What does it mean that shared mental models enable coordination?

Example: An incident response team at a SaaS company was paged at 2:14 AM. Within twelve minutes, five engineers were online and working the problem. No one assigned roles. No one asked 'What should I do?' The senior SRE began checking upstream dependencies. The database engineer queried recent migrations. The application engineer pulled logs from the affected service. The team lead opened the incident channel and began documenting timeline. The on-call product manager notified affected customers. They coordinated seamlessly not because they had rehearsed this specific outage but because they shared a mental model of incident response: who does what, in what order, with what authority. Compare this to a cross-functional project team at the same company that spent forty minutes in a planning meeting debating who was responsible for writing API documentation. Three people thought it was the backend engineer's job. The backend engineer thought it was the technical writer's. The technical writer thought it was the product manager's. The team had no shared mental model of documentation ownership — and the absence of that shared model converted a five-minute task allocation into a forty-minute negotiation. Shared mental models are the invisible infrastructure that makes coordination cheap. Their absence makes coordination expensive.

Try this: Select a recurring team process — a deployment, a sprint planning, a design review, or an incident response. Interview or survey three team members independently, asking each to describe: (1) the steps in the process, in order; (2) who is responsible for each step; (3) what triggers the process to begin; and (4) what criteria determine that the process is complete. Compare the three descriptions. Where they align, you have a shared mental model. Where they diverge, you have a coordination cost that the team is paying every time the process runs — through negotiation, confusion, rework, or conflict. Document the merged model and share it with the team. Ask: 'Is this how we actually work? Is this how we should work?'