Put together a complete brief for AI that brings together the entire module. Return JSON {"data":"...","questions":"...","checks":"...","deliverable":"..."}: data — context (what kind of data, one line, units); questions - 3-4 measurable questions leading to the cause of the fall (trends, cuts, anomalies); checks - how to check AI numbers for convergence; deliverable - what is the output: the main conclusion plus 2-3 priority actions with the expected effect.
A strong analysis is not one question, but a chain: context → measurable questions → verification → conclusion → action. You already know how to do every part; the finale teaches you how to stitch them together. Mechanics: AI responds narrowly to what is asked, so a big question must be broken down into steps, otherwise you will get a streamlined general answer. Lever: lead from the general to the specific. First, WHERE fell - what channel, category, city; then WHY - discounts have eaten up the margin, the average bill has fallen, or an entire category has disappeared. Insider: put forking hypotheses into questions. “If the average bill has fallen, the reason is due to discounts; if the number of checks has dropped, it’s in traffic.” The fork guides analysis in advance instead of wandering: any answer leads to the next step. The second trick: always close the contour with an action with an effect - an analysis without “what to do” is a beautiful obituary, and not help for business. A typical rookie mistake: lumping everything into one giant request “figure out why profits fell.” Break it down into steps and check the numbers on each one - then the conclusion will be solid.
Unlock access to submit solutions for instant AI review.