35 minutes of daily maintenance, compounded over 8 months, is north of 140 hours — and I didn't see that number anywhere in the glowing 'second brain changed my life' threads.
I've been running Obsidian + Notion + Claude Sonnet across three businesses and four client retainers for two years. I tracked context-switch cost across 6 weeks, measured system maintenance time daily for 8 months, and timed re-orientation cost per session over 3 weeks. The ROI is real in exactly two places — and completely absent in one place every evangelist promised it would show up.
The biggest surprise: the "capture everything" mandate made retrieval worse, not better. At month six I had 2,300 notes. Finding the right one under time pressure was slower than re-Googling. I'd built a private internet with worse search. The fix was brutal — I collapsed capture and processing into a single step. If something doesn't have an obvious destination in 10 seconds, I say it to Claude Sonnet and let the conversation log serve as the record. Obsidian is now strictly for things I've already decided matter. That change alone dropped my daily maintenance time significantly.
The ROI that actually held: pasting relevant Obsidian notes directly into a Claude Sonnet session as context before any complex task. Running 12 workers across 4 clients means every context switch otherwise requires a full mental reload. With notes as primed context, I measured roughly a 40% drop in re-orientation time per session over a 3-week tracking period. Not a study — my numbers, my conditions — but they held consistently. Client context retention across long gaps also paid back in ways I can point to: at least three retainer relationships saved by having a dated note with the exact decision and the reasoning when a client emailed six weeks later expecting me to remember everything.
The place it didn't pay back: creative ideation. The Obsidian graph view is beautiful. It has never once surfaced a connection I wouldn't have made anyway.
I wrote up the full breakdown — including the broken daily-note workflow that generated 340 unprocessed notes by month four, and the honest decision guide for whether a second brain even makes sense for your workload — over on dailyfocusmag.com.