AI Citation Rate
Report 2026

Citation Depth: The Quality Spectrum

Not all AI citations are equal. A passing mention (depth score 1-2) and a primary recommendation (depth score 8-10) have fundamentally different impacts on brand consideration. Arfadia's RoGEO framework scores every citation on a 1-10 depth scale.

Source: Arfadia RoGEO framework measurements, GEO-active client accounts Q4 2024 - Q1 2026

Factors That Drive AI Citation Probability

• Brand mentions correlation with citation (Ahrefs)r = 0.664
• Semantic completeness - AIO correlationr = 0.87
• 3rd party vs own domain citation probability6.5x higher
• Citing external sources in content (Princeton)+115%
• Statistics addition in content (Princeton ACM KDD)+41%
• Schema markup comprehensive (SearchVIU)+36% AIO rate
• Expert quotation addition (Princeton)+28%
• Backlink profile correlation with AIOr = 0.18 (weak)

Indonesia AI Adoption Landscape

Industry-Specific AI Citation Dynamics (Indonesia)

Four Structural Reasons Indonesian Brands Are Under-Cited

ChatGPT: Training Data is the Foundation

ChatGPT's primary citation source is training data, not a live web index. When Browse mode activates (31% of queries), it supplements with real-time Bing search results. The strategic consequence: building ChatGPT visibility requires a two-track approach operating on different timescales.

What ChatGPT Prioritizes

• Wikipedia presence (47.9% of ChatGPT's top cited sources)Critical
• Sites with 350K+ referring domains (5x more likely cited)High
• Quora presence (26K+ mentions = 3x citation likelihood)High
• Reddit mentions (219K mentions for similar effect)Medium-High

Gemini / AI Mode: Semantic Completeness Wins

Domain authority's correlation with AI Overview citation has dropped from r=0.43 to r=0.18, while semantic completeness has risen to r=0.87. Structurally complete content now outpredicts backlink authority for Gemini citation. A page not in the top 5 organically can still be cited if its content covers the factual landscape more completely.

Key Gemini Data Points

• Semantic completeness correlation with AIO citationr = 0.87
• AIO citations from pages NOT in organic top 547% of all
• AI Overviews from top-10 pages (BrightEdge)Only 17%
• YouTube in AI Overviews (34% growth over 6 months)23% of citations

Perplexity: Your GEO Measurement Starting Point

Perplexity's transparent numbered citations make it the most directly observable and measurable AI platform for GEO. Every citation is numbered and linked. Citation improvements typically appear within 4-8 weeks of content changes - far faster than ChatGPT's training data pipeline.

Perplexity applies four ranking signals: semantic relevance, authority, corroboration (cross-verification across 3+ authoritative sources), and recency. Indonesia-specific note: Perplexity holds 21.18% of Indonesia's AI chatbot traffic vs only 7.88% globally - optimization delivers disproportionate value.

Google AI Overview: SEO as GEO Foundation

AI Overview is the platform with the broadest Indonesian reach - embedded in Google Search, the default for 92.57% of Indonesian internet users. Traditional SEO skills translate most directly into GEO outcomes here. AI Overview eligibility requires the same quality signals as traditional Google ranking, plus content extractability.

AI Overview in Numbers

The B2B Dark Funnel Expansion

The dark funnel - buyer research that occurs outside attribution stack visibility - has expanded dramatically. 70% of B2B buyers complete their research before making first contact with a vendor. As of March 2026, Similarweb data shows 50.2% of all search queries are AI-assisted. B2B organic traffic is down 23% while brand search is up 14% - the gap represents buyers researching in Perplexity and AI Overviews that analytics cannot see.

94% of buyers use AI during the purchasing process, 97% of research is anonymous, and buyers evaluate an average of 5.1 vendors - but 76% of those vendors are already known to them before the search begins. A B2B brand not cited by AI is invisible to the majority of its potential buyers at the most critical stage of their research journey.

Gartner's Long-Term Vision

Gartner predicts that by 2028, 90% of B2B buying will be AI-agent intermediated, with $15 trillion flowing through AI agent exchanges. 60% of brands will use agentic AI for one-to-one customer interactions. This makes the 2026 GEO investment window not just an optimization decision but a fundamental repositioning: from "findable via search" to "recommended by AI agents."

Revenue Modeling: The GEO Gap

Assumptions: 100 impressions per citation, 10% CTR, 2.8% conversion rate with 4.4x AI-referred premium applied, AOV Rp 350K. Illustrative model - results vary by industry and starting authority.

Step-by-Step Citation Audit

1. 1

   Build Your Query Library (30-50 queries)

   Organize across 4 intent categories: awareness ("what is [service] in Indonesia"), consideration ("best [category] company Indonesia"), comparison ("[Brand A] vs [Brand B]"), and purchase intent ("recommended [service] for [need]"). Include both English and Bahasa Indonesia versions. Run a minimum of 5 repetitions per query per platform - individual runs are not statistically reliable due to AI output variation.

2. 2

   Run Baseline Across All 4 Platforms (Week 1)

   Test each query in ChatGPT, Gemini, Perplexity, and Google (for AI Overview). Record per run: brand present (yes/no), citation depth score (1-10), citation position relative to competitors, and which competitors appear instead of you. This is your competitive gap document.

3. 3

   Calculate Citation Rate and Share of AI Voice

   Citation Rate = (queries where brand appears / total queries tested) x 100 per platform. Share of AI Voice = your citations / (your citations + top-3 competitor citations). Industry non-GEO baseline is approximately 41 total monthly citations across all platforms. Perplexity shows fastest results from optimization (4-8 weeks); ChatGPT shows slowest (4-8 months).

4. 4

   Map AI-Attributed Traffic in GA4

   ChatGPT strips referrer data from paid accounts. AI Mode has 93% zero-click. Most AI-influenced traffic is miscategorized as "direct" in standard analytics. Build a dedicated GA4 segment for known AI platform referral domains (chatgpt.com, perplexity.ai, gemini.google.com, etc.) to capture the measurable fraction.

5. 5

   Monthly Reporting Cadence

   Report to leadership monthly: total citations (all platforms), citation rate by platform, average reference depth score, Share of AI Voice vs top 3 competitors, and trend direction. Connect to revenue: track branded search volume growth in Google Search Console as a proxy for AI-driven brand awareness. Given 40-60% monthly source rotation, quarterly snapshots are insufficient.

Measurement Tools Landscape (2026)

Key Metrics Dashboard

Platform Evolution Forecasts

Bahasa Indonesia AI Parity Timeline

• 2026: Mainstream consumer topics approach near-parity. AI Overviews in Bahasa Indonesia become routine for informational queries.
• 2027: Specialist B2B and technical topics improve substantially. "Sahabat-AI" and sovereign AI programs contribute Bahasa Indonesia training data.
• 2028: Full comprehension parity for mainstream topics achievable if Indonesian brands have built deep content libraries during the 2025-2027 window.

The Agentic AI Future

The most significant medium-term disruption is the shift from AI as an information tool to AI as an action-taking agent. Over 90% of Southeast Asian firms plan to experiment with agentic AI by end of 2026. Gartner projects that by 2028, 60% of brands will use agentic AI for one-to-one customer interactions, and $15 trillion will flow through AI agent exchanges.

For Indonesian brands, this means the optimization target is shifting from "being mentioned in an AI answer" to "being executable by an AI agent." Brands that implement rich schema, structured pricing, and API-accessible product information today are building the infrastructure for agent-accessible discovery tomorrow.