Nobody Knows How Big GEO Is

Every vendor deck in this category opens with a market size slide. I have seen four this quarter, all citing different numbers, all presented with equal confidence.

Here is the short answer: published estimates for 2025 and 2026 put the GEO market anywhere from roughly $390 million to $2.7 billion. Early-2030s forecasts range from about $4.25 billion to $26.85 billion, and cited growth rates run from 13.6% to 50.5%. That is nearly a factor of seven on what the market is worth today and a factor of six on where it is going. The skeptics who say these numbers are unreliable are right, and I would rather say so than quote the one that flatters my industry.

What follows is why the estimates diverge, and what evidence I actually rely on instead.

The spread, laid out

Consider a sample of published research.

One firm valued GEO services at $886 million in 2024, forecasting $7.3 billion by 2031 at 34% compound growth. Another put the global market at $390 million in 2025, reaching $4.25 billion by 2032. A third estimated $848 million in 2025 growing to $19.8 billion by 2034 at 50.5%. A fourth put 2026 at $1.09 billion, reaching $17.1 billion by 2034 at 40.6%. A fifth sized the GEO platform market at $2.70 billion in 2026 with $26.85 billion by 2033, at 13.6%.

Read those growth rates again. The same category is described as compounding at 13.6% and at 50.5%. Those describe different worlds. Over eight years, one is roughly a 3x market and the other is more than a 25x market.

Some forecasts do not even agree with themselves. The fifth report's own start and end figures imply growth of about 39% a year, not 13.6%. The third report's 50.5% rate would take $848 million to roughly $33.6 billion by 2034, not $19.8 billion. Among the reports whose numbers reconcile, growth runs from 34% to 41%.

This is not a rounding difference or normal analyst disagreement. It is a signal that the thing being measured has not been defined.

Why the numbers disagree

Four reasons, none of which require anyone to be dishonest.

The category boundary is unsettled. Does GEO mean software platforms, agency services, or both? Does a Semrush subscription with an AI visibility module count entirely, partly, or not at all? Does an in-house AEO manager's salary count as market spend? Firms answer differently, and each answer moves the number by hundreds of millions. The naming chaos I wrote about earlier has a direct measurement consequence: you cannot size a category the industry has not finished naming.

The revenue base is small enough that individual contracts move percentages. When a market is under $2 billion, a handful of enterprise deals shifts the growth rate materially. Small denominators produce volatile ratios.

The data history is under two years. Forecasting to 2034 from 2024 and 2025 observations means the curve is almost entirely assumption. The difference between 13.6% and 50.5% is not a data disagreement, it is a disagreement about what happens after the period anyone has observed.

And much of this research is commissioned by parties with a position. Not fabricated, but scoped and framed by people who benefit from a particular answer. I hold my own numbers to the same suspicion, which is why I am not publishing a GEO market estimate of my own.

What the skeptics get right here

I want to state their case at full strength, because this is the piece in the series where they land the cleanest hit.

If a category's own analysts cannot agree within a factor of six, the confident market-size slide in every vendor deck is theater. A number presented to two decimal places from a forecast model built on eighteen months of data conveys precision that does not exist. When someone tells you GEO will be a $26.8 billion market, they are reporting the output of an assumption, not a measurement.

Skeptics are also right that hockey-stick TAM projections are a reliable feature of hype cycles. The metaverse had them. NFT infrastructure had them. Producing a large number for a young category is trivially easy and proves nothing.

So if your case for GEO rests on market sizing, you do not have a case. Fortunately, that is not where the evidence is.

What survives scrutiny

Behavior beats forecasts, especially when the forecasts disagree by 6x. Here is the evidence I actually weight, all of which is observed rather than modeled.

