Claude for weekly ops reporting: stop writing Friday reports

The Friday tax

Every small company has a Friday tax. Someone — usually the founder or ops lead — sits down to write the weekly update. What shipped. What didn't. Where we're stuck. Who needs help.

It takes 90 minutes, honestly. Half of it is remembering. The other half is translating into something readable.

Claude can do both halves if you let it.

The setup

Claude reads three sources:

1. Your task system (Linear, ClickUp, Notion, GitHub issues — whatever you use). The last 7 days of completed and in-progress items.
2. Your CRM, specifically deals that moved stages, calls logged, new contacts.
3. Your calendar for the week — just titles and durations, to contextualize where time went.

You connect these via MCP servers (see MCP servers explained) or by exporting CSVs if you're less technical.

The prompt

Generate a one-page weekly ops report for [Company], week of [date]. Use only the data in the attached exports. Structure: (1) 3-sentence headline summary, (2) What shipped — specific, with owners, (3) Deals that moved — company, stage change, size if known, (4) Where we're stuck — issues older than 7 days in progress, (5) Metrics — tasks completed this week vs last, deals moved this week vs last. Tone: direct, no fluff. Flag anything concerning.

You get a draft. You edit for 5 minutes. You send.

Why not just let Claude send it

Because the draft will always contain one thing you'd rather phrase differently, or one judgment call about "where we're stuck" that wants a human eye. The 5 minutes of editing is the point. Without it, reports become wallpaper — technically correct but unread.

What this unlocks

Two things most teams miss.

1. You start noticing patterns. When the report lands automatically every week, you read it. You start noticing "deals that moved" is flat for three weeks. You fix something. The report is doing its job.

2. You can send it to clients too. If you're an agency, the same pipeline generates a client-facing version. Strip internal details. Different header. Same underlying data. Clients see consistent reporting without you hand-rolling each one.

The gotcha: data hygiene

If your Linear is a mess — people not closing tickets, deals sitting in "open" forever — the report will reflect it. This is usually a good thing. It forces the hygiene everyone's been avoiding.

But don't be surprised in week one. Plan for two weeks of cleanup before the output is trustworthy.

---

We run weekly reporting for all our Operations clients. It's the single most-read artifact we produce.