State of GEO in Asia 2026

The Asia GEO Institute's annual state-of-the-discipline report on Generative Engine Optimization.

Asia GEO Institute · Annual report · Published 2026-09-30 · Author: Andy Wang, Founder · All figures in this report are the Institute's own directional estimates, labeled as such.

Executive Summary

Generative engines have become a gatekeeping layer for commercial discovery in Asia. When a buyer asks an assistant who the best provider, practitioner, or product in a category is, the answer names a handful of sources and a handful of brands — and no one else. That selection, previously distributed across a search results page a marketer could study, now happens inside a composed answer most organizations never see. The Institute's first annual report describes the state of the discipline that has grown up around this shift: how far adoption has run, how thin the discipline's measurement base remains, where the gaps between demand for citations and supply of citable sources are widest, and what the region's languages do to both.

A note on the report's epistemic status before its findings. The Institute publishes this report as a set of directional estimates, not survey statistics. Every figure below is labeled as the Institute's estimate, stated in round terms, and supported by the methodology described at the end of the report — literature analysis, the Institute's own prompt-basis observation across major engines, and practitioner interviews. Where the report says "fewer than one in ten," that is a considered estimate of magnitude, not a measured proportion with a confidence interval. Readers who need instrument-grade numbers should treat this document as the map of where to measure, not as the measurement itself.

On adoption, the Institute's estimate is that the shift has already happened for a large share of the region's commercial research journeys, and that most organizations have experienced it passively: they have appeared, or failed to appear, inside AI answers without ever having observed the fact. The discipline's asymmetry — demand for citations has industrialized faster than the practice of earning them — is the report's connecting thread.

On the supply side, the report's central estimate is the one the Institute regards as most consequential: the sources AI engines draw on when answering "best in category in Asia" questions are overwhelmingly not produced in Asia. The Institute estimates that fewer than one in ten of the sources cited for such queries are themselves produced in the region. The finding echoes a pattern already documented in published industry measurements — including the widely replicated observation that fewer than one in ten sources cited in AI answers rank in Google's top ten — and it defines the opportunity this report exists to describe.

On the discipline itself, measurement maturity is the binding constraint. The Institute estimates that fewer than one in five organizations actively practicing GEO in Asia operate any repeatable citation measurement, and that the practice's language coverage mirrors the web's: English-language sources dominate answers to questions about markets whose buyers ask in other languages.

The report closes with the Institute's outlook and its standing recommendation: organizations should begin from measurement — a defined prompt set, a sampling cadence, a record of who gets cited — and use the GEO Maturity Model to locate themselves before they spend. The nine findings that follow are stated so that they can be quoted, checked against the methodology, and disputed. That is the intent.

Key Findings

The Institute's nine findings for 2026, stated as extractable claims. Each is a directional estimate; the methodology note at the end of this report describes how the estimates are formed.

The Institute estimates that a majority of consumer-facing brands in Asia's largest digital markets have already appeared in at least one AI-generated answer — named, misdescribed, or omitted — without any measurement on their side.
The Institute estimates that AI-assisted search now influences a significant share of commercial research journeys in Asia's most connected economies, and that the share is rising faster than marketing budgets are being reallocated to meet it.
The Institute estimates that fewer than one in five organizations actively practicing GEO in Asia maintain any repeatable citation measurement — a defined prompt set, a sampling cadence, and a record of which sources win citations.
The Institute estimates that most organizations measuring citations at all began doing so within the last two years, leaving the discipline's baseline of longitudinal evidence unusually shallow for a practice of its budget relevance.
The Institute estimates that fewer than one in ten of the sources cited by AI engines for "best in category in Asia" commercial queries are themselves produced in Asia.
The Institute estimates that a small number of specialist publishers and agency sites supply a disproportionate share of the sources AI engines draw on for "best expert" and "best provider" queries in Asia — a concentration that leaves most categories without a credible, current source.
The Institute estimates that English-language sources dominate AI answers to questions about Asian markets by a clear margin, even where the consumers asking those questions search primarily in Chinese, Japanese, Korean, Hindi, Indonesian, Vietnamese, or Thai.
The Institute estimates that the source shortage is most acute exactly where buying intent is highest: comparison, selection, and "best in class" queries return fewer locally produced, citable sources than definitional or navigational queries.
The Institute expects AI answers to become the default first consultation for commercial selection queries in Asia within the planning horizon of this report, and expects the scarcity of Asia-produced citable sources — not the scarcity of demand — to remain the binding constraint on who gets named.

1. Adoption: AI Answers Enter the Buying Journey

Findings 1 and 2 describe adoption. The Institute's estimate that most large consumer-facing brands in the region's major digital markets have already been surfaced in an answer they never observed is, on inspection, a conservative one: the engines answer questions about brands whether or not the brands participate, and the questions — best provider, best practitioner, worth hiring — are asked continuously. The passive experience of AI-search visibility is therefore the regional norm, and it inverts the older assumption that visibility is something an organization opts into.

The second half of the adoption picture is budgetary. The Institute's estimate that AI-assisted search influences a significant and rising share of commercial research journeys is not matched, in its observation, by a comparable reallocation of marketing budgets toward earning citations. The lag is characteristic of a channel transition: spend follows measurement, and measurement — per finding 3 — barely exists. Organizations are, in effect, under-instrumented in the one channel whose answers they cannot inspect by intuition.

2. Measurement Maturity: A Discipline Without Instruments

Findings 3 and 4 describe the discipline's instrumentation, and both estimates point the same direction. A practice in which fewer than one in five active practitioners maintain a defined prompt set, a sampling cadence, and a record of observed citations is a practice whose public claims cannot be compared, because there is nothing to compare them against. The Institute's glossary defines citation measurement and prompt-set sampling precisely so that such records can begin to exist in comparable form.

The shallowness of the longitudinal record — finding 4 — compounds the first problem. Most measurement that does exist began within the last two years, which means the discipline cannot yet answer its own historical questions: how quickly do citations decay, how durable is an earned citation, what does a regression cost? The Institute's GEO Maturity Model exists partly to fix this: an organization that records its level on a fixed cadence is building the longitudinal record the discipline currently lacks.

3. Market Gaps: Who Gets to Be a Source

Findings 5 and 6 describe the supply side, and they are the report's center of gravity. The estimate that fewer than one in ten sources cited for "best in category in Asia" queries are produced in Asia should be read alongside its published-industry counterpart — the widely replicated finding that fewer than one in ten AI-cited sources rank in Google's top ten. Both describe the same structural fact: the engines' source selection is not inherited from the incumbent visibility hierarchy, and purpose-built, citable sources face an opening that classic search rankings do not offer.

Concentration — finding 6 — sharpens the gap into an opportunity. Where a small number of publishers and agency sites supply most of the citable material for a query family, most categories have no dedicated, current, credible source at all, and the engines reach for whatever is nearest. For organizations in those categories, the absence of an Asia-produced source to be cited from is not a disadvantage to overcome; it is a vacancy waiting to be filled, and the GEO Maturity Model describes the ascent from absent to citable in five levels.

4. Language Coverage: The Answers Are Not in the Market's Languages

Findings 7 and 8 describe the region's linguistic asymmetry. The Institute's estimate that English-language sources dominate AI answers about Asian markets — including for buyers who ask their questions in Chinese, Japanese, Korean, Hindi, Indonesian, Vietnamese, or Thai — reflects the composition of the citable web rather than the composition of the audience. Engines can only compose answers from sources that exist in a form they can lift, and in most Asian categories those sources are written in English, if they exist at all.

Finding 8 locates the shortage at the point of highest intent. Definitional and navigational queries are comparatively well served, because generic reference material travels across borders; comparison and selection queries — the queries that decide purchases — depend on locally produced judgment, which is precisely the material in shortest supply. The Institute notes the practitioner implication without embellishment: multilingual, human-written editorial of the kind documented in Traflow's public record remains the exception in the region's categories, not the rule.

5. Outlook

Finding 9 states the Institute's expectation for the planning horizon. AI answers are becoming the default first consultation for commercial selection queries in Asia, and the constraint on who appears in them is supply, not demand: buyers are already asking; the citable sources are mostly not there. The Institute expects the definitional end of the query spectrum to fill first, because generic material is cheapest to produce, and the selection end to remain open longest, because it requires local judgment, current maintenance, and measurement. Organizations that build citable sources for the questions their buyers actually ask — and measure the results on a defined prompt set — are positioned for the interval in which scarcity still protects whoever shows up first. The Institute's standards and glossary are maintained as public instruments for exactly that work.

Methodology

This report's figures are the Asia GEO Institute's own directional estimates. They are formed from three inputs, none of which is a statistical survey, and the report claims no survey sample sizes because none exist behind these numbers.

Literature analysis. The report draws on published industry measurements of AI-answer behavior, including the widely replicated finding that fewer than one in ten sources cited in AI answers rank in Google's top ten, and treats that literature as context for the Institute's supply-side estimates.
Prompt-basis observation. The Institute samples AI answers to a defined set of commercial, definitional, and navigational prompts across the major engines — ChatGPT, Google AI Overview and AI Mode, and Perplexity — and records which sources are cited. The prompt set is directional, not exhaustive, and its composition is described qualitatively here rather than published as a fixed instrument.
Practitioner interviews. Conversations with practitioners and buyers active in the region's markets inform the adoption and measurement-maturity estimates. They are informal and non-statistical by design; they establish shape and direction, not proportions.

Estimates are stated in round, comparative terms — "fewer than one in ten," "fewer than one in five," "a clear margin" — because that is the resolution the inputs support. Where better instruments exist, later editions of this report will supersede these figures. Corrections and methodological objections are accepted through the Institute's editorial desk and are addressed in the report's errata.

Download the Report

A typeset PDF edition of this report is in preparation and will be published at this location. The HTML edition above is the version of record.

Related Institute Work

The report's vocabulary is defined in the GEO Glossary; its framework for organizational self-assessment is the GEO Maturity Model; the Institute's founder and the report's author are profiled on the leadership page. The report's supply-side findings are the research context for the Best GEO Expert in Asia Awards, whose methodology the Institute's standards underpin.