Google Says GEO Is Just SEO. Bing Says It's a New Problem. Both Are Right.

Google and Bing both published their positions on AI search optimization in May 2026 — and on the surface they seem to contradict each other. Google essentially told everyone to calm down: GEO is just SEO, the fundamentals haven't changed, stop buying into AI-specific hacks. Bing took the opposite stance: grounding for AI answers is a fundamentally different problem that needs different thinking and different measurement. Read both pieces and you could be forgiven for walking away confused about which one is right. They both are. They're just describing two different jobs.

What Google Actually Said: Calm Down, It's Still SEO

Google's message was disciplined and consistent with everything they've said about their AI features since launch. AI Overviews and AI Mode are not a separate search product running on separate signals. They run on the same core index via retrieval-augmented generation. The pages that appear in AI Overviews are, broadly, the pages that Google already trusts for those queries in organic search. If you rank well and your content is genuinely useful, you have a reasonable shot at AI Overview inclusion. If you don't rank well, no amount of GEO-specific optimization is going to change that.

The tactical guidance was pointed. Google called out a specific list of things practitioners can stop worrying about for AI features: llms.txt files, chunking content for AI readability, rewriting pages specifically for AI, chasing inauthentic mentions, and obsessing over structured data as a GEO lever. They also flagged scaled page generation — creating pages specifically to capture every AI query variation — as a spam policy violation, not a legitimate optimization strategy.

What they did emphasize was the quality of the content itself: unique perspective, first-hand experience, genuine expert insight. Not recycled listicles. Not commodity content that can be found in ten other places. The heuristic they offered was simple enough to be genuinely useful: would your actual visitors find this page satisfying? If yes, you're on the right track. If not, no structured data markup is going to fix that.

Google's broader message was a direct challenge to the GEO consulting ecosystem that had grown up around AI Overviews: most of what's being sold as AI-specific optimization doesn't actually move AI Overviews, because AI Overviews don't run on different signals than organic search. The brands that are winning in AI Overviews are, by and large, the brands that were already winning in organic search — with better content than their competitors.

What Bing Actually Said: The Measurement Problem Is New

Bing's piece was more technical and more philosophical. They weren't arguing with Google's content advice — they were making a different point entirely, about what happens to information once the search system has retrieved it and handed it to an AI model. That step, grounding, is where Bing argues a genuinely new optimization problem exists, even if the underlying crawling infrastructure is shared.

The framing they used is worth sitting with. Traditional search asks: which pages should a user visit? Grounding asks: what information can an AI system responsibly use to construct an answer? These are different questions. Answering the first requires ranking pages by relevance and authority. Answering the second requires evaluating individual facts for provenance, freshness, and conflicting evidence — and deciding whether to include them in an answer or abstain.

The unit of value shift is the key insight. In traditional search, the unit is a document — a page that earns a rank position. In grounding, the unit is a groundable fact: a specific claim with clear provenance that an AI system can responsibly attribute and incorporate into a synthesized answer. That shift has downstream consequences that make grounding optimization meaningfully different from ranking optimization, even when both systems share the same index.

The freshness point deserves emphasis because it illustrates the stakes. In traditional search, stale content drops in ranking — that's a traffic problem. In grounding, stale content produces a wrong answer — that's a trust and accuracy problem. An AI system that cites an outdated statistic, a superseded product specification, or a changed regulatory position isn't just underperforming; it's generating misinformation. That's why Bing argues that measuring and optimizing for grounding quality is a genuinely new problem, not just a variant of the ranking problem.

Bing's practical conclusion was that they are building a separate optimization layer on top of search, not replacing it. The crawling infrastructure is shared. The indexing is shared. But the quality signals that determine whether a fact is suitable for grounding are different from the signals that determine whether a page deserves a high rank. Measuring citation quality — are AI systems using your content accurately, attributing it correctly, and grounding their claims responsibly? — is a problem that requires new tooling.

Why They're Both Right (They're Describing Different Jobs)

The apparent contradiction between Google and Bing dissolves when you recognize that they're operating at different layers of the same stack. Google is talking about the input side: how do you produce content that deserves to be retrieved and cited? Their answer — make it genuinely good, don't game it — is correct. The content signals that make pages worthy of AI Overview inclusion are largely the same signals that make pages worthy of high organic rankings, because both use the same index and most of the same trust signals.

