An ops summary with the numbers filled in

The weekly ops summary usually gets written by someone assembling numbers from four places on a Friday afternoon, which is why it is often late, occasionally wrong, and always written by the person least keen to be doing it.

The writing is not the work. The gathering is.

Fix the numbers before the prose

Decide which figures the summary always contains and where each one lives. The same numbers, the same order, every week.

That consistency is what makes a weekly summary readable — a reader learns the shape and can find what they want. It also makes it automatable, because "gather these six figures" is a task and "summarise the week" is not.

Get the numbers, then write around them

Every Friday at 3pm, write the weekly ops summary.

First gather: [the six figures and where each comes from].

Then write:
  the numbers, with the change against last week
  one paragraph on what moved and why, where you can tell
  anything that looks wrong rather than merely different
  what is carrying over into next week

If a figure is unavailable, say which and why. Never carry last week's
number forward or estimate it.

Under 400 words. If nothing notable happened, a short summary is the
correct output.

"Looks wrong rather than merely different" is the sentence that earns the summary its readership. A report of movements is a table with adjectives. A report that says one figure is implausible given another is doing the thing a person would do, and it is the only part that regularly catches a broken pipeline before the following month.

Send it before people leave

Friday at three, not Monday. The summary is most useful while the week is still in everybody's head, and things that need chasing can be chased before the weekend.

Let it be short on a quiet week

The instruction that a short summary is correct output matters. Weekly reports inflate: nothing happened, but a report was due, so three paragraphs were produced. Two of those in a row and people stop opening it.

Review the shape quarterly

Not the content, the sections. Ops summaries accumulate figures that were added for a reason that has since gone, and nobody removes them because nobody notices. Once a quarter, ask which of the six numbers anyone actually used.

What good looks like

A short summary lands Friday afternoon with the numbers already in it, and occasionally it points out that something looks wrong.

The measure is whether anyone asks a follow-up question. A summary that never generates a question is being skimmed.