Is SEO the Foundation of GEO? Myth, Busted

The one-line myth

“Get SEO right first and GEO follows. GEO is just SEO, upgraded.”

I hear this every week, and it is half true — which is more dangerous than being flat wrong, because it lets you believe you’ve understood the thing.

In the last post, on phased budget allocation, we said “without a foundation, GEO has nothing to stand on.” But “foundation” is doing far too much work as a word. This post pulls it apart, and you’ll find that “is SEO the foundation of GEO” isn’t a yes/no question at all. It’s a question about which mechanism a given AI platform uses to pick you as a candidate.

Three mechanisms. Let’s do all of them.

Why the “foundation” metaphor produces an endless argument

Two camps are currently swinging at each other:

• Camp A: “GEO is an extension of SEO. Do SEO properly and you’re done.”
• Camp B: “GEO and SEO are different things. The foundation metaphor is lazy — don’t be misled.”

The argument never ends because nobody defines “foundation” first. And “foundation” happens to be exactly the kind of word that has one meaning in the dictionary and several completely different meanings the moment you pick up a tool.

Step back: what is a foundation, actually?

By definition, a foundation is the load-bearing layer that holds a structure in place, carries its weight, and stops it being lifted away. Clean definition. It doesn’t change.

But when you actually go and build one, every kind of structure gets something entirely different:

• Pitching a tent: the foundation is a handful of pegs — driven into the ground, guy lines tensioned, and it stands.
• Building a house: the foundation is concrete footings.
• Building a tower: the foundation is piles driven down to bedrock.

Here’s the question that matters: would you walk up to someone pitching a tent and tell them their concrete work is substandard and their foundation is unsound?

No. The tent’s foundation is those pegs. Pour a truckload of concrete under a tent and you haven’t just wasted it — you’ve flattened the tent. The word “foundation” applied to a tent and applied to a tower does not refer to the same thing at all. So “is the foundation important?” is a question that’s always trivially yes. The question worth asking is: for this particular structure, what is the foundation?

Bring the metaphor back to GEO

Asking “is SEO the foundation of GEO” is like asking “is concrete the foundation of a building” — there’s no single answer; it depends what you’re building. And in GEO, different AI surfaces are different structures:

• Some surfaces are tents: the foundation is just the pegs — be crawlable, be discoverable (Mechanism 1). Pouring concrete here does nothing.
• Some surfaces are towers anchored to a search index: the foundation is getting into Google’s retrieval pool (Mechanism 2) — this is where you genuinely do pour concrete, but the concrete isn’t “rank first.”
• Some surfaces are towers already topped out and immovable: they’re built on training data (Mechanism 3) — you can’t touch that foundation, you can only wait for the next building (the next training run).

So Camp A and Camp B argue forever because one is pointing at a tower shouting “obviously you need concrete” and the other is pointing at a tent shouting “concrete? for this?” — both are right, they’re just staring at different structures. Lay the structures out and you don’t have to pick a side.

Below are the three mechanisms, and what the foundation looks like for each.

Mechanism 1: crawlable and discoverable ≠ doing SEO

The lowest bar is this: your content has to be reachable by crawlers and get indexed. If it can’t be fetched, nothing downstream matters — the AI doesn’t know you exist.

But notice how low this bar sits:

• The site is indexable (you haven’t noindex-ed everything, you haven’t blocked crawlers at the door)
• A handful of external links exist, so a crawler can discover that your domain is there
• The content is real text with a sane structure, not a page that only materialises after JavaScript runs

That’s it. This is “don’t be invisible,” not “run a full SEO programme.” A brand-new site with a DR of 0 and no rankings whatsoever clears this bar just fine as long as it’s crawlable and discoverable.

This layer is the tent pegs: get them in and the thing stands; extra concrete is surplus. And plenty of people mistake this minimum bar for “so we’d better max out SEO first” — that’s where the myth is born.

Mechanism 2: AIO — closest to SEO, and “rank = foundation” still doesn’t hold

This is the surface nearest to SEO, and also where the misunderstanding is thickest and most worth clearing up. Many people assume Google’s AI summaries are just the top-ranking pages condensed, so “climb the rankings = get cited.” That’s wrong.

Start with the half that’s genuinely true: Google’s AI Overviews / AI Mode (AIO from here) are built on Google’s own retrieval stack. So “be in Google’s index, and be retrievable for the relevant query” really is a precondition — a rung above Mechanism 1’s “be crawlable.” Of all the AI surfaces, SEO effort pays off most on AIO. Camp A isn’t wrong about that.

But the second half — “rank = foundation” — doesn’t hold, for three reasons:

• AIO uses query fan-out: it splits your question into several sub-queries, runs retrieval on each, and synthesises an answer. The sources it cites are frequently not the top ten for the original query.
• The observed results don’t line up: the URLs AIO cites and the ones you see on the classic blue-link SERP often diverge sharply — the page ranked first goes uncited, while something buried deep, or not in the top ten at all, gets picked. What it’s judging is passage-level relevance (one section of yours happens to answer one sub-question precisely), not a whole-page ranking score.
• “Cited” ≠ “linked” ≠ “ranked first”: AIO may use your content to generate the answer and hand the visible source link to someone else; it may also link you while the substance of the answer came from elsewhere. Being used as material, being linked, and ranking first are three separate problems. Don’t collapse them.

This isn’t theory. I’ve watched clients with beautiful DR/DA numbers take a Google penalty; I’ve watched a domain that AIO used to cite get displaced wholesale by a brand-new site with DR 0. And this is common, not some freak edge case — which nails all three points above: authority scores, search rankings, and AIO citations are three lines running independently.

So even on the surface closest to SEO, the foundation isn’t “rank first” — it’s “get into Google’s retrieval pool, and have a passage precise enough to be picked out.” Set your target here as “climb to number one” and you’ll still miss.

And this is only one of many AI surfaces. Applying this one surface’s rules to every platform is exactly where Camp A goes wrong.

Mechanism 3: an AI that never calls a search tool is immune to your current SEO

At the far end sits the opposite mechanism — and the dividing line isn’t “which country’s model,” it’s whether this AI actually called a search tool while answering you. (People used to split this as “Chinese models vs Western models,” but that cut is out of date: most mainstream models can search now. The difference is whether this particular answer went and searched.)

When an AI doesn’t call a search tool — no live retrieval, answering purely from the knowledge absorbed during training — it is blind to the current state of your site. For answers like that:

• Push your rankings up today and it can’t see it — its knowledge stops at the last training cutoff
• To influence it, your content and your entity have to be present, in sufficient quantity and consistently, in the next training corpus
• That’s slow work. A long game. It can’t be rushed — which is also why real GEO can’t be sold as “results in 24 hours”

One point that gets missed: even when an AI does call a search tool, its search logic isn’t you opening a browser and Googling. It rewrites the query itself, pulls back a few passages, and generates from those — it doesn’t transplant a page ranking. So “I rank first on Google” doesn’t automatically mean “the AI will cite me.” Even AIO, the most search-dependent of the mechanisms, doesn’t hand out answers by rank order — and the purely baked-in-knowledge layer is further from it still.

At this layer, rankings are close to irrelevant. The only thing that helps is the on-site / technical SEO slice — back to Mechanism 1: be crawlable, be indexed, so you at least have a shot at entering the corpus. As for off-site work (backlinks, DR/DA), it has no purchase here at all; and even once you’re in the corpus, a higher ranking adds nothing.

The real problem: for many people, “SEO” means “buy a pile of high-DR/DA backlinks”

Here’s where the actual landmine surfaces.

When somebody says “get SEO right first,” what is the “SEO” in their head? In this market, far too often it doesn’t mean “make the content clear and the entity unambiguous.” It means —

“Go and rack up a pile of high-DR / high-DA backlinks.”