Bing is talking about the output side: once content is in the system, how do you measure whether AI systems are using it correctly, attributing it accurately, and grounding their claims responsibly? Their answer — this is a new measurement problem that needs new thinking — is also correct. The metrics that matter for grounding quality (citation accuracy, fact freshness, source attribution, abstention rates) don't exist in traditional SEO analytics. Google Search Console doesn't tell you whether an AI system cited your content accurately or whether it was used to support a claim your page never actually makes.

What This Means Practically: The Engine and the Instrumentation

Working with SaaS brands through this shift, the framing that has proven most useful is this: traditional SEO is the engine. GEO is the instrumentation layer on top.

The engine still runs on the same fuel — real content, real links, genuine expertise. Google is right that you cannot shortcut your way into AI Overview inclusion by tweaking your llms.txt or chunking your paragraphs differently. AI Overviews select from the same candidate pool that organic rankings do, which means the only reliable path to AI Overview presence is the same path as always: be the genuinely best answer to the query.

But the instrumentation layer is new and necessary. You cannot tell from your existing analytics whether your content is being cited accurately in AI answers, which competitors are being recommended instead of you for your core keywords, or how each AI model characterizes what your brand actually does. Google Search Console tells you when an AI Overview appeared and that you received an impression — it doesn't tell you whether your brand was in the answer. That gap is what Bing is describing as the new measurement problem, and it's a real one.

The mistake most brands are making right now is one of two things: either they are ignoring the instrumentation layer entirely (no visibility into how AI systems are using their content), or they are buying GEO-specific tactics that have no plausible mechanism for improving organic authority and therefore have no plausible mechanism for improving AI inclusion. Both are mistakes. The first leaves you blind to a channel your buyers are increasingly using. The second wastes budget on optimization that doesn't address the actual problem.

The Budget Question

One practical implication of this framing is budget allocation. The SEO budget and the GEO measurement budget are doing different jobs and should be evaluated against different outcomes. Conflating them — either by treating GEO measurement spend as a replacement for SEO work, or by evaluating GEO tracking tools on whether they improve organic rankings — produces bad ROI analysis in both directions.

SEO budget improves the content and authority that feeds the AI retrieval candidate pool. GEO measurement budget tells you whether that content is converting into AI citations and which gaps remain. The second doesn't substitute for the first. A brand with weak content and poor organic authority will have low AI citation rates, and no amount of citation tracking will change that. But a brand with strong organic SEO and no citation tracking is flying blind in a channel that their buyers may be using more than any other.

The instrumentation layer is not expensive relative to the organic program it tracks. Bingly can tell you your citation rate across ChatGPT, Claude, Gemini, and Perplexity — the data Bing says is genuinely new — for a fraction of what most brands spend on a single piece of content. The case for adding that instrumentation is not "it will improve your rankings." The case is that you currently have zero visibility into a channel where your buyers are increasingly making decisions, and for the cost of a modest tool subscription you can see exactly where you stand.

The Timing Is Not Coincidental

Google and Bing both publishing their positions on AI search in the same month is not a coincidence. The GEO consulting market had grown large enough, and the quality of advice circulating in it had degraded enough, that both search engines had a stake in setting the record straight. Google was seeing brands waste effort on tactics that couldn't possibly help (and in some cases, like scaled AI query pages, actively violated spam policy). Bing was watching the measurement conversation stagnate on the wrong questions — people asking "how do I rank in AI search?" when the more important question is "how do I know what AI systems are saying about me?"

Google killing the shortcut tactics is good for the industry. It forces brands back to the fundamentals — producing genuinely better content than competitors — which is where durable search visibility has always come from. Bing building the measurement layer is also good for the industry. It means the question of "how is my brand being represented in AI answers?" is being treated as a serious, solvable problem with rigorous tooling, not hand-waved away as "just do good SEO and hope for the best."

Read together, the two pieces describe a mature, coherent approach to AI search: earn your place in the candidate pool with genuinely good content and real authority (Google's domain), then measure and improve how that content performs once it's inside the AI answer layer (Bing's domain, and the problem Bingly is built to solve). Those are two different jobs. Treating them as one is where most AI search budget gets wasted.

Frequently Asked Questions