FOIHK Case Study: Local SEO & AI Visibility for a Regulated Industry

By the Visora team — the practitioners who ran this engagement in Hong Kong.

FOIHK (Family Office Institute Hong Kong) went from appearing in 0 of 15 AI-engine answers to being recommended #1 by Copilot and ChatGPT, and #3 in Google AI’s "Top 5 family office institutes in Hong Kong" — after a four-part Visora programme of trilingual technical SEO, full-site schema markup, answer-first content, and GEO (Generative Engine Optimization). The engagement is still in progress; every number below is a dated, verifiable measurement. Screenshot evidence: FOIHK before-and-after portfolio.

The Client

FOIHK (Family Office Institute Hong Kong, 香港家族辦公室學會) is a Hong Kong-based industry body and professional community for the family office sector, covering education, research, responsible philanthropy, events, and cross-border exchange. As a regulated financial-services institution, FOIHK serves Hong Kong’s family office professionals, family members, advisors, and institutions in three languages — English, Traditional Chinese, and Simplified Chinese — with plans to expand the proven model to mainland China and Singapore.

The Challenges

When we took over, FOIHK faced the standard dilemma of doing SEO in a regulated industry — plus an increasingly fatal new problem: invisibility in AI search.

The last row deserves expansion. Before the project began, we ran a complete AI-visibility baseline: 5 major AI engines, each asked the same 3 questions ("What is Family Office Institute Hong Kong" / "Recommend the top 5 family office institutes in Hong Kong" / "I am looking for good family office institutes in Hong Kong"). The result: across 15 answers, FOIHK appeared in 0 recommendation lists, and its name was confused with a competitor 3 times. This is the reality more and more B2B institutions face without realising it: clients no longer just "Google it" — they ask AI. And if AI doesn’t know you, you don’t exist.

The Goal

There was no pre-written KPI list. The client’s goal was a single sentence: when someone searches Google, or asks an AI, about family office institutes in Hong Kong, FOIHK must be found — and correctly identified. We decomposed that into two verifiable tracks: Be found — comprehensive local search visibility across all three languages, with core pages fully indexed by Google; and Be recognised — mainstream AI engines correctly identify FOIHK (fixing the name confusion) and include it in "family office institute recommendation" answers. Monthly AI-visibility re-tests, ranking, and indexing monitoring were built on top, with the technical and content architecture designed from day one for the later mainland China and Singapore expansion.

The Strategy: Four Engines Driving One Flywheel

We designed a programme of technical foundation × content system × structured data × AI visibility (GEO) — Hong Kong first, then the proven playbook replicated to new markets. Regulated industries have one peculiarity: every public statement must withstand regulatory and factual scrutiny. So our content strategy was never "stuff keywords" — it was to build FOIHK into a credible source that both search engines and AI engines dare to cite: verifiable identity, traceable sources, structured statements.

Phase 1 — Technical Foundation: Let Search Engines Understand the Site

• Trilingual hreflang architecture: every page carries en / zh-HK / zh-CN / x-default hreflang tags plus canonical links, so Google knows exactly which language version serves which Hong Kong audience — and the same architecture is ready-made groundwork for future market expansion.
• Local search foundation: completed Google Business Profile details and Hong Kong local citations, locking onto "Hong Kong + family office" local searches.
• Technical SEO audit and fixes: site speed, mobile adaptation, and index health checked item by item; static entry files generated for 90 routes so crawlers can fetch everything.
• Schema markup system: six types of structured data deployed — Article, FAQPage, BreadcrumbList, Organization, Event, Person — so search engines never have to guess what a page is.
• IndexNow instant submission: content pushed to Bing and IndexNow-connected engines within seconds of publishing, cutting new-content indexing from "wait for the crawler" to minutes.
• Trust infrastructure: privacy policy, editorial policy, and contact pages completed — not just compliance, but the baseline signals of E-E-A-T.

Phase 2 — Content System: From "Write What We Have" to "Answer What Users Search"

• Keyword research and layering: a BOFU (high-intent) / MOFU (consideration) / TOFU (awareness) matrix, attacking brand and high-intent terms like "Family Office Institute Hong Kong" first, then expanding to decision-stage long tails like "how to choose a family office institute".
• Pillar-and-cluster architecture: content clusters built around three pillars — Education & Research, News & Events, and Philanthropy — interlinked with exact-keyword anchor text.
• FAQ content system: 29 real high-frequency questions (positioning, non-profit status, SFC licensing, fees), each answer written to a "directly quotable" standard — short, complete, standalone sentences that search engines and AI engines can lift verbatim.
• Localised adaptation: the Traditional Chinese version uses Hong Kong written Chinese and local terminology (e.g. 非牟利機構), the Simplified version uses mainland-standard phrasing (e.g. 非營利機構), and the English version addresses an international audience — one topic, three precise expressions.
• E-E-A-T signals: real author bylines (name + title + one-line credentials), a public editorial policy, and disciplined source citation — in a regulated industry, credibility is itself a ranking factor.

Phase 3 — GEO: Make AI Willing to Mention You

This is the leap that separates the project from traditional SEO. For ChatGPT, Gemini, Copilot, Doubao, and Google AI, we did three things:

• Entity clarification: unified FOIHK’s Chinese and English full names, legal name, and profile wording across the site and off-site sources, eliminating the name confusion with peers like FOAHK — the thing AI engines hate most is not knowing who is who.
• Answer-first content form: core facts on key pages rewritten as standalone quotable sentences, paired with FAQPage schema, so AI engines can lift the original sentence directly.
• Off-site entity footprint: completed institutional information on authoritative external sources such as Wikipedia/Wikidata and LinkedIn, so AI engines can cross-verify the organisation.

