Skip to content

weekly-readout: the client report

Run the week Cowork-ready

Writes the weekly report you actually send the client. It pulls Meta, Google, GA4 and your backend (store or CRM), computes the business metrics that match how the account really makes money, detects anomalies against trailing 8-week bands, and writes the narrative.

The thing that makes it different from a dashboard screenshot: it has an anti-cherry-picking gate. A readout that only reports the metrics that went up does not pass. Losses get stated as plainly as wins, because a client who finds the bad news themselves stops trusting the good news.

It also runs a quarterly QBR mode for the bigger review.

  • Every week, per client. It is the deliverable clients judge you on.
  • Quarterly, in QBR mode, for the strategic review.
  • When you need the week’s numbers with their provenance, because someone is going to ask where a figure came from.
Cowork ✓ Desktop ✓

Both. The branded HTML report is self-contained, so it renders anywhere.

“Run the weekly readout for examplebrand

For the quarterly version:

/um-toolkit:weekly-readout QBR for examplebrand, Q3”

  • The account, confirmed by its platform-reported name.
  • Your declared conversion shape (see conversion shapes). This decides the entire metric set, so it is not guessed. An ecom store gets MER, aMER, CAC and CM3. A consultancy or a home-services business gets a two-tier headline instead, because the revenue from this week’s leads has not happened yet.
  • Backend revenue access where possible. Platform-reported revenue alone gets labelled as such.
  • Markdown plus a branded, self-contained HTML report you can send as-is.
  • The headline metrics for your shape, each traced to its source and carrying a trust label from the tracking gates.
  • Anomalies, flagged at 2 sigma against the trailing bands, not vibes.
  • The narrative: what happened, what it means, what happens next.
  • An archived history line, so next week can compare against this one.

You: “Weekly readout for examplebrand.” (a lead-to-close consultancy, 45-day cycle)

It returns a two-tier headline. Banked: 4 deals closed this week, $61k, sourced from leads generated 30 to 75 days ago. Leading: 34 qualified leads this week at $212 each, labelled leading indicator, not revenue. Cost per qualified lead is up 18% against the trailing band, flagged. Narrative names the flat close rate as the real story, not the lead cost.

  • You cannot get an ecom headline on a lead-gen account. That is the point. Reporting a MER on an account whose revenue lands 45 days later is the mistake this prevents.
  • A trust label of YELLOW or RED on a number means the tracking gate is not confident. Fix the tracking before you argue with the number. tracking-doctor does that.
  • Declare sales-cycle-days in your brand guide if your revenue lags. Without it, every lag-dependent row honestly reports NO-DATA rather than inventing a window.
  • Send it, do not paraphrase it. The anti-cherry-picking gate only protects you if the client reads the version that passed it.

Runs inside weekly-account-os. Reads the gate verdicts from tracking-doctor and signal-surgeon. Pairs with meta-account-doctor for the diagnostic half.