File viewer

brand_radar_12082026_0118.md

/app/data/llm/analysis/house730/brand_radar_12082026_0118.md

Analytical commentary: house730.com — Brand Radar data

Top-line assessment

This is a strong dataset with no obvious crisis points. The raw citation counts are modest in absolute terms, but the composition of the data tells a clear story: house730.com is the single most-cited domain in its tracked topic cluster, with a 19% share of all cited pages — the highest of 194 citing domains. That's a dominant position relative to competitors like centanet (104 pages cited, 89 responses) and 28hse (44 pages, 49 responses). The "we're being seen" part of the AI visibility puzzle is working.

The more interesting question is whether the site is being cited enough relative to its content surface area, and whether the questions that trigger citations align with where the business wants to appear.

What dominates: Google AI Overviews

Google AI Overviews accounts for 80% of total impression volume (14,190 of 17,710) and 40 of the 217 tracked questions. Copilot has more raw question counts (111) but negligible volume (730 total — just 5% of Google AI Overviews' volume). This is a Google AI Overviews story. Everything else is noise by comparison.

Takeaway: when evaluating AI-driven visibility, the metric that matters for this site is performance in Google AI Overviews specifically. Copilot, Perplexity, Gemini are rounding errors on volume.

Question profile: hyper-local, transactional intent

The top-10 questions by volume are overwhelmingly Cantonese-language queries with clear transactional or pre-transactional intent:

  • "將軍澳居屋價錢？" (1,300 vol)
  • "香港最新房價？" (1,100 vol)
  • "買村屋要幾多錢？" (900 vol)
  • "益豐大廈歷史？" (800 vol)
  • "馬鞍山錦駿苑 售價？" (800 vol)
  • "玖瓏山好住嗎？" (800 vol)

Every single top-10 question names a specific district, estate, or housing type. These are users close to a decision — price checks, liveability assessments, specific estate queries. This is high-intent traffic. If house730.com is being cited in answers to these questions, it's appearing at exactly the right moment in the user journey.

There's no evidence of informational or top-of-funnel content dominating the citation mix. That's positive.

The 19% share: a glass-half-full number

house730.com owns 160 of 844 cited pages (19%). That's the #1 share among 194 domains. But it's worth looking at what the other 81% represents:

  • Competitor property platforms: centanet (89 responses), 28hse (49), Midland (40), Ricacorp (38). Combined, the big four traditional agencies account for roughly 26% of all responses.
  • Wikipedia: 42 responses. Not a competitor, but a reminder that AI models fall back on encyclopedic sources.
  • Media and content sites: hk01.com (11), spacious.hk blog (18), richitt.com (12). These are content plays picking up citations through blog posts and guides.

The gap between house730 (188 responses) and #2 centanet (89) is a 2.1x lead. That's a meaningful moat. But the long tail of 194 domains suggests the AI citation landscape is fragmented — no single player has lock-in. This is a land-grab moment, not a won battle.

Own pages: breadth, not depth

house730.com has 169 unique pages cited, generating 188 total responses. That's 1.1 responses per page on average. The top-cited own page (www.house730.com/rent/t1b8/) has only 9 responses — out of 217 total questions.

This is a breadth play, not a depth play. The site is cited across many pages, each appearing in a small number of answers. This pattern is consistent with listing pages and estate profile pages being surfaced in response to highly specific queries (e.g., "玖瓏山好住嗎？" triggering an estate page for that development).

No single page is a citation powerhouse. That's not necessarily bad — it means visibility is diversified and not dependent on one high-performing asset. But it also means there's no obvious content template to replicate and scale. Each cited page appears to earn its place through relevance to a narrow query, not through being an authoritative hub.

Copilot anomaly: many questions, no volume

Copilot generates 111 questions (51% of all tracked questions) but only 730 impressions (4% of total volume). Average volume per Copilot question: 6.6. Average volume per Google AI Overviews question: 355.

This suggests Copilot is surfacing the brand in long-tail, low-volume queries. That could indicate Copilot is scraping a wider range of pages or has different citation thresholds. But the volume is negligible. For reporting purposes, Copilot citations are vanity metrics — impressive question counts, no meaningful traffic.

Hypothesis to verify: Copilot's citation behaviour may be driven by Bing's index. If house730.com has strong Bing organic visibility (to be confirmed via separate data), this Copilot pattern would be consistent with that. If Bing organic visibility is weak, this is harder to explain.

What's missing: no negative signal, but no growth signal either

Share of voice is 1.0 and flat over 365 data points. This means house730.com is the only brand being tracked in this cluster — likely a single-brand monitoring setup. That makes the 1.0 SOV uninformative; it's a function of the tracking configuration, not market dominance.

The flat line tells us nothing about whether citations are growing or shrinking relative to the total addressable question set. The AI question landscape is expanding rapidly as models add capabilities and user adoption grows. Flat SOV in a single-brand tracker doesn't tell us whether house730.com is keeping pace with that expansion.

To assess growth, we'd need one of: - Absolute mention counts over time (not provided in this dataset) - A multi-brand SOV tracker including competitors - Google AI Overviews impression trend data from GSC (if available)

The cited pages lens: competitors are in the room

Looking at the top cited pages overall (not filtered to own domain), the #2 most-cited URL is a 28hse.com short-term rental page (7 responses). #3 is a spacious.hk blog post about short-term rentals (5). #4 and #5 are runhotel.hk and weave-living.com, both short-term rental-related.

This cluster suggests AI models are surfacing competitors heavily on short-term rental queries. house730.com's top-cited page (9 responses) is a general rent listing page — it may or may not be winning on short-term rental queries specifically.

Hypothesis to verify: if short-term rentals are a strategic priority for house730.com, there may be a content gap. The data suggests competitors with dedicated short-term rental content are being cited. If house730.com lacks equivalent content depth on this topic, it's leaving citations on the table.

Potential featured finding

The 19% own-page citation share, combined with a 2.1x lead over the #2 domain, signals that house730.com has a structural advantage in AI citations that is not yet fully exploited. The breadth-over-depth pattern (169 pages, 188 responses, 1.1 avg) indicates the site is citation-eligible across a wide surface area but thin on any single topic. If the site consolidates authority on high-volume query clusters (e.g., specific estate pages, price guides, transactional content), it could significantly increase response counts without needing to create net-new content — just strengthen what's already being surfaced.

Key signals

  • 80% of AI impression volume comes from Google AI Overviews (14,190 / 17,710). Copilot (730), Gemini (1,260), and Perplexity (320) are marginal. All strategic AI visibility effort should focus on Google AI Overviews.
  • 19% own-page citation share is #1 among 194 domains, with a 2.1x lead over #2 (centanet, 89 responses). The moat is real but narrow in a fragmented landscape — this is a land-grab window, not a permanent advantage.
  • Top-10 questions are 100% Cantonese, hyper-local, transactional-intent queries on specific districts and estates. The brand is surfacing at the decision moment, not at top-of-funnel. That's the highest-value placement possible.
  • Citation pattern is breadth-over-depth: 169 pages generating 188 responses (1.1 avg). No single page has more than 9 citations. Visibility is diversified but thin — there's no anchor content asset driving disproportionate citations.
  • Copilot generates 111 questions (51% of total) but only 4% of volume. Copilot citation counts are a vanity metric with negligible real-world traffic impact. Do not use Copilot question counts to size opportunity.