Getting named ahead of the others takes evidence only you could have written.
AI Visibility Score, 28 Jan – 4 Feb 2026. A finance master’s, four Chinese models. Read the case- B
- Behaviour
- P
- The person
- E
- Their situation
Be the brand
AI recommends.
What a buyer chooses depends on who they are and the moment they are in. We model both, ask their real questions across ten AI models, and show you exactly where you stand. AI SEO, GEO and AEO by Jeffery Asia.
error
Enter a domain, like yourbrand.com.
arrow_upward
Please enter your brand website to run the free diagnosis.
Search gave everyone
the same ten links.
AI gives everyone
a different answer.
There is no page one to win any more. There is only the answer your buyer gets — and whether you are in it.
Why this is not SEO
AI already knows who is asking.
Every assistant builds a picture of its user from what they have asked before — their situation, their constraints, what they keep coming back to. Then it answers that person.
So the same question does not have one answer to optimise for. It has as many answers as there are kinds of buyer.
What we optimise instead
B = f(P, E)
Four letters, one idea: what a buyer does depends on who they are and what is going on around them. Get both right and you can write their question before they type it.
Lewin, 1951. We point it at AI search.
Turn it by hand
Change who is asking.
Change who is asking.
Change their situation.
Watch the recommendation change.
P
E
B · what they actually type
“I started running a month ago and my knees ache after every run — do I need different shoes, and which ones?”
Who the answer names
- 1A competitor
- 2You
- 3A review site
Injury questions reward brands that explain cushioning in plain words. Your fit guide does — that is why you are named.
Illustrative: a running-shoe brand. In a real run, the people and their questions come out of your own material — never invented.
The part nobody budgets for
Your buyers ask AI thirty different ways.
Your buyers ask AI thirty different ways.
How many answers name you?
Five kinds of buyer. Two situations each. Three questions in every situation. AI answers each of the thirty on its own — so a brand can own one and be invisible in the next twenty-nine.
Most brands have never counted. The ones that do stop guessing where the marketing budget goes.
AI names youAI names someone else
7of 30 questions name you
Being straight about it
Three things decide the answer.
Three things decide the answer.
One of them is ours to move.
Not oursTheir account history
The profile the AI assistant has already built of that user.
This is the jobWhat it reads while answering
The sources it goes and finds, and decides to trust, in that moment.
Not oursThe model itself
Ask it the same question twice and the answer will not read the same. That variation is normal, and it sits inside the margin we expect and measure.
Then we score what came back
Same questions, same models, every day.
MRMention rateAre we named at all?
OFHow oftenIn passing, or throughout?
ARPositionTop of the list, or ninth?
BCRShare of answerHow much is about us?
One score on one day is weather, not climate. We report the trend — and when a month does not move, we say so.
See how AI answers your customers
Compare your brand’s presence across these 10 AI models and answer platforms. Every report shows which models and questions were tested.
Tencent Yuanbao, Xiaohongshu and Google AI Overviews are not measured automatically. Our team can add visibility scoring for those surfaces. Ask us to add them
Swipe to explore all 10 models
ChatGPT
Claude
Gemini
DeepSeek
Qwen
Doubao
Kimi
Hunyuan
Grok
PerplexityBe the answer,
not the tenth link.
Three things this work is supposed to change. Each one is a number, measured the same way twice.
International83.0
Chinese32.6
Half your buyers ask ChatGPT. The other half ask Doubao, DeepSeek or RedNote — which read a different internet entirely.
How both tracks are scoredA content platform
A news portal
An education site
Your own domain0
In one client diagnosis, the Chinese models cited owned domains zero times. The content was fine. It was in the wrong place.
Where content actually goes
See your own three numbers
A baseline takes one run. It is worth having even if you stop there.
Our service cycle
Six steps, one account manager from the first call. Step six opens the next cycle.
- Week 101
Discovery interview
Your account manager learns the audiences you want to win and collects brand and business material — shared with your permission, anonymised where needed.
Brief & data access - Weeks 1–202
Core audiences, first audit
We build your core buyer personas and the situations they buy in, then run the first audit for a baseline visibility score.
Personas · baseline score - Week 203
Kick-off
Together we agree which audiences and business lines to optimise first, and what success will look like.
Agreed targets & KPIs - Monthly04
Optimisation
Content written for each target persona and situation, placed on the platforms the models actually read.
