The GEO Authority Diagnosis
Visora is a Hong Kong SEO and GEO agency that makes B2B brands visible inside AI answers — ChatGPT, Copilot, Gemini, Perplexity, and Doubao. The GEO Authority Diagnosis is the same 15-minute assessment we run at the start of every engagement, packaged as a free template: a 6-engine visibility test, a 7-dimension GEO scorecard, and the scoring bands that tell you whether AI would recommend you — or your competitor.
The paradox: a healthy website, an invisible brand
When we audited a Hong Kong industry institute in August 2026, their website scored 92/100 on SEO fundamentals — clean technicals, full schema, fast responses. Their AI Recommendation Score was 41/100. Two of six AI engines listed them at all; Copilot and Doubao actively did not recommend them. The website was healthy. The problem was whether AI engines knew the company existed. (Audit snapshot, 3 Aug 2026.)
That gap is now the norm for established B2B firms — professional institutes, law firms, family offices, consultancies. Buyers increasingly shortlist vendors by asking AI. If you are not in the answer, you are not on the shortlist, and no amount of on-site SEO fixes that.
What’s inside the template
| Section | What you do | What it tells you |
|---|---|---|
| A — 6-Engine Visibility Test | 3 questions × 6 AI engines (ChatGPT, Copilot, Gemini, Google AI, Perplexity, Doubao), scored /18 | Are you in the answer when prospects ask AI who to hire? |
| B — 7-Dimension GEO Scorecard | Entity Clarity, Quotable Facts, FAQ Coverage, Comparison Positioning, Structural Clarity, Authority Signals, Freshness — scored /70 | Does AI have anything quotable to cite about you? |
| C — Authority Gap | Combined score mapped to five bands: Invisible → Emerging → Contested → Established → Authority | Where you stand today, in one number |
| D — Competitor Blind-Spot Check | Cross-check your Google top-10 against AI citations | The opening your competitors haven’t seen |
The benchmark: what happened when FOIHK ran this
The Family Office Institute Hong Kong (FOIHK) ran this exact diagnostic as the first step of its engagement. The full write-up — with dated before/after screenshots — is in the FOIHK case study. Headline outcomes:
| Measure | Result | Snapshot date |
|---|---|---|
| Baseline | 0 of 15 AI answers mentioned FOIHK; name confused with a competitor 3 times | Jun–Jul 2026 |
| Copilot | Not listed → recommended #1; wrong name corrected | Re-test 26 Jul 2026 |
| ChatGPT | "Unable to display" → ranked #1 | Re-test 26 Jul 2026 |
| Google AI | Not listed → Top 5, ranked #3 | Re-test 26 Jul 2026 |
| Name confusion | 100% of errors corrected (Copilot, Gemini) | Re-test 26 Jul 2026 |
| Google ranking | #2 for "family office institute HK" | 17 Aug 2026 |
By the numbers
Quotable facts, with sources — the same standard the template asks of you:
- 15 probes × 6 engines — per full diagnostic — the self-assessment is the 3-question screening version
- 38/70 — FOIHK’s 7-dimension GEO score on a brand-new, technically excellent website (audit, 3 Aug 2026)
- 9 of 9 — top-ranking Google domains in FOIHK’s category cited by no AI engine at all — Google rank ≠ AI visibility (3 Aug 2026)
- 2 of 6 — AI engines listing FOIHK at baseline — Copilot and Doubao explicitly did not recommend them (3 Aug 2026)
What we still haven’t fixed — and why we tell you
One AI engine (Doubao) still mis-states FOIHK’s name, and Google’s Chinese full-name search still surfaces the competitor. We publish this because a diagnosis you can trust shows the unfinished work. Every client gets the same treatment: monthly re-tests, same day, same questions, dated snapshots — everything shown, nothing hidden. (Status as of the 26 Jul 2026 re-test.)
From self-score to full diagnosis
Self-scores are a screening tool, not a measurement. The full GEO Authority Diagnosis runs 15+ probe queries per engine across 6 engines, measures recall, reach, convergence and factuality, audits your off-site entity footprint (Wikipedia, Wikidata, LinkedIn, directories), and produces a per-engine verdict matrix with a dated re-test baseline. We run a fixed number each month, and you keep the full report either way.
- Book the full diagnosis — Fixed monthly slots — report is yours to keep
- Free 60-second AI visibility scan — See how AI engines describe your business today
- Read the FOIHK case study — The full before/after, with dated evidence
Frequently Asked Questions
What is the GEO Authority Diagnosis?
A 15-minute self-assessment that scores how visible your brand is to AI engines. It combines a 6-engine visibility test (3 questions each across ChatGPT, Copilot, Gemini, Google AI, Perplexity and Doubao, scored /18) with a 7-dimension GEO scorecard (/70), then maps the total to five bands from Invisible to Authority. It is the same instrument Visora runs at the start of every client engagement.
What is GEO, and how is it different from SEO?
SEO optimises for Google’s ranked links. GEO (Generative Engine Optimization) optimises for the AI recommendation layer — whether ChatGPT, Copilot, Gemini and Perplexity name you when a buyer asks who to hire. A site can score 92/100 on SEO while AI engines score the brand 41/100; that gap is what GEO closes.
Is the template really free?
Yes. Enter a work email and the download unlocks immediately, in English, Traditional Chinese and Simplified Chinese. We send one follow-up with practical tips; you can unsubscribe in one click.
What results did this method produce for FOIHK?
Five weeks after baseline: Copilot went from not listing FOIHK to recommending it #1, ChatGPT from "unable to display" to #1, and Google AI placed it #3 in its Top 5; 100% of name-confusion errors were corrected (re-test, 26 Jul 2026). Google ranked it #2 for "family office institute HK" (17 Aug 2026). Full evidence: the FOIHK case study.
What if an AI engine confuses my brand with a competitor?
That is the most dangerous failure mode — AI is actively sending your prospects elsewhere. It is usually fixable through entity clarification: consistent naming, a quotable one-sentence definition of what you are, and an off-site entity footprint (Wikipedia, Wikidata, LinkedIn). FOIHK’s confusion with a near-identically named competitor was 100% corrected on Copilot and Gemini within five weeks.
How often should we re-test?
Monthly, on the same day, with the same questions — AI answers drift weekly. Visora’s engagements include dated monthly re-tests against your baseline, so movement is measured, not guessed.
Last updated: