brand_radar_12082026_0255.md

Bupa’s brand is appearing consistently in AI-generated answers across Google AI Overviews, Google AI Mode, Gemini, Copilot and Perplexity, but the citations powering those mentions rarely come from Bupa’s own site. This is the dominant pattern in the data.

What stands out

• 100% share of voice is an artefact of tracking a single brand. The SOV is 1.0 and flat over 365 days, but only because the Brand Radar is monitoring Bupa alone. No competitor-comparison signal can be drawn from this metric. The real story is in the citations.

• AI answers mentioning Bupa are overwhelmingly fuelled by third-party sources.
  Of the 138 tracked AI questions where Bupa appears, the AI models collectively drew on 389 distinct pages from 187 domains. Bupa’s own domain provides only 26 pages (6.7%) and accounts for just 29 responses (21% of total mentions). The remaining 79% of Bupa mentions are attributed to external pages.

• Competitor and comparison-site domains dominate the citation landscape.
  Bowtie (bowtie.com.hk) alone accounts for 47 pages cited and 48 responses, nearly double Bupa’s own count. 10Life, hkvhis.com, MoneyHK101, AXA, and others all outrank Bupa’s domain in total responses. The most-cited individual pages are from comparison articles like moneyhk101.com’s VHIS compare (29 responses), galvoxstar.com (17), and hkvhis.com (16). Bupa’s top cited page is its homepage with just 10 responses.

• The volume is heavily concentrated in Google AI Overviews.
  Google AI Overviews delivers 8,750 of the 12,420 total impressions (70%) from only 24 questions — an average of 365 impressions per question. Google AI Mode contributes a further 2,460 from 30 questions. Copilot has the highest question count (54) but only 660 impressions, while Gemini (19 questions, 420 impressions) and Perplexity (11 questions, 130) barely register. This suggests the biggest opportunity and the highest-value visibility is in Google’s AI-powered SERP features.

• Top queries are high-intent, HK-specific, and dominated by voluntary health insurance (VHIS) comparison.
  The question “香港自愿医保哪家好？” generates 2,100 impressions in Google AI Overviews, followed by “自願醫保幾錢？” (800), “What is the best insurance company in Hong Kong?” (700), and variations on VHIS comparison and medical insurance quality. Direct brand queries (“Bupa有中文客服吗？”, “Bupa是哪里的公司？”) also appear, with moderate volume (400 each) and are correctly citing Bupa’s own customer-care and about pages.

What’s concerning

• Own-site citation share is dangerously low. Bupa’s authoritative content on its own products, plans, and VHIS information is not being surfaced as the AI’s primary source, even when the answer mentions Bupa by name. This means the brand’s messaging in AI answers is filtered through third-party interpretations, often on pages that also promote competitors.

• Bupa’s AI visibility is dependent on comparison sites. If a major aggregator like 10Life or Bowtie changes its content, ranking, or becomes inaccessible to crawlers, Bupa’s AI mentions could drop or be recast in a less favourable light. There is no evidence that Bupa controls the narrative; the data suggests it is riding on the coattails of competitor domains.

• The gap between question count and volume for Copilot/Gemini/Perplexity is suspicious. Google AI Overviews delivers enormous volume per question, yet Copilot triggers 54 brand mentions with minimal volume. This may indicate that many Copilot-cited questions are extremely long-tail or that the volume model in Ahrefs underestimates Bing-based AI traffic. Either way, the return on effort for optimising non-Google AI platforms appears skewed unless additional traffic data (e.g., analytics referrals) contradicts the modelled volume.

Connections and implications

• The co-occurrence of high-impression VHIS comparison queries with low own-site citation share is the core strategic problem. Bupa is present in the answers, but the user clicking a citation is more likely to land on Bowtie or a comparison site. This means Bupa is donating brand visibility while competitors capture downstream engagement. This is not a backlink problem — it’s a content authority and source preference problem in LLM retrieval.

• The fact that Bupa’s own pages are cited for direct brand queries (客服, about) but not for comparison queries indicates the AI models treat Bupa as a known entity but do not trust or retrieve Bupa’s pages as the best source for competitive evaluations. Bupa’s VHIS comparison page (bupa.com.hk/en/medical-insurance/vhis-info/voluntary-health-insurance-schemes-comparison-2026/) gets only 4 responses despite being directly relevant to the top volume questions. This suggests it is either not indexed effectively by AI crawlers, not recognised as sufficiently comprehensive, or outranked in retrieval by third-party resources.

• Data quality flag on total responses: The sum of responses from the top own pages listed appears to exceed the reported 29 total own-domain responses (top 10 pages alone sum to 33). This discrepancy is small but indicates potential double-counting or page consolidation in Ahrefs’ reporting. It does not alter the overall conclusion that own-site citation share is under 25%.

Potential featured finding

The single most impactful root cause is the severe under-citation of Bupa’s own domain (6.7% of cited pages, 21% of responses) on high-volume AI overview queries. This one finding explains why competitor domains are over-represented, why Bupa’s visibility does not translate into direct site engagement from AI answers, and why the brand’s AI presence is fragile. If Bupa’s VHIS and medical insurance content were preferred as the source in AI answers, it would simultaneously reduce competitor citation, improve message accuracy, and likely capture clicks directly. This is the finding that connects to crawl, indexation, and content quality signals — and the one that would unlock the most value if resolved.

Key signals

• Bupa appears in 100% of 138 tracked AI questions (estimated 12,420 total impressions), but 79% of those mentions use third-party sources, not Bupa’s site.
• Bupa’s own domain contributes only 26 of 389 cited pages (6.7%) and 29 of 138 responses (21%); Bowtie alone furnishes 48 responses.
• Google AI Overviews delivers 70% of all impressions (8,750) from just 24 questions — the highest-value channel for AI visibility.
• Top volume queries are VHIS comparison searches that should map directly to Bupa’s own VHIS comparison page, yet that page receives only 4 responses.
• The brand’s AI presence in comparison contexts is almost entirely mediated by competitor and aggregator domains, making it vulnerable to shifts in third-party content.