31% of my week — not doing work, not switching between tasks — just reconstructing context I'd already had and abandoned.
I spent six months tracking where my time actually went while running ad ops automation, a content publishing operation, and a CRM sync service across four clients. I went in assuming deep work was eating roughly 50% of my hours. After six weeks of strict observational logging in Toggl (tagged at task level, started in real-time — never reconstructed at day's end), the real number was 31%. Context-switching overhead came in at 38%. The remaining 31% was a category I hadn't even named: re-reading Slack threads to find where a project stood, reopening 14 browser tabs to rebuild a mental state I'd dropped 48 hours earlier, rereading my own Notion docs because I hadn't left a clear resumption point.
The fix wasn't better time-blocking. I tried that — seriously tried it, three weeks of Pomodoro followed by 90-minute hard-blocked calendar sessions. Individual task output nudged up. Weekly throughput across all four clients went down. When you're managing 12 active automation workers and a client's ad account can get flagged at 6pm on a Friday, rigid blocking just means you feel organized while falling behind. What actually moved the number was attacking the orientation overhead directly. I built a dead-simple Notion database I call the "resumption layer" — the only hard rule is that you can't close a task without writing one sentence: the next physical action and where the relevant context lives. Not a summary. A resumption point. Three weeks later, orientation work dropped from 31% to 17%. Roughly five hours a week I stopped spending on reconstructing my own mental state.
The second change was subtler: I stopped tagging Toggl entries by project and started tagging by cognitive mode — build, review, communicate, admin. Project-level data tells you where time went. Cognitive-mode data tells you what your brain was actually doing. Turned out "review" was running at nearly the same mental cost as "build" for me, and I'd been stacking review sessions immediately after deep build blocks and wondering why I was useless by 3pm.
I wrote up the full breakdown — including how the logging workflow itself broke in week four and what I replaced it with — over on dailyfocusmag.com.
