A queue is not a list of wishes. It is a set of decisions with owners, and a week is what those decisions cost.
Monday: triage, 90 minutes
Northline Studio, five client sites, one person.
new items this week: 14
- harbor and pine 4 (2 audit findings, 1 comment, 1 search note)
- cedar and slate 2 (1 access alert, 1 client request)
- brightwater 3 (2 comments, 1 audit finding)
- foxglove 4 (3 comments, 1 request)
- meridian 1 (1 redirect finding)
states after triage
queue 7
reply 3
defer 2 with dates in the next cycle
dismiss 2 blog idea, brand colour
queue cap breach: foxglove reached 9 items, so the oldest defer was
moved to next cycle and recorded.
Tuesday and Wednesday: work, 3 hours
shipped
- harbor and pine: three service titles rewritten, one pull request merged
- cedar and slate: search console reconnected, six days of data backfilled
- meridian: two 404s redirected, one hop each
not shipped
- harbor and pine: the description rewrite waits for the client's wording
Thursday: read, 90 minutes
Reading day is not verdict day. Two changes shipped two weeks ago are still inside their window. This week’s read only confirms the data is arriving.
changes inside their window: 3
changes whose window closed: 1
verdict recorded: inconclusive, with the comparison set noted
Friday: close, 60 minutes
queued at the start of the week 12
shipped 4
deferred with a date 3
dismissed with a reason 2
still queued 3 each with a named next action
interruptions
- one outage, cedar and slate, 40 minutes on Wednesday
- one client call, 25 minutes on Thursday
The three still-queued items are the honest part of the log. A week that closes with everything empty usually means the queue was filtered before it was counted.
Acceptance: the week is closed when
- every queued item has a state, and a next action if it stayed open
- every shipped item has a verdict or a window that is still open
- dismissed items keep their reasons
- interruptions are recorded, because they explain the week
- the count at the start equals the count at the end plus the work
Troubleshooting
| Symptom | Likely cause | What to do |
|---|---|---|
| Triage takes two hours | Items arrive without evidence | Ask the agent for evidence, and send it back |
| The queue never shrinks | The cap is not enforced | Move the oldest defer and record it |
| Two interruptions ruin the week | No rule for outages | Adopt the exception rules from the roster plan |
| Verdicts pile up | Nothing closes windows | Reserve Thursday for the readings |
| Friday close slips | It is treated as optional | Book it, and keep it to 60 minutes |
What this log does not show
The log records decisions, not effort. A week of reading with nothing shipped looks empty here, and it can be the correct week.
It also does not show the two hours a client spent answering a question. Nor does it show the day the developer spent on a preview that never merged. Those costs are real, and they belong in the retainer conversation.
Limits
This shape assumes one or two sites and one person. A larger roster uses the same shape with one site per day, and the Friday stop does not move.
A week with four interruptions is not a process failure. It is a week where the plan made the cost visible.