Articles live where AI reads - Daily05
Track & measure
The same questions, re-run every day. Scores, citations and competitors, reported in plain language.
Report · first within 30 days - Each cycle06
Review & re-audit
What moved and why. Personas refreshed with new market signals, and the next cycle opens with a fresh audit.
Review · next cycle’s plan
Real work. Specific evidence.
Three things worth a closer look: the work, the people, and the research underneath it.
Helping finance applicants find SMU in AI answers
First we measured where SMU stood: which applicant personas and situations AI already recommended it for, and where it disappeared. Then we built a content strategy written for AI to read.
For SMU’s Master of Applied Finance, JefurryAxis mapped applicant needs, generated tailored articles around the decisions those applicants faced, and placed them on content platforms that AI models retrieve from.
69.177.2
AI Visibility Score · 28 Jan–4 Feb 2026
Five of our ten articles were cited as sources — four by Doubao, one by DeepSeek — and only under the target persona’s questions.
Reported by Jeffery Asia across Qwen, Doubao, DeepSeek and Yuanbao for the case’s applicant profiles. A visibility result, not a measure of enrolments or revenue.
Read the full case study → · on jeffery.asia ↗NUS EMBA
A partner in AI search, not just a client of it. NUS EMBA and our team have worked on AEO, GEO and AI SEO together from the early days — testing the method on real applicant questions, reading the results side by side, and sharpening how an executive-education decision gets measured.
Executive education · A long-term AEO and GEO research partnership.
Read the case on jeffery.asia ↗Research accepted at ANZMAC 2026
“Persona-Driven AI Engine Optimization in Zero-Click Search” has been accepted for an oral presentation following peer review, as confirmed in the invitation to Dr. Jia Yun.
Accepted for oral presentation · Conference to be held 30 Nov–2 Dec 2026 in Auckland.
The study behind the paper → · ANZMAC 2026 conference site ↗
Built by Jeffery Asia
JefurryAxis is the AI visibility product of Jeffery Asia, a Singapore growth agency working across education, healthcare, finance and cross-border services since 2016. The research, the measurement method and the delivery team all come from there.
A team drawn from the National University of Singapore, Nanyang Technological University, the Massachusetts Institute of Technology and the University of Melbourne — with deep brand-research experience, long roots in Chinese and Southeast Asian culture, and solid academic and technology research behind every engagement.
Singapore Prestige Brand Award 2025Established Brands · organiser’s announcement ↗
Alibaba Cloud MVP 2025Dr. Yun Jia, founder · profile ↗
IAI Tourism Marketing 2022Bronze Award · company recognition ↗
Organisations we work with
Across education, finance, government and cross-border services
FAQ
Frequently Asked Questions
What is B = f(P, E), in one sentence?
What someone does depends on who they are and the situation they’re in. We use it to predict the questions your buyers ask AI — and check whether you’re in the answer. Lewin’s equation, 1951; the application to AI search is ours.
How is this different from SEO?
SEO gets you a link on a results page. This is about being the recommendation inside the answer — where there is no page one to be seventh on.
How do you measure it?
Brand-free questions — three international models and three Chinese ones, each asked more than once. We score whether you’re named, how often, how high, and how much of the answer is about you, then ask exactly the same questions again, on a regular cycle.
Which platforms do you cover?
ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek, Qwen, Doubao, Kimi, Hunyuan. Yuanbao, RedNote Ask and Google AI Overviews through team-assisted runs. Which ones matter depends on where your buyers ask.
Can you guarantee we’ll be recommended?
No, and nobody can. What we can do is optimise for the specific personas you want to win — and that approach has been tested. Our peer-reviewed research, “Persona-Driven AI Engine Optimization in Zero-Click Search”, found that articles written for one persona were cited only for that persona, while the brand’s AI Visibility Score rose from 69.1 to 77.2 — the targeting holds. And in our RedNote case, the same method took a brand from zero to one: absent from the answer before, recommended and cited after.
How does JefurryAxis relate to Jeffery Asia?
JefurryAxis is the AI visibility product built by Jeffery Asia, a Singapore growth agency. The research, the method and the delivery team come from there.
What does a first engagement look like?
A baseline before a proposal — where you stand, which buyers you’re losing, and which sources the models read to decide. Worth doing even if you stop there.
Be the AI
Recommendation
Every day, people in your market ask AI which brand to trust, and it answers with a name. We help make sure that name is yours — starting with a clear view of where you stand today.