AI chat assistants are being used very differently than many marketers assume. Rather than serving mainly as shopping companions or direct search substitutes, they are primarily tools for cognitive work — writing, planning, analysis and learning — according to a study from AI SEO agency Dejan. In 2026, “optimizing for AI” requires understanding how people actually interact with assistants, instead of presuming they behave like buyers in a store.
Dan Petrovic, director of Dejan, examined 3.9 million conversational turns — representing 613 million words and 4.4 billion characters — to identify real-world usage patterns. His findings upend several common beliefs about AI-driven content strategy.
Usage patterns: most chats are brief and task-focused
- Typical chat length (median): 2 turns (one prompt and one reply)
- Median session size: 430 words
- More than 80% of sessions stay under 1,000 words
- Only 4.2% of sessions go beyond 2,500 words, usually for editing, summarizing, tutoring, coding or data analysis
- Average session length: 732 words, pulled upward by long document inputs
- Assistant output volume: roughly 1.5 times the user’s word count
Petrovic observed that user input typically represents just 16–17% of the total session content, with the assistant producing the vast majority of the text.
Dig deeper: What happens when no one clicks anymore
Intent breakdown: commercial behavior is a minority
Petrovic reviewed 24,259 labeled sessions across 42 intent types. He found that 64.6% of all sessions showed no commercial intent at all. Most people turned to AI assistants for activities such as:
- Brainstorming (7.7%)
- Planning (6.5%)
- Emotional support or casual conversation (6.2%)
- Analysis (5.7%)
- Learning (4.7%)
- Text transformation, including summarization and translation (4.6%)
- Content production (3.9%)
Even within the 35.4% of sessions that did involve commercial intent, most were at the top or middle of the funnel:
- Awareness phase: 10%
- Consideration: 8.5%
- Discovery plus decision support: 6.9% combined
- Transactional assistance: 4.8%
- Post-purchase help: 5.1%, such as troubleshooting and how-to guidance
Key insight: AI chats are cognitive workflows, not simple queries
Many marketers and SEOs still approach AI with a search-centric mindset, assuming conversations resemble keyword queries. Petrovic’s research indicates the opposite. People mainly use assistants to support multi-step thinking and execution, not to jump straight into buying.
“AI chat use is overwhelmingly non-commercial,” Petrovic said. “Users see assistants as co-pilots — not salespeople.”
Implications for AI content strategy
Marketers should rethink how they optimize for AI by:
- Focusing on content that serves awareness and early-stage exploration
- Developing structured, richly contextual resources that agents can reference and reuse in longer workflows
- Reducing reliance on transactional keyword tactics in AI-oriented content
Bottom line: Winning AI visibility isn’t about forcing intent — it’s about aligning with it. Assistants are evolving into tools for thinking, not just for closing sales. Content that helps people learn, plan, write, compare or evaluate is far more likely to surface in real-world AI interactions.
The report. How do people use AI assistants?
Fuel up with free marketing insights.
Email:
MktoForms2.loadForm(“https://app-sj02.marketo.com”, “727-ZQE-044”, 16293, function(form) {
// form.onSubmit(function(){
// });
// form.onSuccess(function (values, followUpUrl) {
// });
});
The post Why AI content strategies need to focus on tasks not transactions appeared first on MarTech.