← The Growth Lab · AI SHOPPING
From Keywords to Conversations: How Shopper Intent Is Changing
Oct 14, 2026 · Read time: 3 min · DV8 Pacific
Search advertising was built on the keyword. A shopper typed two or three words, and brands bid to appear next to them. AI assistants work differently. Shoppers do not type keywords into a chat. They describe their situation, and that changes what brands need to understand about their customers.
A search is a query; a conversation is a brief
Compare two requests. In a search box: "oyster sauce." In an AI assistant: "What do I need to make beef chow fun for four people tonight? I want to avoid MSG." The first tells you the product category. The second tells you the meal, the number of people, the timing and a constraint. An AI assistant uses all of it to decide which products to suggest.
Conversational requests carry several layers of intent at once: the underlying need, the context, the constraints, personal preferences, the criteria for deciding, and how close the shopper is to buying. Some of this is stated outright. Much of it is implied and only becomes clear as the conversation continues.
Think in need states, not keywords
The most useful way to organize this is around need states: the specific situations in which a product is the right answer. Take a matcha brand. Keyword thinking produces a list of variations on "matcha powder." Need-state thinking produces a different list: people looking for a coffee alternative, people wanting afternoon energy without another coffee, beginners trying matcha at home, and people who want café-style lattes in their own kitchen.
It is the same product in four different situations. Each calls for a different message, a different product page and, in AI advertising, different guidance to the platform about when the ad is relevant. ChatGPT's ad system already works this way. Instead of buying keywords, advertisers write "context hints" that describe what the product does, who it helps and when it is useful, and the platform matches that description to conversations.
Why this matters more for Asian brands
Many Asian products need context for US shoppers. A sauce, snack or skincare product that needs no explanation in its home market may need a recipe, an occasion or a comparison to make sense in the US. Need-state thinking forces that translation. It shifts the question from "what is our product called" to "in which American situations is our product the best answer."
How to build a need-state map
Start by collecting the real questions shoppers ask about your category. Customer service logs, reviews, retailer Q&A sections and community forums are good sources. Group the questions into need states, and for each one write down the context, the constraints and what a shopper would need to know to choose your product.
The map is useful in several places at once. It shows which situations your product pages should address so AI assistants can match your product to them. It gives AI ad campaigns a structure, with one group of ads per need state. And once campaigns run, results by need state show which situations bring the most valuable customers, an insight that can shape positioning, packaging and product development.
Sources: OpenAI, Ads Help Center guidance on ad groups and context hints, 2026; DV8 Pacific analysis.