Search Intent is not Task Intent

Intent is the most desirable data point for advertisers. With advertising in AI becoming the next big category, advertisers need to understand how different intent looks in AI versus search.

Digital advertising has categorized users by demographics, interests, and search keywords for two decades. But users don't interact with AI assistants the way they use Google. They're delegating entire workflows or using assistant as thought partners. This behavioral shift breaks traditional targeting taxonomy.

The Usage Data

It was 23 years ago that the categories of informational, navigational, and transactional search were coined by Andrei Broder. This was later expanded to a fourth “commercial” by Google. It’s tempting to fit AI search into the same categories, but it would be far from the truth. AI isn’t search, it’s a new paradigm for interacting with information.

Analysis of millions of conversations reveals distinct usage patterns: non-work messages now comprise 70% of ChatGPT usage (up from 53% in June 2024), while coding accounts for 36% of Claude usage versus 4.2% for ChatGPT. Users select specific assistants based on task categories, not information retrieval.

Anthropic's Economic Index analyzing 4+ million conversations found 57% augmentative behavior (collaborative work) versus 43% automative behavior (full task delegation), with increasing directive usage where users hand off complete workflows.

Search Intent vs. Task Intent

Search intent

• User: "best running shoes"

• Advertiser targets: keyword "running shoes," demographic male 25-34, interest in fitness

• Context: single query, no prior relationship

Task Intent

• User: "I'm training for a marathon in 6 months. Can you make me a training plan as well as an equipment list? I specifically need new shoes to help with overpronation, budget $150"

• Advertiser must target: fitness goal (marathon training), timeline (6 months), medical requirement (overpronation), budget constraint ($150)

• Context: multi-turn conversation, accumulated user needs

The targeting taxonomy must shift from attributes to task parameters. Users are articulating goals, constraints, and requirements across conversation threads.

Why Traditional Targeting Categories Break

Traditional ad platforms categorize by demographics, interests, and search intent (informational, transactional, navigational).

AI assistants reveal user needs through task objectives, resource constraints, workflow stages, and multi-session context.

Example: A user asking "help me plan meals for the week" reveals dietary constraints, cooking skill level, budget preferences, family size, and time availability, all from conversational context.

Traditional targeting would use demographics and interests. Task-level targeting uses accumulated conversational context to understand the specific workflow being delegated.

What This Means for Advertising

The data available in AI assistant workflows gives advertising a better chance of landing relevant recommendations. At Koah, we're building targeting infrastructure for task-based workflows that captures minimal personal information while recommending highly relevant content based on workflows (meal planning, trip research, code debugging) rather than demographic interests.

Early results: Our ads achieve 6% CTR on mobile and 1-2% on web because they're matched to active tasks, not demographic profiles. When someone is delegating meal planning for a diabetic diet, an ad for low-glycemic recipes directly supports their workflow.

Users interact with AI assistants as workflow delegation tools, not search engines. Advertising taxonomy must evolve from demographic and keyword targeting to task parameter matching. Traditional targeting categories won't work when users are delegating complex workflows rather than expressing simple search queries.

Koah builds AI-native advertising infrastructure for task-level intent. If you're monetizing an AI app, reach out here.