Purpose
Track the nanny's days off. The founder + wife are monitoring two distinct things (per founder 2026-06-03):
- Disruption / burden — days off where NO coverage was available, so the founder + wife scramble and lose their own work time. "Family comes first." This is the cost metric.
- Suspected opportunistic pattern — the founder's hypothesis: last-minute "sick" call-outs may line up with when backup happens to be available (grandparents in town, daughter's aunt in town, etc.). He WFH and observes the household; today's (June 3) sudden sickness surprised him — no signs of illness Mon or Tue, and his parents happened to be in town. He is NOT anti-sick-day ("I want her to take sick days when she is ill") — the concern is specifically sick days that look timed to available coverage rather than actual illness.
So each sick day carries two questions beyond category: was backup coverage available? and were there observable signs of illness? The log instruments both so the pattern can be tested with real data, not a hunch.
Schema
Each entry is a single-line bullet:
- YYYY-MM-DD - <CATEGORY> - <DURATION> - coverage:<none|incidental|planned> - backup-present:<yes|no|unknown> - illness-signs:<yes|no|unknown|n/a> - last-minute:<yes|no> - <notes>
Categories (the reason): NANNY-SICK | NANNY-VAC | FAMILY-VAC | NANNY-OTHER
- NANNY-SICK — last-minute call-out, claimed illness
- NANNY-VAC — vacation the nanny requested in advance
- FAMILY-VAC — founder + wife on their own vacation (planned by them)
- NANNY-OTHER — personal day, family emergency, other discretionary
coverage — was childcare in place that day?
none— no childcare; founder/wife scrambled or lost work. DISRUPTIVE.incidental— coverage happened but was NOT pre-arranged (grandparents/aunt happened to be around). Not disruptive that day, but not reliable.planned— coverage arranged in advance, or the day was planned by the family itself (FAMILY-VAC, advance NANNY-VAC). Not actionable.
backup-present — was alternate caregiving available that day (grandparents/aunt/etc. in town)? This is the variable the founder suspects she keys off of. yes|no|unknown.
illness-signs — for sick days only: did the founder (WFH, observes them) see actual signs of illness? yes|no|unknown. n/a for non-sick categories.
Duration: full-day | half-day | early-leave | late-arrival | hours: N
Headline metrics (since tracking start 2026-05-13; refreshed 2026-09-25)
Tracking began 2026-05-13 — counts are window-to-date, NOT calendar-YTD (per founder 2026-06-03). Anything before 2026-05-13 is unrecorded.
- Total days off (all reasons): 10
- Disruptive — sick day, NO coverage (the burden): 5 → 05-13, 08-26, 09-08, 09-09, 09-25. Signal A is TRIPPED: 4 no-coverage days in the last 30 days (08-26, 09-08, 09-09, 09-25) vs the threshold of >2.
- Pattern-watch — sick day with backup-present:yes AND/OR illness-signs:no: 2 of 7 → 06-03 (backup yes, no signs), 08-26 (no backup, no signs). 09-08/09-09 = reported-not-observed; 06-30 and 09-25 read genuine (09-25: clear exposure to the daughter's cold).
- Planned / not actionable: 3 → 05-22 + 05-25 (family's Miami trip), 09-07 (Labor Day).
Pattern test (founder's hypothesis: sick days cluster on backup-available days): 1 of 7 sick days had backup present (06-03). The data now points the other way: most sick days land with NO backup, which is why they hurt. The live problem is disruption frequency, not timing.
By reason (total-time view):
| Category | Count since 2026-05-13 |
|---|---|
| NANNY-SICK | 7 |
| NANNY-VAC | 0 |
| FAMILY-VAC | 2 |
| NANNY-OTHER | 1 |
Entries (oldest-first)
2026-05-13 - NANNY-SICK - full-day - coverage:none - backup-present:no (inferred from founder's 2026-06-03 "this time my parents were in town" contrast) - illness-signs:unknown - last-minute:yes - founder iMessage 12:31 ET, no detail at the time. The disruptive one; does NOT fit the backup-available hypothesis.
2026-05-22 - FAMILY-VAC - full-day - coverage:planned - backup-present:n/a - illness-signs:n/a - last-minute:no - founder+wife in Miami (long-weekend trip); gave nanny off. "On us." Logged retroactively 2026-06-03.
2026-05-25 - FAMILY-VAC - full-day - coverage:planned - backup-present:n/a - illness-signs:n/a - last-minute:no - founder+wife in Miami (long-weekend trip); gave nanny off. "On us." Logged retroactively 2026-06-03.
2026-06-03 - NANNY-SICK - full-day - coverage:incidental - backup-present:yes (founder's parents in town) - illness-signs:no (founder WFH, saw no signs Mon/Tue; sudden, surprised him) - last-minute:yes - founder iMessage 09:42 ET. Covered by grandparents so not disruptive, but fits the backup-available + no-illness-signs pattern the founder is watching.
2026-06-30 - NANNY-SICK - full-day - coverage:incidental (family tag-team, resolved same-day: wife Michelle at nap time → Aunt Lauren took her for lunch → founder takes her after his meetings clear so Lauren can drive back to Ft. Lauderdale) - backup-present:yes (Aunt Lauren, covered lunch before heading out) - illness-signs:yes (sniffling yesterday; "genuinely didn't sound good this morning" per founder) - last-minute:yes - founder iMessage 07:47–09:01 ET. Context: already a short week — nanny planned-off Thu 7/2 (standing Thursday off), likely off Fri 7/3 (July 4 observed), and all of next week (family's own Miami summer trip ~7/6+, = FAMILY-VAC, on the family). GENUINE sick day — real illness signs AND the backup was about to leave (inconvenient timing). Does NOT fit the opportunistic-pattern hypothesis; evidence-against. Founder absorbed part of his own afternoon, so a partial burden, but not a no-coverage day.
2026-08-26 - NANNY-SICK - full-day - coverage:none (fell to founder; he moved 3 internal calls to 8/27, none client-facing) - backup-present:no (sister-in-law out of town; wife had patient visits she could not miss; grandma had watched her the night before and had to get to her own work; daughter also kept home from preschool's half day because of a hand-foot-and-mouth outbreak there) - illness-signs:no (founder saw no signs; nanny said "see you tomorrow" on leaving 8/25) - last-minute:yes - founder iMessage 10:22 + 10:25 ET. Second no-coverage day on record (first: 2026-05-13). Fits BOTH watched dimensions (no backup available, no illness signs) - but note this is the first no-backup + no-signs day; 6/3 was backup-present + no-signs, 6/30 was genuine.
2026-09-07 - NANNY-OTHER - full-day - coverage:n/a - backup-present:n/a - illness-signs:n/a - last-minute:no - Labor Day, standard recognized holiday, planned in advance. Not part of the sick-day pattern watch.
2026-09-08 - NANNY-SICK - full-day - coverage:none - backup-present:no (no grandparent, no sister available; founder himself already in the negative on his own PTO balance) - illness-signs:reported-not-observed (nanny told them a stomach bug; founder not questioning legitimacy here, explicitly said "I really wish our caretaker would tough it out a bit more" — frustration is about impact/timing, not suspicion) - last-minute:yes - founder iMessage 13:05 ET. Third no-coverage day on record (prior: 2026-05-13, 2026-08-26). Compounding context worth keeping: wife Michelle is still recovering from surgery this week (tired, in pain, cannot lift or chase their daughter or help her on stairs), so this lands as a genuine two-adults-down day, not just a one-day nanny gap. Does not fit the opportunistic-pattern hypothesis — evidence-against (he's not suspicious of this one, just under real strain).
2026-09-09 - NANNY-SICK - full-day - coverage:none - backup-present:no (Michelle stayed home from work instead — she's recovering from surgery and "pushed herself too hard yesterday" doing childcare, so this is a sick day covered by a not-actually-resting recovering adult, not real backup) - illness-signs:reported-not-observed (stomach bug reportedly started Tue morning, continued Wed, vomiting reported — founder: "we are just taking her word for it") - last-minute:yes - founder iMessage 12:53-12:58 ET. Second CONSECUTIVE no-coverage day (09-08, 09-09); 4th on record overall (05-13, 08-26, 09-08, 09-09), 3 of 4 within the last 14 days — crosses the disruption threshold. Founder flagged this could extend to a full week off, not just today. New standing fact: nanny's regular day off is Thursday (not sick-related) — exclude future Thursday absences from this log unless she's ALSO out sick that day; her Monday 9/7 absence was the Labor Day holiday, already correctly excluded.
2026-09-25 (Fri) - NANNY-SICK - full-day - coverage:none (on founder + Michelle) - backup-present:no ("no backups around") - illness-signs:plausible-exposure (fever reported by "nana"; not observed by founder, but a clear exposure path: daughter had a head cold → ear infection over the weekend, antibiotics from Sun, no fever; nana watched her Mon-Wed as normal; daughter back to school Thu) - last-minute:yes - founder iMessage 10:24 + 10:28 ET. Reads GENUINE: caught it from the daughter she cared for all week; evidence-AGAINST the opportunistic hypothesis (no backup available). 5th no-coverage day on record; 4th in the last 30 days (08-26, 09-08, 09-09, 09-25). 7th NANNY-SICK overall. "Nana" = the daughter's nickname for Andreina (founder confirmed 10:30 ET).
Context added 2026-08-26 (founder, iMessage 10:29 ET)
- Nanny = Andreina. The daughter calls her "Nana" (founder 2026-09-25); "nana" in a founder message means the nanny, not a grandparent. Daughter's preschool (3s class) meets 5 days/week; the family pays for 4 and chooses which day she stays home. That choice has been made last-minute, which leaves the nanny unable to plan mornings.
- Founder is pushing for a STANDARD skip-day so Andreina can plan to be there that morning. His read on 8/26: a predictable schedule would have let her say "I planned to be there by noon, I can still come then" instead of calling a full sick day. Treat that as his working hypothesis about the mechanism behind the no-signs days, not as established.
- Follow-up: once the fixed skip-day is chosen, record it here and check whether later no-signs sick days cluster on or off that day.
When to surface
Two separate signals, surfaced when either trips:
A. Disruption (cost): days with coverage:none.
- More than 2 no-coverage sick days in any 30-day stretch, OR more than 1/month sustained over a rolling 3-month window.
B. Opportunistic pattern (legitimacy): the founder's hypothesis that sick days time to backup-availability.
- Surface once N is meaningful (≥4-5 sick days) IF the share with
backup-present:yesAND/ORillness-signs:nois clearly disproportionate. Report it as a ratio with the raw entries, never as an accusation — the log's job is evidence, the founder's job is the conversation. - Until N is meaningful, just keep instrumenting. Do NOT surface a "pattern" off 2-3 data points.
Capture discipline (going forward)
When the founder reports a sick day, capture both new dimensions at log time:
- backup-present? (were grandparents/aunt/anyone available that day)
- illness-signs? (did the founder, who WFH, observe actual signs of illness)
If he doesn't volunteer them, it's fine to ask once, briefly — these two fields are the whole point of the pattern watch.
Related
- [[index]] - life project