AI can now review every call, every email and every deal, and give feedback at a scale no manager can. The research shows it helps — but not everyone equally.
What AI coaching looks like
- Call scoring against your methodology (discovery questions, next steps, talk ratio).
- Suggested improvements after each call.
- Role-play practice with an AI buyer for objection handling and new products.
- Deal-level prompts: missing stakeholders, no next meeting, unclear budget.
What the research found
AI coaches help the middle most. In a field experiment published in the Journal of Marketing, Luo, Qin, Fang and Qu found an inverted-U pattern: middle-ranked sales agents improved most with an AI coach, while bottom- and top-ranked agents gained little. Bottom performers were overloaded by the volume of AI feedback; top performers resisted it. Agents coached by a combination of AI and human coaches did better than with either alone.
AI assistants help newer people most. In customer support, Brynjolfsson, Li and Raymond found a generative AI assistant raised productivity 14% on average and 34% for novice and low-skilled workers, with minimal impact on the most experienced. The AI spread the know-how of top performers to newer staff.
What this means in practice
- New hires: AI coaching and role-play can shorten ramp; pair it with a human who filters the feedback to one or two priorities.
- Middle performers: the biggest gains — give them the tools and the time.
- Top performers: don't force it; ask them to help shape the AI's playbook instead.
- Managers: use AI for the data (what happened on calls), and spend your time on the conversation (what to change and why).