Independent editorial published by Digital Signet. Some links are affiliate links; we only recommend things we would recommend anyway. Dollar figures are models, not guarantees: every assumption is visible on the calculator page. Last verified April 2026.
contextcost.com · 2026 edition · last verified April 2026
This site translates the research of Gerald Weinberg, Gloria Mark, Mihaly Csikszentmihalyi, Paul Graham, and the DORA team into two dollar calculators an executive committee can act on. The math is open. The assumptions are visible. Sister publications: featurebloat.com, codesmellcost.com.
Methodology and assumptions: /calculator. Prefer a meeting-specific number? /meeting-cost-calculator.
Deep-dive at /meeting-cost-calculator.
The term borrows from computer science, where a context switch is the CPU saving one process state and loading another. In knowledge work the cost is several orders of magnitude larger. A CPU context switch costs microseconds. A knowledge-worker context switch costs tens of minutes.
Gerald Weinberg captured the organisational version in Quality Software Management, Vol 1: Systems Thinking (Dorset House, 1991, pp 284-285). His table shows what happens when an engineer carries more than one project at a time: at two projects, each receives 40% of productive effort (20% lost to switching); at three projects, each receives 20%; at four, 10%. The compounding is severe and non-obvious.
"The waste on two projects is not twice the waste on one. It is ten times the waste."
These estimates were heuristic, not laboratory measurements, and Weinberg said so. They remain the most-cited figures in the field because nothing more rigorous has replaced them for knowledge-work organisational concurrency. See The Research for the full treatment.
Gloria Mark and colleagues at UC Irvine studied knowledge workers in their natural habitat over multiple years. The finding in "No Task Left Behind?" (CHI 2005) and confirmed in "The Cost of Interrupted Work" (CHI 2008): after a single interruption, the average worker takes 23 minutes and 15 seconds to return to full cognitive engagement with the original task.
Interrupted workers compensate by working faster in the short term, but with measurably higher stress and more errors. The speed increase is a coping mechanism, not a productivity gain. Sophie Leroy's attention-residue research (OBHDP 2009) shows that even after you technically return to Task A, residue from Task B continues to impair performance. The true switch cost is larger than the 23-minute number captures.
Not all interruptions cost the same. A severity-ordered taxonomy:
Email, PR digests, newsletters. Lowest cost: you choose when to process. Expand this class.
Planned meetings, standups. Known in advance, budgetable. Toxic only when back-to-back.
Slack DMs with red dots, @-mentions, urgent email. Unpredictable timing plus urgency signal.
Tap on shoulder, ringing phone, desk drive-by. Terminates any deep-work block immediately.
Full taxonomy with policy responses: /interruption-taxonomy.
This site is the third panel of a tech-editorial trilogy with featurebloat.com (adding features is visible, removing them is invisible) and codesmellcost.com (refactoring is invisible, shipping is visible). All three argue the same thesis at three zoom levels: visible work crowds out invisible work, and invisible work determines whether an organisation ships anything meaningful in five years.
Digital Signet runs two-week attention audits. We map your calendar, inventory your interruption channels, measure your real focus-time, and deliver the memo that protects your team's best hours.
Email Oliver about an attention auditFor a 20-person engineering team at a fully-loaded $120/hr with 4 switches/day and Weinberg's 20% loss assumption, the annual cost is approximately $270K. For a 50-person team the number exceeds $670K. Use the calculator above with your own numbers. The formula is: headcount x hourly rate x hours/day x switches/day x loss% x working days/year.
Gerald Weinberg's table in Quality Software Management, Vol 1 (Dorset House, 1991, pp 284-285) shows that at two concurrent projects, each receives 40% of productive time (20% lost); at three projects, 20% each; at four, 10% each. These are engineering heuristics, not laboratory results, and Weinberg says so explicitly. They remain the most-cited estimate because nothing more rigorous has replaced them.
Yes, for sustained cognitive work. The literature (Rogers and Monsell 1995, Rubinstein et al. 2001, APA 2006 review) is consistent: task-switching imposes a switch cost paid at the start of every new task. For software engineers whose value comes from holding large mental models, the loss is at the severe end of the range.
Paul Graham's July 2009 essay argues that managers work in hour-long appointment slots, while makers (programmers, writers) need half-day or whole-day unbroken blocks. A single 11am meeting destroys the 8am-11am productive block and leaves a broken fragment too short for hard work. See /ic-vs-manager-maker-schedule for the calendar diagram.
Csikszentmihalyi's interviews in Flow (Harper & Row, 1990) suggest roughly 15 minutes of uninterrupted concentration before flow onset. Combined with Mark's 23-minute refocus finding, an interruption costs at minimum 15-23 minutes before full productive engagement resumes.
Updated 2026-04-27