1. Marketing explained
Eleven scenarios at the same party. Guess the term before the reveal.
Scenario 1
You see someone attractive at a party. You walk over and say:
“I’m rich. Marry me.”
Direct marketing
Scenario 2
You get the person’s telephone number from someone else and call to say:
“I’m rich. Marry me.”
Telemarketing
Scenario 3
You walk around the party telling everyone:
“I’m rich. Marry me”
Mass marketing
Scenario 4
You email everyone at the party, sending them all the same message over and over again saying:
“I’m rich. Marry me.”
Spam marketing
Scenario 5
You give your friend a hundred dollars. Your friend walks over and says:
“That person over there is rich. You should marry him.”
Advertising
Scenario 6
Your friend willingly walks over and says:
“That person is rich and will make a good husband.”
Public relations
Scenario 7
The attractive person hears from her friend:
“I heard that he’s rich and will make a good husband.”
Word-of-mouth
Scenario 8
The attractive person walks over to you and says:
“I hear you’re rich.”
Brand recognition
Scenario 9
You do not say anything.
You arrive in a chauffeur-driven Rolls-Royce and make a large charitable donation during the party.
Positioning
Scenario 10
You do not say anything.
All the available attractive people Google your name.
SEO marketing
Final scenario
They search your name and ask:
“Is he a good catch?”
AI agent search
“We used to market to the person at the party. Increasingly, we are also marketing to the intelligence standing between us and that person.”
-- the thesis of this lecture
2. From 4Ps to Purple Cow (the basics)
Three businesses, three different battles -- the frameworks win different kinds of fight, they do not simply replace one another.
Three frameworks, three battles
| Framework | The battle | Source of advantage |
|---|---|---|
| 4Ps | Coordinating a market offer | A coherent commercial system |
| 7Ps | Delivering an experience consistently | Operational excellence |
| Purple Cow | Becoming impossible to ignore | Remarkability and conversation |
“Each framework wins a different fight -- the question is which fight you are actually in.”
Example 1 · The 4Ps
Kellogg’s Corn Flakes
Product, price and promotion were competent. Placement -- being in the aisle, at eye level, in every grocer in the country -- was what made the category.
Example 1 · The 4Ps
Even the master gets a P wrong
- India, 1994
- Priced at nearly double the local competitor, distributed only in metro stores, and promoted as a replacement for idlis and vadas.
- The result
- Sales fell 25% within months, before a full repositioning fixed it. One misjudged P dismantled the other three.
“Which P is most frequently delegated in your organisation, even though it can destroy the other three?”
Example 2 · The 7Ps
Singapore Airlines
- Product
- The route, seat, cabin class and journey
- Price
- The fare architecture and premium
- Place
- Booking channels, route network, partnerships
- Promotion
- Advertising, reputation and loyalty communication
- People
- Cabin crew and customer-facing employees
- Process
- Booking, check-in, boarding, service recovery, baggage
- Physical evidence
- Aircraft, lounge, uniform, menu, cabin, digital interface
People, process and physical evidence are the product
- A confusing booking flow contradicts a calm, premium promise.
- A poor disruption process destroys trust instantly.
- An indifferent employee becomes the brand.
- A dirty lounge invalidates millions spent on advertising.
“Our promotion promises ______, but one of our processes communicates ______.”
Example 3 · Purple Cow
Tesla and the electric car category
- The conventional belief
- A responsible car requires sacrifice. Tesla’s distinction was not merely “electric” -- it made electric performance conspicuous, giving owners a story worth retelling.
- What that bought
- Owners became demonstrators and advocates, not just customers. Performance challenged category assumptions and sparked conversation well beyond typical car buyers.
“A Purple Cow does not mean being eccentric. It means giving the right audience a reason to remark.”
The progression
- 01 -- 4Ps
- Construct a commercially coherent offer.
- 02 -- 7Ps
- Make the organisation deliver the promise.
- 03 -- Purple Cow
- Make the delivered promise worth discussing.
- 04 -- AI era
- Make the evidence discoverable, understandable and credible to machines acting for humans.
“The Purple Cow assumes people can see the cow. What happens when an AI stands at the entrance to the field and decides which three cows the customer should consider?”
3. The Forbes transition
What happens to a century-old institution when the mechanics of discovery change underneath it.
Transition case
Forbes
Founded by B.C. Forbes in September 1917 -- over a century old, famous for its billionaires list and “30 Under 30.” Unlike a search-dependent startup, Forbes is exactly the kind of legacy institution executives assume is safe from disruption.
What happened
40%
YoY traffic decline across the top-50 news domains (Similarweb)
50%
Google traffic decline across 2,500+ sites (Chartbeat)
59%
YoY visits down, per the Forbes CEO -- to 75.4M monthly visits
Forbes CEO Sherry Phillips confirmed to Press Gazette (Nov 2025) that AI Overviews providing lengthy answers atop search results meant users no longer felt the need to click through to the original source -- hitting programmatic ad revenue specifically.
The key distinction
- Not this
- “Forbes failed because it refused to adapt to AI.”
- What actually happened
- Forbes acquired True/Slant in 2010 specifically to become “digitally led” and built an open contributor network that multiplied content volume and search footprint -- a strategy that worked brilliantly for the search-engine era.
“Forbes solved the previous distribution problem so well that its entire model became a bet on people continuing to click through from search.”
The response
Sherry Phillips has reframed strategy around loyal audience communities and direct relationships, diversifying revenue toward live events, branded content and licensing -- areas an AI Overview cannot intermediate.
“A century of brand authority did not protect Forbes from a change in the mechanics of discovery. What protects it now is value that lives outside the search click.”
4. Live search experiment
Google AI Overviews and AI Mode are both live in Singapore. Rather than describe the shift, we will search together, right now.
Stage 1 · Predict
“Where is the best place to propose in Singapore?”
- Which three locations appear?
- A commercial venue, or a public location?
- What evidence is used?
- Which publications or sites get cited?
- Who can appear at all?
- Can any business surface without paying for an ad?
Stage 2 · Search live
Examine the page top to bottom
- Is there an AI Overview above the organic links?
- Which places are named, and which sources are cited -- official sites, publishers, Reddit, review sites, travel platforms?
- Which businesses benefit even without their own site being cited?
- How much of the decision can be made without a single click?
The exact output varies by account, location and time -- that variability is itself part of the lesson.
Stage 3 · Increase intent
“What’s the best place for steak in Singapore?”
Before searching -- what does “best” mean here? Budget, location, ambience, halal, cut, occasion, group size.
- “Best steak restaurant for a quiet anniversary dinner.”
- “Best steak under S$100 per person near Marina Bay.”
- “Best halal steak restaurant in Singapore.”
- “Compare three options on atmosphere, price and recent reviews.”
Stage 4 · The funnel changed
Who performs the integration?
- Traditional search
- Search → open links → read reviews → compare prices → shortlist → decide. The user integrates.
- AI search
- Interpret context → search subtopics → synthesize evidence → eliminate → shortlist → answer objections. The AI integrates.
“The marketer is no longer competing only for a click. The marketer is competing to become part of the AI’s consideration set.”
Stage 5 · Working backwards
Why did this business appear?
- Reviews
- Strong review volume, with consistent facts across platforms.
- Third parties
- Mentions in reputable publications, and recent discussion.
- Basic facts
- Clear location, price and service information.
- Machine legibility
- Strong category association, machine-readable information.
“Which department owns this outcome?” -- no longer just marketing: PR, customer experience, operations, data, web engineering, partnerships and reputation all appear in the answer.
“Google did not merely find documents. It formed an opinion -- it interpreted ‘best,’ gathered witnesses, evaluated alternatives, and created a shortlist.”
-- the move from search engine to answer engine
Two discovery systems
| Dimension | Traditional SEO | AI answers & agents |
|---|---|---|
| Primary unit | Webpage | Answer, claim, entity or product |
| User behaviour | Short keywords | Detailed conversational requests |
| System output | Ranked links | Synthesized recommendation |
| User’s role | Researcher and integrator | Delegator and evaluator |
| Marketing objective | Rank and win the click | Enter the consideration set |
| Typical query | “Steak Singapore” | “Quiet steakhouse under S$100 near Marina Bay” |
| Value of third parties | Backlinks and referral traffic | Corroboration, authority, sentiment |
| Success measure | Rank, impressions, clicks | Mentions, citations, recommendation share |
| End state | User visits a website | Agent may act without a website visit |
5. AI marketing principles
Discovery, trust, dual-audience writing, the long tail, machine legibility and measurement.
Principle 1 · Discovery
From ranking to consideration
A brand can rank well for its own name and still be absent from the unbranded questions that decide purchases:
- “Best accounting platform for a five-person Singapore consultancy.”
- “Alternative to Salesforce for a regulated SME.”
- “Best chemistry learning platform for an IB student who struggles with calculations.”
A company may win category prompts but fail once the user adds realistic constraints -- which is why AI discovery is a long-tail problem.
The new discovery map
- 01 -- Category
- Best solutions in this category?
- 02 -- Problem
- How should I solve this problem?
- 03 -- Comparison
- Product A vs B for my situation.
- 04 -- Constraint
- Best option under a budget or regulation.
- 05 -- Risk
- What are the disadvantages?
- 06 -- Validation
- Is this company trustworthy?
Principle 2 · Trust
The AI as a cautious recommender
AI trust ≈ claim clarity × independent corroboration × source authority × consistency
A conceptual model, not a published score. The multiplication matters: if any one factor is near zero, the recommendation becomes fragile.
AI-visible credibility evidence
- Named people
- Executives and experts with verifiable credentials.
- Independent coverage
- Media mentions and industry-association listings.
- Case studies
- Named organisations with measured outcomes.
- Original research
- Proprietary data nobody else can generate.
- Transparency
- Published pricing, limitations and policies.
- Consistency
- Matching product data across site, marketplaces and directories.
Most cited sources are not yours
of AI-surfaced sources in many CPG and finance searches come from publishers, user-generated content and affiliates -- not the brand’s own site (McKinsey). A slogan is a claim; a customer result, named case study, independent review or credible citation is evidence.
Principle 3 · Two readers
Writing for humans and machines
| Humans need | AI systems need |
|---|---|
| Relevance | Clear topic identification |
| Explanation | Extractable answer passages |
| Confidence | Verifiable claims |
| Narrative | Logical structure |
| Judgment | Comparisons with stated criteria |
The answer-first pattern
- Weak
- “Our decades of passion and commitment have allowed us to create world-class, transformative solutions for the modern enterprise.”
- Stronger
- “We provide a multi-tenant school-management platform for private K–12 schools in Southeast Asia -- admissions, billing, attendance, parent communications, role-based access.”
Information gain
AI can generate generic explanations cheaply. What differentiates you is what a generic model cannot independently invent:
- Original data
- Surveys and internal benchmarks.
- Experience
- First-hand implementation detail.
- Named cases
- Case studies and proprietary methods.
- Position
- Contrarian but defensible conclusions.
“The goal is not more content. It is more uniquely useful evidence.”
Principle 4 · The long tail
The long tail matters more, not less
- Old query
- “CRM software”
- AI-era query
- “Which CRM suits a ten-person Singapore consultancy that uses Microsoft 365, requires data export and has no full-time administrator?”
The second query carries substantially more purchase context -- and generative engines match this well via semantic similarity.
Build content around combinations
- Persona
- Problem
- Industry
- Geography
- Budget
- Occasion
- Regulation
- Integration
“Does this page contain enough evidence to answer the user’s specific situation, or did we merely insert their keywords into generic copy?”
Principle 5 · Machine legibility
The invisible storefront
- Structure
- Descriptive titles, headings and structured product data.
- Real questions
- FAQ content built from questions customers actually ask.
- Accessible HTML
- Facts in text, not locked inside images.
- Stability
- Stable URLs, with pricing and policies kept current.
Not a trick to manipulate a model -- it reduces ambiguity and makes accurate retrieval easier.
Cross-functional ownership
- Marketing
- Positioning, questions, content, measurement.
- PR
- Independent authority and media evidence.
- Engineering
- Crawlability, structure, feeds, data quality.
- Customer success
- Reviews, cases, authentic customer language.
- Leadership
- Expert visibility and institutional credibility.
AI marketing cannot live exclusively inside a content team.
Principle 6 · Measurement
The new executive dashboard
- AI mention rate
- Share of tracked prompts where the brand appears at all.
- Recommendation rate
- Share of prompts where the brand is positively shortlisted.
- Citation share
- Share of cited sources that support the brand.
- Sentiment
- Positive, neutral, mixed or negative framing.
A measurable monthly routine
- 13% → 32%
- Semrush’s tracked AI share of voice, over two months of targeted content changes (vendor case study).
- Select 30–50 commercially important prompts.
- Test across Google, ChatGPT, Perplexity and Gemini.
- Record mentions, citations, sentiment and competitors.
- Correct gaps, create the missing evidence, repeat monthly.
6. Collective credibility
Help one another become truthfully easier to verify -- legitimate, distributed evidence, not coordinated fake reviewing or manufactured consensus.
Credibility actions
- Expert contribution
- Real insight in articles, panels and podcasts -- not generic endorsement.
- Legitimate citation
- Cite because it helps the audience, not because of a reciprocal deal.
- Case-study partnerships
- Document problem, criteria, results and limitations honestly.
- Data collaboration
- Pool anonymised data for original, citable industry benchmarks.
- Visibility audits
- Quarterly tests of each other’s key prompts across AI platforms.
- Warm introductions
- To journalists, associations, podcast hosts and standards bodies.
The quarterly visibility audit
- Is the company mentioned at all -- and is it described accurately?
- Which competitors appear instead, or alongside?
- Which sources influence the answer?
- What evidence is missing, and what negative or outdated information is persisting?
AI outputs vary between runs -- track trends, not one single answer.
7. Commitment and close
One important, verifiable fact -- made easier to discover within thirty days.
Within 30 days
“What is one important, verifiable fact about your organisation that you will make easier to discover?”
- 01 -- The fact
- 02 -- The evidence
- 03 -- Where it appears
- 04 -- Who can help
Three eras, three questions
- Then
- “How do we reach the customer?”
- Digital
- “How do we rank when the customer searches?”
- Now
- “When an intelligent intermediary investigates the market, will it find enough evidence to trust us, understand us and recommend us?”
“Visibility can be purchased -- but credibility has to be constructed. And the organisations in this room can help one another construct it.”
“The customer’s agent asks whether you are really rich, reliable and worth meeting. Your advertisement says yes. Your website says yes. But the agent checks everybody else.”
-- already happening