Capital has moved, with consequences for the people deploying it. By March 2026, at least $255 million in venture funding had gone into pure-play GEO startups in roughly eighteen months, with Profound accounting for about $155 million of that. Profound raised a $96 million Series C at a $1 billion valuation in February 2026, with Lightspeed leading and Sequoia and Kleiner Perkins participating. It followed that with a $180 million Series D at a $1.8 billion valuation in September 2026. Adobe closed a $1.9 billion acquisition of Semrush in April 2026 on an explicit AI visibility thesis, after diligence, board approval, and regulatory review. HubSpot acquired XFunnel and shipped a product within six months.

Buyers are hiring, with salary bands attached. Citizens Bank posted an Answer Engine Optimization Manager at up to $171,000. SailPoint posted an AEO and GEO Manager at $101,400 to $170,864. Stripe, AWS, Amgen, Experian, and Publicis have all created dedicated roles. A company creating a permanent six-figure position has made a more costly commitment than any analyst writing a forecast.

Practitioners report spending intent directly. Conductor surveyed more than 250 enterprise digital leaders, including CMOs, and found 94% planning to increase AEO investment in 2026, with 97% reporting a positive impact in 2025. That is a survey with its own limitations, but it measures stated behavior rather than extrapolating a curve.

None of that tells you the market is worth $17 billion in 2034. All of it tells you the category is real, which is the question actually under dispute.

How to read a market-size claim

Three questions I ask, and I would apply them to anyone including me.

What exactly is being counted, and does that definition match how you would spend money? A number that bundles agency services with platform subscriptions is not a software TAM.

What is the observation window behind the growth rate? Under three years of history means the forecast is an assumption with a decimal point.

Who commissioned it, and does the scope favor them? Not disqualifying, but it tells you which direction the definitional choices leaned.

For your own planning, none of this should matter much anyway. The relevant question is not how large the category becomes. It is what share of your buyers currently begin their research in an AI assistant, which is measurable inside your own funnel rather than in a published forecast. My data at GrackerAI (disclosure: my company, a GEO platform for B2B SaaS) shows 4 in 10 B2B security buyers now start vendor research inside AI assistants. I would rather you test that against your own pipeline than accept it because I published it.

Here is my question: if you removed every market-size figure from your AI visibility business case and kept only evidence from your own funnel, would the case still hold? If yes, the forecasts were never load-bearing. If no, you were not making an argument, you were citing one.

Frequently Asked Questions

How big is the GEO market in 2026?

Published estimates disagree substantially. Figures for 2025 and 2026 range from roughly $390 million to $2.7 billion depending on the research firm and how the category is defined. Early-2030s forecasts range from about $4.25 billion to $26.85 billion, with compound growth rates cited between 13.6% and 50.5%. That is roughly a six-fold spread on the forecast and nearly seven-fold on current size.

Why do GEO market size estimates vary so much?

Four reasons. First, the category boundary is unsettled, so firms differ on whether to count software platforms, agency services, in-house salaries, or SEO suites with AI modules. Second, the revenue base is small enough that individual enterprise contracts move growth rates. Third, the data history is under two years, making long forecasts largely assumption. Fourth, much of the research is commissioned by parties with a commercial position on the answer.

What evidence for GEO is more reliable than market forecasts?

Observed behavior rather than modeled projections. At least $255 million in venture funding went into pure-play GEO startups in the roughly eighteen months to March 2026, and Adobe closed a $1.9 billion Semrush acquisition in April 2026 after full diligence and regulatory review. Profound reached a $1 billion valuation, and later $1.8 billion. Companies including Citizens Bank, SailPoint, Stripe, and AWS created dedicated roles with salary bands reaching $171,000. Conductor found 94% of 250-plus enterprise digital leaders planning to increase AEO investment in 2026.

How should I evaluate a market size claim for a new category?

Ask what is being counted and whether that definition matches how you would actually spend, since bundling agency services with platform subscriptions does not produce a software TAM. Ask how long the observation window behind the growth rate is, since under three years of history means the forecast is mostly assumption. And ask who commissioned the research and whether the scoping decisions favor them.