3 Billion or 28.6 Billion Yuan? China's GEO Market Size Reports Contradict Each Other — Here's What Brands Should Actually Believe

If you're considering a budget line for Generative Engine Optimization, try something first: search for the size of China's GEO market. You'll find a spectacle.

Analysys says the 2026 market is roughly 3 billion yuan. iResearch-adjacent firm iiAnalyse says 4.8 billion. A joint IDC–CCID estimate says 20.3 billion. The China Academy of Information and Communications Technology (CAICT) puts it at 28.6 billion. Dig into earlier press coverage and you'll find versions ranging from 6.2 billion to 94.2 billion — a 30-fold spread between the smallest and largest.

Same country, same year, same industry — four authoritative numbers in open conflict. And as 36kr noted drily today: nearly all of these figures originate from GEO agencies' own press releases. An industry that can't even agree on how big it is is busy selling authority to everyone else.

This article won't tell you which number is right. Instead, here is something more useful: three questions that dismantle any market-size report.

First, Understand: the Numbers Fight Over "Scope," Not Honesty

None of the four estimates is lying. Each quietly drew a different statistical boundary. Grasp this, and you've decoded 90% of industry reports.

The narrow scope (Analysys's 3 billion) counts only "dedicated GEO service fees" — money companies explicitly paid for something labeled GEO. On that accounting, the market really is young: about 250 million yuan in 2025.

The broad scope (CAICT's 28.6 billion) sweeps in everything: companies' in-house content teams, AI marketing tool subscriptions, media spending, and agency operations costs. Two employees producing AI content internally get counted as part of "the GEO market."

The other two estimates — 4.8 billion and 20.3 billion — simply picked positions along this spectrum.

So who's right, 3 billion or 28.6 billion? Both. And neither is useful — because total industry size was never the first metric a brand decision-maker should look at.

Three Things That Are Actually Certain (Where Nobody Argues)

Here's the interesting part: once numbers leave "market size" — the easiest figure to inflate — every dataset suddenly converges. These are the figures worth remembering:

First, the user base has crossed the tipping point. CNNIC's 2026 report on generative AI development shows China passed 700 million generative AI users in the first half of 2026 — over 50% penetration — with 76% of users turning to generative AI to answer questions. iResearch adds that in Q3 2026, generative AI queries already account for 41.2% of Chinese-language search, nearly double the prior year, with a projected 57% share by Q3 2027.

Second, decision contexts are migrating faster than consumer contexts. In high-stakes categories — B2B procurement, healthcare, finance, education — generative search already exceeds 52% of queries. A Harvard Business Review China survey found 66.3% of B2B procurement decision-makers use AI search brand recommendations to shortlist vendors, and 54.7% of consumers say they doubt a brand's prominence if AI search never mentions it.

Third, enterprises are already voting with their feet. An October 2026 EqualOcean survey of 1,300 CMOs found 31.7% have deployed GEO and 44.2% plan to within 12 months — meaning 75.9% of companies have written GEO into their digital marketing strategy.

Notice what these three datasets share: they describe behavior, not money. Behavior is hard to inflate — different researchers using different methods keep reaching the same conclusions. That convergence is the real temperature of this industry.

Three Questions That Dismantle Any Market Report

The next time an agency waves a white paper at you — "GEO is growing 1,100% this year, you'll miss the window if you don't sign now" — return serve with these three questions:

Question one: where does your statistical boundary fall? Dedicated service fees only? Or in-house content, tool subscriptions, and media spend included? The same market differs tenfold across boundaries. If they can't articulate the boundary, discard the number.

Question two: who paid the money, and to whom? Money enterprises pay to external service providers is the real market; internal costs and agency-to-agency subcontracting get double-counted easily. When most of a "market size" comes from agencies' self-reporting, assume sales motivation is baked in.

Question three: what decision of mine does this number actually inform? This is the killer question. Whether the industry is 3 billion or 28.6 billion yuan has almost no causal bearing on whether your brand should invest in GEO this year, or how much. What you should examine is the AI-search query share in your category, your brand's current visibility rate inside AI answers, and what your competitors are doing. Macro numbers manufacture anxiety. Micro data guides action.

A More Honest Framework

If I had to summarize China's 2026 GEO market in one sentence: there is no debate about direction — only about pace; all the noise is in the money figures, all the consensus is in the behavior figures.

The direction is so clear it needs no white paper endorsement — 700 million people are already asking AI, 41% of queries already run through generative search, and three-quarters of enterprises have written it into strategy. The only remaining question is where your industry and your brand sit on the migration curve.

Categories with complex parameters and considered purchases — home appliances, 3C, beauty, mother-and-baby — plus advice-heavy services like automotive aftermarket and travel, are judged by multiple practitioners to be the first GEO battlegrounds to mature. If your brand lives there, the market-size debate is irrelevant. Your campaign has already begun; the only choice is whether to win it or sit it out.

The Bottom Line

Every emerging industry passes through a "number inflation" phase — everyone needs a big-enough market story to win budgets, investment, and clients. That's not a crime. But you need immunity.

Real opportunity never lives in the page numbers of a market-size report. It lives in one plain fact: when your customer asks AI "which brand should I choose in this category," whether the answer includes you — and whether it describes you accurately. Managing that answer is worth more than reading ten white papers.

FAQ

Q: So which agency's GEO market figure should I trust?

A: Trust none completely, dismiss none entirely. Scope matters more than the number: the narrow figure (~3 billion yuan) reflects real money paid to GEO specialists and tells you how mature the service market is; the broad figure (~28.6 billion) includes in-house content and tooling, telling you how hot the overall ecosystem runs. Read together, they reveal the industry's outsourcing ratio — most money still sits in in-house and tools, and professional services are just getting started.

Q: An agency is pressuring me to sign with "the market is growing 1,100% this year." How do I respond?

A: Answer macro anxiety with micro data. Ask for your category's AI-search query share, your brand's current AI visibility baseline, and 12-month effect curves from comparable clients in your industry. Agencies that can produce all three are worth talking to. Agencies that only narrate industry trends will likely deliver industry-trend-grade work.

Q: Will the GEO market really reach 9 billion yuan by 2027?

A: On the narrow scope, Analysys projects roughly 9 billion yuan for 2027 — a tripling in a year. The behavior data behind that projection is real: AI search's query share is heading toward 57%. But remember: a growing market doesn't lift every agency, and it certainly doesn't guarantee returns on every yuan you spend. In the standards era, the industry is stratifying fast — money placed with white-hat providers and money placed with black-hat providers are heading for very different endings.