The three deals to work this week, waiting on Monday

What are the five questions?

This is the whole prompt. The only edits it needs are your stage names and thresholds:

• "Open my CRM and answer these five questions about every open deal above [amount]. Use numbers and deal names, no generalities."
• "1. Which deals changed stage since last Monday, and in which direction? 2. Which deals have had no activity (no email, no call, no note) in 14 days or more? List them oldest-silence first. 3. Which close dates moved, and by how much? 4. What is the total weighted pipeline now versus last Monday, and which deals account for the difference? 5. If I can only work three deals hard this week, which three, and what is the specific next action on each?"
• "Format: questions 1–4 as tables, question 5 as three short paragraphs. Cite the record you read for every claim. Change nothing in the CRM. This is read-only."

What a good run looks like

The stalled-deals table earns the habit on week one.

A good run also names the boring cause in question 4: "weighted pipeline fell because the Acme close date moved to Q3, not because anything was lost."

Question 5 is where you grade it. The right answer ties evidence to action: "Meridian has gone quiet 16 days after a positive demo. Re-engage through the champion, not the buyer." If you get "focus on your biggest deals", require each recommendation to cite the activity log line behind it.

Make it a Monday routine

Schedule the prompt for 7am Monday and the review is waiting before standup.

The week-over-week comparisons in questions 1, 3 and 4 work better as a routine, because the companion holds last Monday's run as the baseline instead of recomputing history.

Teams pair it with CRM data entry automation, since the review is only as good as the activity log, and a companion that files call notes keeps that log honest. The write side of Attio is on the Attio integration page.

What should you verify before trusting the numbers?

Recompute the weighted total yourself once.

It is one filter and one column sum, and a match calibrates the whole run cheaply. Then check the activity definition: some CRMs log automated emails as activity, which empties the stalled-deals table. If your list looks short, add "automated sequence emails do not count as activity" to question 2.

Keep the read-only line even after you trust it. When you want writes, split them into their own prompt with its own approvals, the way sales teams using agents separate reporting from data entry.

Frequently asked questions