From Paid Reach to Earned Inclusion
Rethinking Brand in the Age of AI
Mark Sage - 11 min read - 13/06/2025
Google CEO Sundar Pichai recently described our current moment as a platform shift “bigger than the internet.” His comment followed Google I/O 2025, where the company unveiled a suite of AI-driven features that radically reshape how we search, discover, and interact digitally.
Search has already evolved dramatically — from the early days of static directories like Yahoo, to today’s keyword-driven engines that crawl the web for content. But conversational AI is now redefining the model entirely. Tools like ChatGPT and Gemini don’t just retrieve results — they infer our intent. They aren’t simply scanning for matching keywords; they’re drawing on humanity’s collective knowledge to anticipate what we really mean.
For marketers, this shift is profound. Visibility is no longer about bidding for attention or gaming SEO — it’s about having something relevant to say.
Context is now king.
Whether a consumer is using search, a shopping app, or browsing a recipe site, agentic AI is increasingly doing the heavy lifting — interpreting intent and returning actionable answers. Soon, we won’t search for products. We’ll just ask for what we need, and the agent will decide what shows up.
This isn’t just a UI improvement — it’s a structural shift.
This article explores how intent-rich, needs-based queries — previously typed into Google — are now interpreted by conversational AI; and importantly, what that shift means for brands.
At the core of this are two converging forces reshaping the role of AI in discovery:
The growing importance of content, context, and intent alignment in AI-driven interactions.
The rise of agentic platforms that filter, automate, and increasingly replace human decision-making.
To understand these forces better, a number of queries/prompts were used for everyday categories within the Hong Kong market — a compact, high-density environment of 7.5 million residents, combining modern retail infrastructure with an active general trade ecosystem.
The objective was to see how today’s AI interprets and responds to real-world prompts and uncover the hidden logic behind what gets surfaced, what gets skipped, and why.
So, lets jump in!
In the era of conversational AI, discovery is no longer driven by keywords. Users now ask fully formed, intent-rich questions like “What’s a good suncream I can use every day?”
These aren’t keywords — they’re fully stated needs or ‘jobs to be done’ that express both the core ask (“I want suncream”) but also underlying intention (“Recommend me a lightweight, effective, non-irritating suncream suitable for daily use in a hot and humid climate.”).
The search query is now a prompt, and the search results are now more nuanced. Not simply a list of links that might match your keywords – but a genuine attempt to answer your question within a single response.
In traditional search —
Brands compete to rank, to be clicked, and to persuade
In the agentic world —
Agents pre-filter options based on user needs, data, and efficiency.
The agent is now making the choice — it’s trying to smartly infer what you want to do, and then providing a comprehensive, single answer to it. That answer is not just information though — increasingly it includes the ability to carry out the action — placing the order, cancelling the service or making the enquiry.
This means you’re not persuading a human — you’re informing an algorithm.
And it changes the marketing question from “How do we stand out?” to “How do we avoid being filtered out before we’re even seen?”
It also isn’t some future state or pipedream — Google ‘Buy for Me’ or OpenAI’s Shopify integration already show how this is going work.
So, showing up is not simply a matter of prioritisation, it will soon be a matter of whether you have any chips in the game.
To understand how AI actually handles real-world prompts, a set of common “jobs to be done” queries were used to test it — asking not just for product recommendations, but also where to buy.
The goal wasn’t only to see who appeared in the results, but to uncover why.
This wasn’t intended as a scientific study or technical audit. Instead, it was a practical exploration, with the prompts reflecting how real consumers might ask for help in everyday situations, and focusing on how AI interprets intent.
The prompts used were:
- “Where can I buy fresh chicken?”
- “How do I make sweet and sour pork and where can I get the ingredients?”
- “What’s a good suncream I can use that I can use every day?”
- “My daughter is 1 year old — what would be the best swimming nappy for use in the pool”
Across these prompts, the results were striking — major retailers were often absent or referenced only generically (“supermarkets”). Instead, specialist retailers were prioritised and named, and eCommerce marketplaces beat out branded retail sites.
For large retailers this is a critical blind spot — you’re essentially being ignored.
There are three core reasons for this:
Weak eCommerce content — Pages are structured for transactions, not discovery.
Lack of contextual framing — No narrative around why a product matters or how it’s used.
Inferred intent bias — The AI interprets what’s not said as much as what is.
