News
Real case: HKUNIFI store went live on 22.09 — first 8 days of GEO results, coverage 17.1% and 2/5 keywords now cited by AI (with screenshots)
hkunifi.com, GNS's own UniFi store, went live on 22.09.2026. First report: coverage 17.1%, AI mention rank #2.4, SOV 5.5%, and 2 of 5 target questions cited.
Anson Ng
16 min read
GNS's own UniFi store, hkunifi.com, went live on 22.09.2026. Today (30.09.2026) the first GNS-GEO report is out, and this post covers the first batch of real numbers.
The short answer: coverage 17.1%, AI mention rank #2.4, SOV 5.5%, and 2 of 5 target questions already cited by AI. These are early numbers, eight days after launch, not an endpoint. Below we list which keywords are working, which are still gaps, and three things that should not be over-read.
The store's GEO foundations (AI crawler policy, llms.txt, structured data, entity links) were published item by item in an earlier post: own case: the GEO foundations behind hkunifi.com. This post is about the results.
First report, 30.09.2026
| Metric | 30.09.2026 |
|---|---|
| Coverage | 17.1% |
| AI mention rank | #2.4 |
| SOV | 5.5% |
| Questions cited | 2 / 5 |
Scope: 1 website, 1 alias, 5 keywords. Launch date: 22.09.2026.
Five target questions: two are already cited
| Keyword (as shown in the report) | Type | Coverage | AI mention rank |
|---|---|---|---|
| 公司想將網絡外判俾人托管，每個月俾月費… (outsourced managed network, monthly fee) | Service query | Not yet | — |
| 我想將防火牆、交換機、Wi-Fi 同 CCTV 整合… (integrate firewall, switch, Wi-Fi and CCTV) | Service query | 50% | #3.1 |
| 我想幫間鋪頭換商用 Wi-Fi，唔想再用家用 router… (business Wi-Fi for a shop) | Service query | Not yet | — |
| 香港公司買UNIFI產品 (buying UniFi gear in Hong Kong) | Product query | 21% | #4.7 |
| 香港邊度買 UniFi 行貨最穩陣？有無官方渠… (where to buy official UniFi stock) | Channel query | 29% | #2.5 |
The report footer reads "2/5 distinct keywords cited". All five answers are generated by Google AI Overview / AI Mode. Keywords are shown as they appear in the report, some truncated by the interface.
What is working, and what is still a gap
Coverage has started (3 keywords):
- Integrated management interface (firewall / switch / Wi-Fi / CCTV): coverage 50%, rank #3.1 — the best of the five
- 香港邊度買 UniFi 行貨最穩陣 (where to buy official UniFi stock): coverage 29%, rank #2.5 — the best rank
- 香港公司買 UNIFI 產品 (buying UniFi gear in HK): coverage 21%, rank #4.7 — in the answer, but low in the list
Not mentioned at all (2 keywords):
- 公司想將網絡外判俾人托管（月費託管） (outsourced managed network, monthly fee)
- 我想幫間鋪頭換商用 Wi-Fi，唔想再用家用 router (business Wi-Fi for a shop)
Both are service-style queries: the user has not decided which product to buy, they are asking who can do it for them. That kind of answer needs concrete service content — scope, process, cases, how quotes work — while our pages at this stage lean towards a product catalogue. The next step is dedicated content for those two service queries.
Three things not to over-read
1. "2/5 cited" and "3 keywords with coverage" are different things
Three keywords show non-zero coverage, meaning the brand appears in the AI answer, but the report counts only two as cited. Mentioned means the name appears; cited means the answer includes the brand's content or a link to it. We covered the distinction in detail elsewhere: a guide to AI search metrics.
2. Adding up the per-keyword figures does not match the dashboard
The five keywords read not yet / 50% / not yet / 21% / 29%. Treating the gaps as 0% gives an average of roughly 20%, while the dashboard shows 17.1%. The difference most likely comes from the tool testing several phrasings of each keyword and counting all of them in the dashboard, while the table shows one representative value per keyword. We use the dashboard figure as the headline number and publish the per-keyword detail alongside it, so readers can check both.
