AI Search Lab
AI Search Scenario Frame | Mentions, Citations & Referral
Serious AI Search evaluation starts with a fixed query baseline and keeps brand mentions, source citations, and AI referral traffic in separate columns—so teams do not confuse one favorable answer with validated outcomes.
Simulated cases · Not client results
When assessing AI Search, separate mentions, citations and real referral traffic
These two simulated scenarios show common visibility challenges: a B2B software brand absent from AI answers, and a Hong Kong consultancy with fragmented entity signals. We use hypothetical challenges, actions and recording frames to show how to distinguish brand mentions, source citations and real referral traffic.
Case 01／B2B SaaS
From absent in AI recommendation answers to stable appearance on target queries
Challenge
This B2B software company ranked well on Google, but when prospects asked ChatGPT for tool recommendations, answers named competitors every time. Classic SEO performance did not automatically become AI recommendation visibility: brand, product use and differentiation had not formed a clear entity for the model, and core pages lacked content suitable for comparison and citation.
What we would do
The team starts with a citation-gap analysis—recording which questions competitors are mentioned on, which pages answers cite, and which content formats build trust. Core service pages are then rebuilt with product entities, fit scenarios and differentiation; authoritative FAQs aligned to AI question patterns are added; and structured data makes brand, product and content relationships easier for machines to identify.
- Establish fixed target queries and a competitor-mention baseline.
- Rewrite service pages for entities, use cases, audience, integrations and comparisons.
- Create independently citable FAQ and answer paragraphs.
- Add Organization, Product／Service, Article, FAQ and relationship schema.
- Continuously retest mentions, cited sources and AI referral sessions.
Recorded results
During the observation window, the brand began appearing on previously missed target queries and recorded traceable AI referral enquiries. Because raw accounts, query lists and identifying data are not public, this page does not claim percentages or multiples. The verifiable conclusion is that entities, citable content and ongoing retests together improved recorded answer coverage.
Case 02／Hong Kong professional services
Turn offline professional reputation into a service entity AI can understand
Challenge
This Hong Kong consultancy has deep professional experience and offline reputation, but no AI-answer visibility. When prospects ask Gemini or Perplexity about key advisory topics, competitors are named and the firm is absent. The problem is not only content volume—it is missing consistent, verifiable public relationships among bilingual brand names, consultant specialisms, service topics and region.
What we would do
We build a topical-authority programme around core advisory domains, adding definitions, processes, FAQs and expert viewpoints for each. The team also develops near-encyclopaedic neutral reference content, unifies company and consultant entities, and establishes traceable citation trails in high-authority publications so models can confirm specialism across sources.
- Unify Chinese and English brand names, service taxonomy, region and consultant identity.
- Build core pages and interlinked content clusters for main advisory topics.
- Add authors, credentials, methods, applicability and limits.
- Earn professional publishing and trusted external mentions—no fake reviews or anonymous endorsements.
- Measure brand mentions, source citations, answer accuracy and enquiry referral separately.
Recorded results
During the observation window, the brand moved from absence on main advisory topics to accurate mentions across several related questions, with traceable AI referral sessions. Because underlying records are client-confidential, this page is directional case illustration only; the record supports clear brand entities and cross-source professional signals as important improvement directions for later retests.
Shared execution process
From citation baseline to entities, content and external evidence
Citation Audit
Mark questions where the brand is mentioned, fully absent or misdescribed; record competitors and actual cited sources.
Entity Optimisation
Organise brand, people, product, service, region and sameAs relationships; remove name collisions, inconsistent descriptions or wrong classification.
Authority Content
For questions models must answer, build clearly structured, well-evidenced, independently extractable definitions, comparisons, processes and FAQs.
Citation Distribution
Build verifiable brand signals in real, relevant professional publications, reviews, communities, databases and other sources AI commonly retrieves.
Controlled Retest
Retest with the same queries, platforms and record fields; check AI referral, qualified leads and concurrent activities at the same time.
Build your own baseline
Cases are not forecasts; your query set is the decision starting point
Industry, brand foundations, model versions, content authority and external sources all affect outcomes. Build a repeatable test first—then you know the real gaps and sensible priorities.
Test brand mentions, citations and competitive gaps across major AI platforms.
Request an AI Search AuditFAQ about AI Search Scenario Frame | Mentions, Citations & Referral
Is this a real client case study?
No. It is a simulated teaching frame for diagnosis, intervention, and measurement—not a testimonial or proven project result.
Do example numbers apply to every brand?
No. Models, industries, baselines, query sets, and investment differ. Nothing here is a guarantee of mentions, citations, or commercial outcomes.
What baseline fields should AI Search cases record?
Platform/entry point, date, region, fixed queries, mentions, citations, answer accuracy, attributable referral, and concurrent SEO/PR/marketing activity.
Are brand mentions the same as source citations?
No. A brand can be named without a link; a citation means the answer shows or links a specific source. Log them separately.
When can something be called a verified AI Search case?
Only with verifiable underlying logs, clear baseline and window, client authorization or anonymization, concurrent-activity disclosure, and stated result limits.
Want a baseline of how AI answers describe your brand?
Request a free Audit