These challenges really all come down to how the AI is making some initial decisions based on the question to create a wider scope — basically taking the prompt and embellishing it to try and guess the ‘latent meaning’.
Prompt — “Where can I buy fresh chicken?”
What the AI ‘Heard’ — “Probably already knows about supermarkets. Looking for fresher, better-quality chicken — maybe organic or wet market-sourced.”
Prompt — “What’s a good suncream I can use every day?”
What the AI ‘Heard’ — “Recommend something lightweight, non-greasy, and good for Hong Kong humidity — ideally locally available.”
Prompt — “Best swimming nappy for 1-year-old”
What the AI ‘Heard’ — Suggest a trusted, parent-recommended, possibly reusable swim nappy that’s well-suited to active babies and found in specialty stores.”
This intent bias shapes the response, which can in turn filter out familiar, mass-market options. The model assumes that if you’re asking, it’s because you don’t want the obvious.
For example, in response to:
“How do I make sweet and sour pork and where can I get the ingredients?”
The AI interpreted this as: “I want this dish to turn out well. I don’t usually cook it. So give me the best ingredients and sources -not just supermarket options.”
That logic leads it to recommend premium meat retailers over chains like Wellcome or PARKnSHOP. Similarly, with the swimming nappy, the word “best” — meant as a casual prompt — was read literally, returning top-rated specialty products over basic disposables.
Why does this happen? Well, it all comes down to contextual storytelling.
The Hong Kong speciality retailers like Healthy Express, Farmers Market, and Streamline Sports, actively framed their products with narrative and use-case value — origin stories, health claims, recipe integration. These content signals gave the AI confidence they’re relevant to the inferred need.
Healthy Express Example
Meanwhile, the larger retailers relied on familiarity and foot traffic. But agents don’t reward familiarity (or allow it to be bought — for the moment) — they reward structured meaning.
In other words, if you’re not embedded in the story, you’re not included in the response.
AI models like ChatGPT, Gemini, and Claude are designed to be helpful, not promotional. That means their answers are built around usefulness, not market share.
When choosing what to surface, the AI models use three foundational principles that underpin the content selection: -
1. Narrative relevance — AI favours content that matches the user’s intent in a real-world context -not just keywords. In short, LLMs are trained on stories, not slogans.
Weak Match — “Nam Ngư is Vietnam’s #1 fish sauce. Buy online now.”
Strong Match — “My grandmother always used Nam Ngư when she made thịt kho tàu — it gives the pork a slightly sweeter depth you don’t get from others”
2. Semantic relationships — AI doesn’t look for exact matches. It builds a conceptual map: “Nam Ngư” → “Vietnamese cooking” → “braised pork” → “home recipes”. Brands that appear within those connections are far more likely to surface
3. Use-case integration — The model prioritises products as solutions. It promotes items that clearly show how, when, and why they’re used — especially when paired with recipes, how-to guides, or community commentary.
In the tests, this played out clearly.
Even though the Hong Kong mainstream retailers like Wellcome, PARKnSHOP, Watsons and Mannings stocked the items in question, they were rarely prioritised. Instead, the agent surfaced local sites like HKTVmall, Baby Central, or specialist blogs — not because their prices were lower or their reach broader, but because their content was clearer, more structured, and more relevant to the task.
This isn’t simply about SEO. It’s about how easily your data can be understood by the AI and how clearly your brand fits into the story the agent is trying to tell.
In this new landscape, AI doesn’t care who sells the most — it cares who solves the problem best.
This means, if your brand isn’t discoverable in that context then it doesn’t matter how dominant you are offline. You’re “agentically” invisible.
Brand Marketing is Key
Fixing technical content gaps is necessary, but it’s not sufficient.
The deeper issue revealed in these tests is one of storytelling. Whether the search is for chicken or suncream, AI doesn’t just want a product listing — it wants context. And this is where brand must evolve.
Traditionally, brand building was about mental availability — gaining broad awareness through media reach. That still matters, but in an AI-first world, it’s no longer enough.
We’ve reached a brand inflection point.
Today, your brand must speak to two audiences simultaneously:
Consumers, who respond to emotion, identity, and habit.
Agents, who respond to structure, clarity, and contextual relevance.
Agents aren’t persuaded by creativity or emotional resonance. They’re not moved by colour palettes or slogans. They filter. They prioritise. And they choose based on whether a brand makes sense in context.
