File viewer

brand_radar_12082026_0223.md

/app/data/llm/analysis/bupa/brand_radar_12082026_0223.md

Here's my analytical commentary on the Bupa AI Brand Radar data.

Analytical Commentary

The AI impression-to-mention ratio is exceptionally strong

Bupa generates 12,420 estimated AI impressions from only 138 questions tracked. That's roughly 90 impressions per question — an unusually high ratio that means Bupa is surfacing in high-volume queries. The top 5 questions alone account for 4,800 impressions (39% of total), all in the 600–2,100 monthly search volume range.

This is not a visibility problem. Bupa already has 100% share of voice among the single brand tracked, and that share has held flat at 1.0 for the entire 365-day history. The flat trendline, combined with zero competitor tracking, tells me the Ahrefs setup isn't configured for competitive monitoring yet. The 1.0 SOV figure is essentially meaningless as a competitive signal — it just means Bupa is the only brand in the tracked set.

Google AI Overviews dominates the volume but Copilot dominates the mention count

The source split reveals an important structural imbalance:

  • Google AI Overviews: 24 questions → 8,750 impressions (70.4% of total volume)
  • Copilot: 54 questions → 660 impressions (5.3% of total volume)

Copilot generates more than twice as many AI overviews containing Bupa as Google AI Overviews does (54 vs 24), but the Copilot queries are low-volume — roughly 12 impressions per mention. Google AI Overviews averages 365 impressions per mention. This means Google is surfacing Bupa in the high-intent, high-volume commercial queries ("香港自愿医保哪家好？" at 2,100 volume), while Copilot is mentioning Bupa in a long tail of lower-volume research/comparison queries.

The practical implication: losing Google AI Overviews visibility would be catastrophic. Losing Copilot would barely move the needle on traffic.

Bupa owns only 6.7% of its own AI citations — this is the core finding

Of 389 total cited pages across all AI sources, only 26 pages (6.7%) are from bupa.com.hk. This means 93.3% of the time an AI chatbot cites something about Bupa, it's citing a third-party page — a competitor's blog, a comparison site, an aggregator.

Bupa ranks 5th among cited domains with 29 responses from 27 pages. Bowtie (a direct competitor) leads with 48 responses from 47 pages — 66% more responses than Bupa. The top four domains ahead of Bupa are all comparison/aggregator/competitor properties:

  1. bowtie.com.hk — 48 responses (competitor)
  2. 10life.com — 35 responses (insurance comparison platform)
  3. hkvhis.com — 33 responses (VHIS comparison site)
  4. moneyhk101.com — 31 responses (financial comparison blog)

This is not just a "Bupa doesn't control its AI narrative" problem. It's an active competitor advantage. Bowtie is cited nearly twice as often, and its content is what AI chatbots use when discussing the Hong Kong insurance market. Every time a user asks "香港自愿医保哪家好？" (2,100 volume), the AI answer draws from comparison sites and competitor content — not from Bupa.

Bupa's own cited pages are shallow — homepage dominates, key product pages are invisible

The top own pages cited:

  1. www.bupa.com.hk/en/ — 10 responses (the homepage)
  2. VHIS comparison page — 6 responses
  3. 2026 VHIS comparison page — 4 responses

After that, it drops to customer care, about pages, and contact pages at 2–3 responses each. The homepage being the top-cited own page is not a good sign — it means AI models are citing Bupa's brand presence, not Bupa's content. The VHIS comparison pages are doing modest work (10 responses combined), but there are no citations to product detail pages, claims process pages, network hospital pages, or any content that would drive conversion.

Meanwhile, Bowtie's 47 cited pages almost certainly include deep product and educational content that feeds the AI models with substantive information.

The question intent is high-commercial but Bupa-specific queries exist

The top 10 questions split into two categories:

Market-level commercial queries (high competition): - "香港自愿医保哪家好？" (2,100) - "自願醫保幾錢？" (800) - "What is the best insurance company in Hong Kong?" (700) - "醫療險哪一家比較好？" (600) - "Which is the best medical insurance in Hong Kong?" (400)

Bupa-specific queries (brand capture): - "Bupa有中文客服吗？" (400) - "Bupa是哪里的公司？" (400)

The Bupa-specific queries together represent 800 impressions. These are people already in Bupa's orbit asking basic qualifying questions. The fact that these surface in Google AI Overviews means Bupa is capturing brand interest, but the volume is one-sixth of the single biggest market-level query.

The strategic question: if AI chatbots answer "哪家好" by citing Bowtie and comparison sites, how many of those 2,100 high-intent impressions convert to Bupa customers? The current data suggests very few, because Bupa's content isn't what the AI models are reading.

Comparison sites are the AI citation backbone — this may indicate a structural vulnerability

The top 10 cited pages are dominated by comparison content: moneyhk101 (29 responses), galvoxstar (17), hkvhis (16), alea.care (15), jumpstart (14), hkvhis again (14), bowtie (14), consumer.org.hk (12). Eight of the top 10 are comparison/aggregator properties. Only AXA (12) and Bowtie (14) are direct competitors with their own pages cited.

This pattern suggests AI models in the Hong Kong insurance vertical are trained on and preferentially cite comparison content, not brand content. If this is systemic (rather than Bupa-specific), it may indicate that all insurers in the market face the same structural disadvantage — but Bowtie's 48 responses vs Bupa's 29 proves some brands can break through.

Data quality flag: own pages count discrepancy

The digest_cited_pages shows own_pages_count: 26, but digest_cited_domains.own_domain shows pages: 27. This is a 1-page discrepancy within the same dataset. It's minor (3.8%) and doesn't change the analysis, but it suggests Ahrefs' Brand Radar has edge-case counting logic worth noting if we're quoting these figures in a proposal.

Key signals

  • Bupa owns only 6.7% of AI citations about itself — 26 own pages cited out of 389 total. When AI chatbots answer questions about the Hong Kong insurance market, 93.3% of citations go to third parties.
  • Bowtie outranks Bupa 48 to 29 in AI responses despite Bupa being the brand under analysis. A direct competitor has 66% more AI voice in Bupa's own tracked queries.
  • Google AI Overviews drives 70.4% of all AI impressions (8,750 of 12,420) from only 24 questions. The top question alone ("香港自愿医保哪家好？") accounts for 2,100 impressions — 16.9% of total volume. Visibility here is existential.
  • Bupa's own cited content is top-heavy and shallow — the homepage gets 10 responses (34% of all own citations), while customer care, about, and contact pages fill out the rest. No product detail pages, network pages, or claims pages appear in the top 10 own URLs.
  • 2,100-volume insurance comparison queries are being answered with competitor and aggregator content — the top cited domains are comparison sites, not insurers.

Potential featured finding

Bupa's near-total absence from its own AI citations is a single root cause expressing as multiple symptoms: low own-domain share (6.7%), competitor citation dominance (Bowtie at 48 vs Bupa at 29), and homepage-as-proxy-for-all-content signal. If Bupa can get its key commercial pages (VHIS product details, network coverage, price comparison frameworks) cited by AI models at rates comparable to comparison sites, it would simultaneously improve share of voice, capture high-volume commercial queries currently lost to aggregators, and reduce competitor citation advantage — all without needing to change anything about technical SEO or backlink profiles. This is a content structure and citation authority problem, not a domain authority problem.