Prompt of the day

Share-loss forensics: who is taking it, where, and with what lever

Share is slipping and the meeting is Thursday. Build the suspect list and the confirmation plan before opinions fill the vacuum.

You are a competitive insights analyst running a share-loss investigation. I will paste my share trend and whatever cuts I have (retailer, region, segment, price tier, competitor set). Produce:

A) A SUSPECT TABLE, ranked by likelihood: competitor, where the loss is visible in my cuts, the lever they appear to be using (price, promotion, distribution, innovation, media), and the evidence status — shown in my data vs hypothesis.
B) The THREE sharpest confirmation cuts to run next, each with what a positive result looks like.
C) A 3-sentence EXECUTIVE SUMMARY that clearly separates what is confirmed from what is still hypothesis, so nobody leaves the meeting quoting a guess as a fact.

My data: [PASTE: share trend by period, cuts available, competitor list, anything known about competitor activity]

Rules: Do not invent, estimate, or extrapolate any figure — if a number is not in the data I give you, write "not provided" and flag it. Mark every claim I should verify against my syndicated data or internal reporting before using it externally. Never include retailer-confidential terms or personally identifiable shopper data.

Why it works — Share-loss meetings run on anecdote unless someone shows up with a ranked suspect list and an evidence column. The confirmed-vs-hypothesis split is the discipline that keeps a forensic read from becoming the office rumor mill with a chart on it.

Ask AI to grade its own answer and redo it as a 10

Models settle for 'good enough' on the first pass. One self-critique loop routinely lifts quality more than a longer original prompt.

After any output you find just okay, reply: 'Rate that 1-10 against what a top expert would produce, explain the gap, then rewrite it as a 10.'

1. Get the first draft.
2. Send the rate-and-improve line.
3. Repeat once more if needed.