This is commonly filed under “off-site SEO.” And the equation “off-site SEO ≈ buying backlinks” is itself the product of years of black-hat racing to the bottom — off-site SEO originally meant genuine signals from off your site (how others mention and cite you), not links you paid for. The market is saturated with vendors selling backlinks, satellite sites, PBNs (private blog networks) and link swaps, everyone benchmarking against a number from a third-party tool, drifting further off course as they go. A few things need saying plainly:

• DR (Ahrefs) and DA (Moz) are proprietary metrics from third-party tools, not Google ranking scores. Google has said repeatedly that it has no such thing as a “domain authority score.” The number you’re chasing isn’t one the search engine looks at.
• The moment a metric becomes the target, it stops being a good metric (Goodhart’s law). DR/DA were meant to estimate a site’s strength; once the whole market is buying links in order to move DR, the number decouples from real authority and becomes a scoreboard game.
• Useful SEO was never about link counts. It’s about content that can be understood, a clear entity, and real, editorially-chosen citations — people linking you because your content is good, not because you paid.

Narrowing “SEO” to “buy high-DR/DA backlinks” is already outdated and risky within SEO. Carry it over to GEO and it becomes a double disaster.

Why this misunderstanding is doubly bad for GEO

First: AI engines aren’t reading a DR number. What decides whether you get cited is your entity’s authority and consistency, topical consensus, and the diversity of your citation network — whether several mutually independent, credible sources are talking about you. A pile of bought links from a single, heavily interlinked PBN provides “backbone = 1” borrowed authority; it won’t carry candidacy in an AI’s eyes, and spam adjacency can turn it into an outright negative signal.

Second: it makes you spend money in the wrong place and bet your time on the least useful mechanism. You think you’re laying a foundation; you’re stacking a fake wall that the next algorithm sweep will clear. Come the next platform update or the next training run, the bought assets go to zero and only the genuine entity signals survive.

Which is why everything we handle is white-hat: DR 0 on a new site is normal. You get there on real local signals, a clear entity, and solid content, working your way into the candidate pool step by step. No shortcut, but no landmine either.

One table: which mechanism are you stuck on?

An important caveat first: the table below is organised by mechanism, not by platform. In the real world a single platform mixes all three and switches between them per query — ChatGPT sometimes searches and sometimes answers from baked-in knowledge; Google doesn’t return an AIO for every query. So use this table to locate which layer you’re stuck on, not to “look up what to do for the platform I use” — that second use drops you straight back into the one-platform-one-rule trap.

Not one cell in that table answers “go chase DR/DA.” That isn’t a coincidence.

So — is SEO the foundation of GEO?

The precise answer:

• “Crawlable and discoverable” is the minimum bar for all of GEO — that layer counts as foundation, but the bar is low, and it is not the same as a full SEO programme.
• Even on AIO, the surface closest to SEO, the foundation is “in Google’s retrieval pool with a passage precise enough to be picked,” not “rank first” — and being cited, being linked, and ranking first are three different things.
• For an AI that never calls a search tool and answers from baked-in knowledge, your current SEO is nearly irrelevant; what counts is real, long-term presence.
• And at every layer, “chase high DR/DA backlinks” is not the foundation — that’s a myth the market’s race to the bottom warped into place.

Next time someone tells you “get SEO right and GEO follows,” ask them one question back: when you say “SEO,” do you mean making the content and the entity solid, or buying a pile of backlinks? That one question sorts out whether they know what they’re talking about.

Next step: work out which mechanism has you stuck

The question was never “should we do SEO.” It’s which mechanism is currently blocking you — not crawled yet? Not making it into AIO’s candidate material? Or content that never entered the corpus at all? Three blockers, three different fixes, and you can’t guess which from intuition. You measure.

👉 Run a free GEO audit — the report gives you scores across 12 dimensions and tells you which layer your problem sits on, rather than sending you off to buy another batch of backlinks.

Diagnosing the blocker, fixing the right mechanism, tracking it quarterly — that’s what a managed GEO service does, as opposed to selling you a one-off list of backlinks: [email protected]

GEO advanced series. Further reading (introductory): “How to Split Budget Between GEO and SEO”