AI-assisted discovery is already material in Indonesia, but the evidence needs careful reading. Jakpat found that 71% of its 1,423 respondents had used AI; among that AI-user subgroup, 74% selected ChatGPT.
For two decades, Indonesian digital marketing optimized for one destination: Google's blue links. That assumption quietly broke over the last eighteen months. Here is the data, all of it public and sourced, on how Indonesians actually discover brands now.
The adoption numbers
The infrastructure shift: AI search speaks Bahasa now
Google's AI Overviews began appearing for Indonesian users in October 2024. Google announced AI Mode in Indonesia and Bahasa Indonesia on September 9, 2025. The practical implication is that Indonesian-language pages now serve both ordinary Search and Google's AI features, under the same published eligibility requirements.
What actually gets a brand recommended by AI
Large-scale citation studies through 2025–2026 are remarkably consistent about where AI answers source their recommendations:
- "Best X" list pages dominate. In Ahrefs' study of 26,283 cited URLs, "best X" list pages made up 43.8% of cited page types for recommendation queries, covering both third-party lists and brands' own; 35% of those list pages came from low-authority domains. Small publishers win here.
- Listings appeared frequently in one large vendor study.Yext reported that listings made up 42% of citations across its sampled verticals. That is not the same as saying 42% of every local-service answer comes from directories.
- Brand mentions beat backlinks. Web mentions correlate with AI visibility at r≈0.66–0.71; backlinks at just r≈0.1–0.22. YouTube presence is the single strongest correlated signal (r≈0.737) (Ahrefs, 75,000 brands).
- Freshness is rewarded. ~65% of AI-cited pages were published within the past year (Seer Interactive, 5,000+ URLs).
- Language matching matters. AI engines prefer citing sources in the query's language. In Weglot's study of 1.3M+ citations, translated sites saw up to 327% more AI Overview visibility in non-primary-language queries. For Indonesia: Bahasa queries get answered from Bahasa sources.
What this means for Indonesian marketers
The playbook changes in three concrete ways:
- Check more than one search index. OpenAI lists Bing among ChatGPT Search's third-party providers, so Bing indexation can matter alongside Google. It is not documented as the pipeline for every response. Rendering matters too: many major AI crawlers, including OpenAI's, ClaudeBot, Meta's, ByteDance's, and PerplexityBot, do not currently render JavaScript (Vercel/MERJ crawler study); Gemini benefits from Googlebot's rendering and Applebot can render JS. JavaScript-only content risks being missed by most AI crawlers.
- Invest off-site. The citations that drive AI recommendations live on directories, editorial lists, review platforms, YouTube, and communities, not your homepage. Earned presence is the new link building.
- Publish in both languages. English content wins English queries; Bahasa content wins Bahasa queries. Mixed-language pages underperform in both pools.
Our GEO services in Indonesia combine standard search eligibility, evidence-led content, source strategy, and controlled measurement. The free Epilog GEO readiness toolchecks technical preparation and includes a clearly labelled experimental single-model observation. It does not measure cross-engine visibility.
Frequently Asked Questions
Which AI assistant do Indonesians use most?
Jakpat's 2025 survey, as reported by DailySocial, found that 71% of all respondents had used AI. Among the respondents who used AI, 74% selected ChatGPT. Those figures use different denominators and should not be combined.
Is Google AI Mode available in Bahasa Indonesia?
Yes. Google announced AI Mode in Indonesia and Bahasa Indonesia on September 9, 2025. Standard Google Search eligibility applies; Google does not require special AI schema or an AI text file.
What should brands do about AI search in Indonesia?
Start with ordinary search eligibility on Google and Bing, useful language-matched pages, accurate entity information, and legitimate third-party corroboration. Then measure what selected AI products visibly cite or mention through a repeated, logged panel.