3. "Not yet" and "—" are not a verdict of 0%
For uncovered keywords the report shows "not yet" for coverage and "—" for rank, because there is no rank without a mention. That should not be read as a decline: the store is only 8 days old, so this is missing data rather than worsening performance.
Comparing with New Win Tec: different timing, not directly comparable
Our other case, New Win Tec (latest report), was already at 5/5 cited two weeks after launch and is now at 95.7% coverage. The two should not be compared directly:
- New Win Tec: B2B services (tail lift repair and installation), a narrow set of target queries, few competing brands
- HKUNIFI: ecommerce products, where AI answers usually list several authorised dealers and retail channels, so the denominator is far larger
That is also why SOV is 5.5%: the brand is being mentioned, but the market itself has many competitors listed in the same answers.
Citation trend (22/09-30/09)
The trend chart shows our brand line (blue) starting near zero on 22/09, rising to a peak around 28/09 (roughly 5 on the axis), easing to about 3, then recovering to about 4 on 30/09. The comparison lines move between 0 and 3 over the same window.
One honest note: in this screenshot both comparison lines carry the same label, so we cannot reliably tell which brands they are. We therefore report only the shape of the trend and do not name them.
Next steps
- Add dedicated content for the two service-style keywords (outsourced managed network, business Wi-Fi replacement)
- Strengthen product pages with citable specifics (models, comparisons, lead times, installation scope)
- Build third-party citations (dealer pages, directories, customer reviews)
- Keep monitoring weekly on the same scope and judge on the 4-8 week trend
Two previously disclosed gaps still stand: hkunifi.com/sitemap.xml still returns 404 (only sitemap-index.xml is served), and there is still no visible link to geo.gnshk.com anywhere on the store's pages.
How to verify
The figures above are taken from the 30.09.2026 GNS-GEO report. You can reproduce them yourself:
- Google search "香港公司買UNIFI產品": https://www.google.com/search?q=%E9%A6%99%E6%B8%AF%E5%85%AC%E5%8F%B8%E8%B2%B7UNIFI%E7%94%A2%E5%93%81
- Google search "香港邊度買 UniFi 行貨最穩陣": https://www.google.com/search?q=%E9%A6%99%E6%B8%AF%E9%82%8A%E5%BA%A6%E8%B2%B7+UniFi+%E8%A1%8C%E8%B2%A8%E6%9C%80%E7%A9%A9%E9%99%A3
AI answers change over time and the numbers move; GNS-GEO monitors weekly and compares on the same scope so the trend stays comparable.
Summary
HKUNIFI went live on 22.09.2026. After 8 days: coverage 17.1%, AI mention rank #2.4, SOV 5.5%, and 2 of 5 target questions cited. Product and channel queries have started to gain coverage; the two service-style queries remain the main gap and are the next priority.
To see where your own brand actually stands in AI search, start with the free AI visibility report.
FAQ
What are the figures in HKUNIFI's first GEO report?
hkunifi.com went live on 22.09.2026. The first GNS-GEO report on 30.09.2026 shows coverage 17.1%, AI mention rank #2.4, SOV 5.5%, and 2 of 5 target questions already cited by AI. Scope: 1 website, 1 alias, 5 keywords.
The report says 2/5 cited, but 3 keywords show coverage. Why?
Because being mentioned and being cited are two different metrics. Three keywords show non-zero coverage, meaning the brand appears in AI answers, but the report footer counts only 2 as cited, meaning the answer includes the brand's content or link. The two should not be conflated, so we report them separately.
Is 2 out of 5 after 8 days a success?
All we can say at this stage is that coverage has started. Eight days is too small a sample to judge direction, and ecommerce queries usually return several retailers in one answer, so a single brand's coverage and SOV are naturally lower. We monitor weekly on the same scope and will judge on the 4-8 week trend.
Why is SOV only 5.5%?
SOV is a relative share. AI answers for UniFi product queries typically list several authorised dealers and retail channels, so the denominator is large. The store has also only been live 8 days, so accumulated third-party citations are still limited. A low share is expected at this stage.