The weekly number review
Lesson 6 of 6 · Run it like a shop
1 · