The shift isn't over until the next shift can run it: the 5-line handover log

The shift isn't over until the next shift can run it

Most handovers are conversations, and conversations evaporate - exactly at the moment of maximum tiredness and minimum patience, which is when they're held. The fix is one rule: the handover is written before it is spoken. The talk is the Q&A for the log, not a replacement for it.

What actually travels between shifts is state, and it comes in three kinds:

Open loops - everything started and unfinished, each with an owner and a next deadline. Skip this and the new shift redoes finished work while the unfinished work quietly dies.

Current risks - what's degraded or running on a workaround, plus the early-warning sign for each. Skip this and the incoming shift treats a waving flag as scenery.

Decisions already made - what was decided, by whom, who was told. Skip this and the new shift re-litigates settled questions and sends customers the same news twice.

The five-line entry (one per shift, copy-paste):

1. What happened - facts with timestamps ("19:42 - checkout errors ~1 in 4; 21:10 - cleared").
2. Still open - loop / owner / deadline / done-looks-like ("6 re-runs / Sam / 11:00 / confirmation emails sent").
3. At risk - the fragile thing + its early warning ("sync on retry queue; if last-synced > 30 min, it's failing again").
4. Who knows - names. "Riverside Cafe emailed twice, answered twice, promised morning confirmation."
5. If worse - threshold + named escalation ("error rate > 10% or queue > 45 min: duty line, then Alex").

Line 4 is the one small teams skip and customers feel. It's the difference between the morning shift sending one promised confirmation and a third apology email to someone who's already been apologized to twice.

The overlap rule: ten minutes, scheduled at the shift boundary, paid - never "when things quiet down," because things never quiet down. In a two-person company the hot handover is you at 17:25 writing to you-at-9am, who is effectively a stranger.

The repeat-back: the incoming shift reads the top three loops back in their own words. Thirty seconds; every misunderstanding it catches was otherwise going to be discovered at 15:00 by someone who never agreed to it.

Log the quiet shifts too - two lines: "quiet, nothing open." Baseline plus proof the system runs daily. A log that only appears on bad shifts starts to feel like a confession, and the fastest way to avoid confession is to stop noticing things.

Worked example on the page: a flapping payment gateway (19:42-21:10, ~25% of checkouts failing, provider green), six failed orders, one twice-emailed customer. Four minutes of writing at shift end; the morning reads in ninety seconds, re-runs the six orders by 10:40, and sends exactly one follow-up - the one that was promised.

Full template - the three kinds of state, the five-line entry, the open-loops fields (owner / deadline / done-looks-like), the six rules, the paste-in block, and the five traps:

https://hive80-lab.github.io/ops-notes/shift-handover-log-template.html

(For the fortnight-long version of the same move, the annual leave coverage checklist: https://hive80-lab.github.io/ops-notes/annual-leave-coverage-checklist.html - and what tomorrow morning does with tonight's log: https://hive80-lab.github.io/ops-notes/after-action-report-template.html)