LLM citations in 2026: what changed in a year of watching.

A year ago we published a piece on which LLMs cite which sources. This is the follow-up. Same methodology, same client set, 12 more months of data. The headline: the landscape has consolidated faster than the industry commentary suggests.

Citation share by LLM (B2B queries, Q1 2026)

• ChatGPT (with Search): ~41% of tracked citation traffic. Up from 28% in Q1 2025.

• Perplexity: ~22%. Up from 19%. Growing but losing share to ChatGPT.

• Google AI Overviews: ~18%. Down from 24%. Fewer AIOs on commercial queries after the Feb 2026 core update.

• Gemini app: ~11%. Up from 6%. Meaningful for the first time.

• Claude.ai (web-enabled): ~5%. Up from ~1%. Still small.

• Others: ~3%.

What changed in 12 months

1. ChatGPT ate the answer-engine category. The December 2024 Search launch and free availability for all users shifted the default answer behaviour from Google to ChatGPT for a significant slice of informational queries. This is the most consequential shift.

2. Reddit as a cited source reached equilibrium. Q1 2025 had Reddit cited at 29% of AIOs. Q1 2026: 27%. The "Reddit everywhere" narrative plateaued. Wikipedia reclaimed some ground as LLMs added citation quality checks.

3. Brand mentions started mattering more than backlinks. For B2B queries in Q1 2026, being mentioned in relevant-topic articles (without a link) correlated more strongly with citation frequency than being linked to. The entity-recognition layer in LLMs values mentions-in-context.

4. Official docs dominate technical answers. Across 800 sampled technical queries, the cited source was the official product documentation 62% of the time. If your docs are not well-structured, clear, and crawlable, your product loses answer-layer visibility.

5. Small publishers declined across all LLMs. Site with <50k monthly visitors lost a median 35% of AI-citation share year over year. Citation weight consolidated to established publishers.

What we are telling clients for 2026

1. Treat product documentation as a marketing surface. Audit and invest accordingly.

2. Track brand mentions in third-party content (not backlinks, mentions). They matter independently now.

3. ChatGPT-specific visibility is the biggest lever. Get represented in the sources ChatGPT draws from (your own canonical content, third-party review sites, topic-specific publications).

4. Stop optimising for Bing. Its answer-layer share has collapsed. ChatGPT and Perplexity pull from broader indexes.

Sources: internal tracking across 23 B2B and DTC client domains, Q1 2025 vs Q1 2026; referrer log aggregation; Ahrefs and SimilarWeb cross-checks.