A self storage GEO case study, measured at the 3.5-month mark. Over 14 weeks, Hingewise built a GEO/AEO foundation for V-Box Self Storage: a Wikidata entity plus 8 schema blocks (0 Rich Results errors), +37% organic traffic, and V-Box becoming the #2 AI-cited source in Vietnam for self storage. Most of all: we fixed what AI says about V-Box.
Over 14 weeks, Hingewise built a GEO/AEO foundation for V-Box Self Storage: a Wikidata entity plus 8 schema blocks (zero Rich Results errors), +37% organic traffic, and V-Box becoming the #2 AI-cited source in Vietnam for self-storage. All figures verified on Semrush and Wikidata, Aug 18, 2026.
V-Box rents self-storage units in Ho Chi Minh City. At the start, they faced three hard problems at once, plus a fourth that few notice.
Instead of pouring budget into low-volume keywords, we chose Generative Engine Optimization: building the entity infrastructure that lets AI models find, cite, and recommend V-Box.
In a young category, the brand that builds a clean entity profile earliest has the best shot at becoming the default source AI leans on, before competitors realize the game has shifted.
Structured data (schema)
0
8 JSON-LD blocks
Google Rich Results errors
2 warnings
0 errors · 0 warnings
Bilingual hreflang
0
Site-wide (en · vi)
Published content
~8
82 (66 VI + 16 EN)
URLs in sitemap
~8
84
GA4 measurement
0 events
5 live events
V-Box is now one of very few self-storage brands in Vietnam with a Wikidata entity, the trust anchor LLMs read to verify that a brand genuinely exists.
10 citations · just behind MyStorage (11) · far ahead of the rest. Semrush AI Toolkit, Aug 17
9% of 80 clusters · the low composite comes from coverage, not citation quality
Still lowest of five · gap to the leaders narrowed from 16 (May) to 12 (Aug)
1,761 (May) to a peak of 2,973 (Jul), then 2,413 (Aug) · Authority Score 15 to 17
Monthly organic traffic, 2026 · Semrush, VN database. The project began just as traffic bottomed out (May), then climbed to a peak of 2,973 (July).
What matters is not the total, but where the growth came from. V-Box’s homepage jumped from around 324 to 1,200 visits a month, a gain of 876, nearly half (48.9%) of all traffic, carrying genuine rental intent (top keywords: storage units, storage near me).
The Excel-template blog post still pulls about 28% of traffic with the wrong intent. That’s on the roadmap to fix, not something we count as a win.
On the composite AI-visibility score, V-Box still sits behind the leaders, understandable for a new brand covering few topics. But behind that total sits a telling detail: when V-Box does appear, AI cites it almost as often as the leader. The low composite comes not from citation quality but from narrow coverage (7 of 80 clusters), and the gap to the leaders narrowed from 16 points (May) to 12 (Aug). The leading indicators are all moving the right way.
The low composite score doesn’t come from citation quality, but from narrow topic coverage: present in 7 of 80 analyzed topic clusters. The leading indicators are all moving the right way, and the AI-score gap to the leaders narrowed from 16 points (May) to 12 (Aug).
AI citations in Vietnam · Semrush AI Toolkit, Aug 17, 2026. V-Box ranks 2nd, just one citation behind the leader.
This is the newest part, and the strongest evidence. At baseline (9-10 Jul) AI described V-Box wrongly on 8 facts. After the foundation went live, on 26 Aug we asked the same questions again on ChatGPT (web search on) and Gemini, and screenshotted both moments.
Evidence for a simple claim: fix the source and you fix what AI says.
We measured at the 3.5-month mark, so not everything is fixed yet. Two gaps are open, and both point to the same root cause: AI repeats whatever the published source still says.
V-Box appears in 7 of 80 topic clusters. All the empty space is room to grow.
This is an analytical conclusion, not a promise: V-Box does not lack citation strength, it lacks reach.
No made-up numbers. Our QC process caught 5 pieces of unverified data before they shipped, including a Wikidata ID in the source brief that turned out to be a rapper’s music album, not a storage company. Every ID was verified through the API before use.
Adapting to platform changes in real time. Mid-project, Google discontinued the Q&A feature on Google Business Profile. We caught it in time, stopped that workstream, and shifted resources to FAQPage schema and the AI answer layer.
Real E-E-A-T standards. Conflict of interest declared per Wikimedia policy, address data updated to the latest administrative changes.
The foundation is in place; results need time for AI and Google to re-crawl and for coverage to widen. We do not promise rankings, because nobody controls the AI algorithms. We commit to building the signals AI uses to choose sources, and to measuring transparently every month.
AI visibility (Semrush)
22/100
Narrow the gap to the 30–34 group
Topic clusters present
7/80
Expand coverage
Correct-intent traffic (homepage)
1,200/mo
Overtake the old wrong-intent traffic
Unit count AI cites
250 (older published figure)
338 confirmed (Cat Lai 88 + Phuoc Long 250)
The unfinished work shows the method. At measurement time, AI cited the figure V-Box had published (250). Working with V-Box operations, we confirmed the current total is 338 (Cat Lai 88 + Phuoc Long 250), and the published data is being updated to match. AI did not invent this, it repeated the published number, so once the data is current and consistent, AI answers correctly too. At the 6-month gate, we update in full.
GEO is optimizing a brand to be found, cited, and recommended by AI engines like ChatGPT, Gemini, and Google AI Overviews. Unlike traditional SEO that targets keyword rankings, GEO focuses on entity profiles, structured data, and citation share.
Yes, and that’s exactly when it matters most. For V-Box in the emerging self-storage category, Hingewise didn’t chase keyword volume; it built a digital identity (Wikidata entity, schema, AI-readable content) to become a source AI cites. The result: V-Box reached the #2 AI-cited source in Vietnam despite almost no established search demand.
Hingewise uses the Semrush AI Toolkit to track mentions, citations, and an AI-visibility score against the competitor set. Every number carries a date and source, and any “AI says X” claim is manually re-verified before it goes into a report.
The foundation (entity, schema, content) takes around 14 weeks. Visibility results need more time for AI and Google to re-crawl and for topic coverage to widen, so Hingewise measures and reports monthly rather than promising rankings.
Yes, often. AI repeats the information it reads; if your data is old, inconsistent, or thin, AI will get it wrong. For V-Box, 8 facts were wrong at baseline; after fixing the source, most are correct again. GEO is not only about being mentioned, but being mentioned correctly.
If you are in a young category with thin search demand, or AI already names your brand but gets the facts wrong, the fix rhymes: build a clean, verifiable source that AI can read and cite. You do not need high search volume to be recommended. You need to be the clearest, most consistent answer to the questions your customers actually ask an AI.
Or are competitors taking your place every time a customer asks ChatGPT, Gemini, or Google AI? Get a free GEO Snapshot: how you show up today, who’s being cited in your place, and the three things to fix first.