Phase 4 — Continuous Optimisation (In Progress)

• Monthly AI-visibility re-tests: fixed question set, fixed engines, before/after comparable.
• Keyword ranking and new-opportunity monitoring; content updated monthly with a visible "last updated" date.
• IndexNow submission logs and crawler-visit statistics tracked continuously.
• Market-expansion roadmap: replicate the SEO + GEO playbook proven in Hong Kong to mainland China and Singapore — the trilingual architecture and content system were designed for this from the start.

Early Breakthrough: AI Visibility Re-Test (2026-07-26)

Five AI engines plus Google brand search, each asked the same three unified questions. Before-and-after results:

Key Highlights

• Copilot: from "not listed" to #1 recommendation for "good family office institutes".
• ChatGPT: the same type of question went from "unable to display" to ranked #1.
• Google AI’s "Hong Kong Top 5 family offices" list: FOIHK entered at #3.
• AI name confusion (Copilot / Gemini) fully corrected — the brand’s entity identity was legitimised inside the AI world.
• Trilingual hreflang deployed with zero errors; Google now correctly maps languages to regions.

Baseline Metrics (Recorded 2026-08-17)

An honest note: as of 2026-08-17 the programme is still being deployed in phases and has not fully landed. This baseline exists so every future gain is measured against a real starting point.

Early SERP Rankings (Google Incognito, 2026-08-17)

• "family office institute HK" — #2 (behind only FOI.ASIA; an early signal of brand-term breakthrough).
• "family office training HK" — #22 (an education term, FOIHK’s home turf, with clear headroom).
• Other core terms ("family office hong kong", "香港家族辦公室", "家族辦公室課程") — not yet in the top 20; listed as the next attack targets.

Competitor Baseline (2026-08-17)

How We Read the Baseline

• The content flywheel is already spinning: FOIHK’s output (~6 posts/month) exceeds every direct- competitor association — FOAHK nearly dormant, FOI.ASIA at zero, RFO dormant for 2.5 years; only AOF, PWMA, and the media-type rivals still publish actively.
• The gap concentrates in domain authority and index scale: DR 3.2 versus competitors’ 24–76, and 13 indexed pages versus hundreds — exactly what link building and continuous indexing must close.
• Brand terms show an early breakthrough: ranking #2 for "family office institute HK" indicates entity clarification and brand optimisation are working.

An Honest Note: What Is Still Being Optimised

As of 2026-08-17, the full rollout is incomplete and the gaps are real: Doubao shows no movement on the three test questions yet; the Chinese full-name search still surfaces FOAHK; organic traffic and keyword rankings are still compounding. Every gap is on the next phase’s optimisation list, tracked with monthly re-tests. Reporting what we have not achieved yet is exactly what makes this case study trustworthy.

What This Project Proves

Technical foundation × content strategy × structured data × GEO — together they form a growth flywheel that keeps running. Hong Kong is only the first stop.

• Regulated industries can do SEO — regulatory constraint is not a ceiling on content; being "credible, verifiable, and quotable" is the moat of this industry.
• Local SEO has no shortcuts — a unified technical architecture plus localised content execution are both indispensable. Master the Hong Kong market first, then replicate the proven playbook to more markets.
• AI visibility is the next battleground — from "0 appearances in 15 AI answers" to "#1 in two engines", GEO is not a concept; it is a hard metric that can be tested question by question and optimised item by item.

Appendix: Full AI-Visibility Test Data

The three unified test questions: Q1 "What is Family Office Institute Hong Kong"; Q2 "Can you recommend me the top 5 family office institutes in Hong Kong"; Q3 "I am looking for good family office institutes in Hong Kong". After measurements taken 2026-07-26.

Full-Name Recognition Rate (2026-07-26)

See the Evidence — Then Measure Yourself

• FOIHK screenshot portfolio — dated before-and-after captures from every engine
• All client results — proof, not promises
• Free 60-second AI visibility scan — your 0–100 score across 6 engines
• GEO Hong Kong — how Generative Engine Optimization works for HK businesses
• Visora services — scans, platform plans, GEO, done-for-you SEO

Frequently Asked Questions

Can regulated industries like family offices do SEO?

Yes. The FOIHK case study shows that regulatory constraint is not a ceiling on content — being credible, verifiable, and quotable is the moat of a regulated industry. Visora built FOIHK into a source that search engines and AI engines dare to cite: verifiable identity, traceable sources, and structured statements.

What is GEO and how is it different from SEO?

SEO targets Google’s blue-link rankings; GEO (Generative Engine Optimization) targets the AI recommendation layer — whether ChatGPT, Gemini, Copilot, Doubao, and Google AI name and recommend you. In this project, GEO moved FOIHK from 0 mentions in 15 AI answers to #1 recommendations in Copilot and ChatGPT.

How long did it take for AI engines to reflect the changes?

The baseline measured 0 mentions across 15 answers before the work began. The first re-test on 2026-07-26 already showed #1 recommendations in Copilot and ChatGPT and a #3 entry in Google AI’s Top 5 list — retrieval-level changes appeared within weeks, while deeper "trained memory" absorption continues to be tracked monthly.

Is the FOIHK project finished?

No — it is an ongoing engagement. As of 2026-08-17 the deployment is still in phased rollout: Doubao results and Chinese full-name search are still being optimised, and every gap is tracked with monthly AI-visibility re-tests against the recorded baseline.

Last updated: 29 August 2026