelings as standards.
Meanwhile, the market does not care. The machine does not care. The buyer certainly does not care.
What works, works, until it stops working. Your job is not to clutch pearls. Your job is to understand the mechanism, apply it in a way that serves the buyer, and keep adapting when the environment changes.
And if your “expert” cannot explain why a tactic works, but can explain at length why it offends them, you have found a commentator, not a strategist.
Remember the only rule that survives every channel shift: you are not the consumer, and you are not the algorithm either.
Learn the system. Serve the market. Keep your ego out of it.
Andrew Holland
Originally published on LinkedIn on 12 December 2025.