"Deep work is not passive. It is not something that happens to you."

Deep work blocks: concrete schedules teams actually use

Cal Newport's Deep Work (Grand Central, 2016) offers the clearest practitioner guidance: 3-4 hours of genuine deep work per day for experienced practitioners, up to 5 hours for elite outliers. Newport draws on Anders Ericsson's deliberate-practice research, particularly from Peak: Secrets from the New Science of Expertise (Ericsson and Pool, Houghton Mifflin, 2016), which found approximately 4 hours per day as the sustained ceiling for deliberate practice among elite performers.

For most engineering teams, the realistic and defensible target is two 90-minute deep-work blocks per engineer per day. That is 3 hours of protected, interrupted-by-nothing deep work - less than Newport's maximum, more than most teams currently achieve, and achievable without restructuring the entire organisation.

The two-block model leaves 4+ hours per day for meetings, collaboration, code review, Slack, email, and administrative work. It is not monastic. It is defensible.

Three concrete templates for real teams, not idealised individual schedules:

Template A works for small teams (under 15 people) where individual schedules are easy to coordinate. Template B works for tech leads who need morning maker time but afternoon meeting availability. Template C is the highest-impact whole-team intervention: one no-meeting day per week, applied consistently across the whole team, creates 52 guaranteed deep-work days per engineer per year.

Calendar blocks erode without active defence. The operational discipline that keeps blocks intact:

• Block titles that signal intent. Label the block "Deep work - do not schedule over" explicitly in the calendar entry. Unnamed blocks get scheduled over by well-meaning colleagues who cannot see what is in them.
• Slack status. "In deep work until 10:30 - reply after" during block hours. This converts potential Class 3 (unplanned async) interruptions into Class 1 (batched) without requiring a team-wide policy change.
• Phone in a drawer or blocker running. Opal or Freedom running during block hours. For engineers with high phone-distraction habits, this is the single highest-leverage individual intervention. See /tools-that-help for specific recommendations.
• Team norm on interruption severity. The team agreement: interruptions during a deep block require a named severity (Class 3 or Class 4 on the interruption taxonomy). Class 1 and 2 interruptions are deferred to the next available window.
• Manager cover. The engineering manager commits to rescheduling any meeting that a team member is pulled into during their deep-work block. This is the leadership behaviour that makes the norm credible.

Five patterns that consistently undermine deep-work block programmes:

• The block is on the calendar but the person checks Slack every 7 minutes. This is fake deep work producing real context switching. The fix is environmental, not motivational.
• The block is protected individually but on different schedules across the team, so the team is always 'someone is unavailable' and synchronous availability drops below the point where anyone can schedule a meeting.
• The block is 90 minutes but follows a long meeting that has run over, so it starts ragged and fragmented.
• The block is labelled '9-12 deep work' but nothing is specifically scheduled inside it, so it drifts to email and Slack without anyone deciding that is what they wanted.
• The leader does not model it. A VP Engineering who runs 20 meetings per day sends a clear message to the team about what the real expectation is, regardless of what the policy document says.

A four-week rollout that has worked in multiple engineering organisations: