Somantra AI’s Australian Citation Report said it analyzed 2.4 million citations across seven months of Australian insurance data and found a clear split in how ChatGPT and Google surfaced sources. In July 2026, ChatGPT cited insurance brand homepages 12.7 times more often, while government explainer pages held an average position near 2.0.
That matters because insurance sits in a high-trust category where the difference between a brand page, a government explainer and a third-party source can change the answer a consumer sees. The report’s core signal is not just that brands are visible, but that citation behavior shifts by intent: homepages appear to carry more weight when the query is closer to branded discovery, while foundational public-sector pages still win on authority for coverage, eligibility and policy questions. The measurement also reinforces a practical point for Australian insurers, which is that AI citation performance is not a single ranking game across all queries.
The report fits into a broader run of Somantra AI insurance research in 2026. In June, the company published AI Search Visibility: Australian General Insurance Brands - May 2026 Report and said it tracked 20 insurance brands across 34,278 real consumer conversations. On July 17, it published The AI Search Anomaly: How 38 Domains Contaminated Insurance Advice Across AI Search Engines, which pointed to low-quality and harmful insurance advice appearing in AI search. The company also published brand-versus-brand pages for Suncorp vs Budget Direct and CGU vs Budget Direct, both describing more than 34,000 conversations about the Australian insurance industry across ChatGPT and Google AI answers.
For insurers, the content move is specific. Homepage architecture needs to do more than state a brand name and a slogan. It needs clear product entry points, plain-language explanations, and entity signals that help a model identify who the insurer is, what it sells and which policy family answers which question. Product pages and FAQ pages need the same discipline, with coverage terms, exclusions, claims steps and eligibility written in language a model can parse without guessing.
The same data also argues for a stronger trust layer beyond owned media. Government-style educational pages, regulatory references and third-party validation can help on questions where AI systems favor authority over promotion. The report does not turn citation share into proof of causation, but it does show that Australian insurers are already being judged on how well they organize content for discovery, explanation and trust in two different answer engines at once.