That means brands must now embed their story not just in ads — but in data. Not just in memory — but in meaning. Success comes from creating semantic associations that AI can recognise and retrieve. Distinctive assets still matter, but so does a deeper consistency, which means repeating your story across content, channels, and structured metadata until it becomes machine-readable.
If you don’t tell your brand story in a form AI can understand, no agent will tell it for you.
The challenge is that modern marketing has become obsessed with paid attention — banner ads, search ads, social ads. These tend to dominate strategy because they’re measurable — but they’re also disposable. They disappear when the budget dries up.
But for AI, it’s brand stories that matter.
Brand is now a data signal
AI agents don’t “feel loyal” to a brand — they weigh it, rank it, or exclude it based on the data they’ve been trained on. This means your brand equity has to be encoded in the data.
Context is now a marketing channel
AI doesn’t respond to what you paid to place, and instead it responds to what’s embedded in context. Where and how your brand appears in content has become the primary way it’s surfaced by AI.
Equity is now algorithmic
Brand equity in this world becomes a composite of digital signals, not recall surveys. Its strength is no longer what people say they remember but it’s what agents choose to include.
This essentially rewrites the rules of visibility:
Familiarity no longer ensures presence.
Availability is irrelevant without contextual relevance.
Visibility is driven by structured content and intent-matched messaging.
To compete in this new world, brands must get back to what matters — being relevant, being useful, and being part of the story. Because that’s what agents are now trained to find.
Loyalty has a role to play
As we’ve seen, AI can misread intent based purely on language. When I ask for the “best swim nappy,” I’m often not asking for the top-rated product as reviewed by parents, I’m simply looking for a useful suggestion. But the AI interpreted “best” literally, prioritising premium or niche options that may have been misaligned with my actual need.
This reveals a central tension in agentic commerce — language is imprecise, but AI is literal. Without behavioural context, even the smartest agents risk misunderstanding what we actually want.
And that’s where one of the most undervalued assets comes in — the loyalty programme.
In an agent-driven world, context is everything, and loyalty data provides it. Purchase patterns, price sensitivity, brand preferences, repeat frequency — all of it can enrich AI models and bridge the gap between what’s said and what’s meant.
Historically, we’ve used loyalty data to drive segmentation and push offers. But now it can do more. It can serve as the foundation for a digital twin — a dynamic, behavioural profile that helps AI understand each customer more personally and more precisely.
That means knowing when “best” means most affordable, most convenient, or most familiar — not most awarded. It means knowing which brands a customer trusts, which categories they shop regularly, and how often they repurchase. And it means tailoring responses not just to the prompt, but to the person behind it.
In this new landscape, loyalty isn’t just a CRM tool — it’s the context engine that makes brands relevant at the moment of decision.
We’ve spent decades learning how to build brands that people remember. But in an agentic world, being remembered is no longer enough. You also have to be understood by machines.
AI doesn’t reward awareness.
It rewards relevance.
That means the brand that shows up is not the brand that shouts the loudest, but the one that best answers the question. And increasingly, that question isn’t being asked by a human. It’s being asked for them by their agent — an agent that is looking beyond the question itself to infer more meaning.
This is the real disruption — In this new paradigm,
familiar names risk fading digitally if they aren’t linked to the right context.
Brand equity will be judged not simply by perception surveys or media budgets, but by how well your story has been embedded in the language, structure, and data that AI uses to make decisions.
This isn’t just an existential crisis for search companies like Google —
It also creates urgency for brand marketers.
We need to extend our brand thinking from being a set of assets to be managed to also now include a network of meanings to be made machine-readable.
We need to see the value sitting within our loyalty programme — more than simply a CRM tool, it now has the potential to be a context engine, powering a digital twin for a loyal customer (and their look-a-likes)
We need to change our performance marketing narrative from paying for visibility to earning inclusion — because if you’re not in the decision logic, you’re not even in the race.
Brand marketing in an AI-powered world doesn’t mean abandoning emotion, creativity, or storytelling. But it does mean evolving those stories. Structuring them, surfacing them, and embedding them into the semantic frameworks that intelligent agents will increasingly rely on.
Tomorrow’s consumer journey won’t start with your product or service — it will start with their prompt.
And the question every brand must now answer is…
Are you even in the response?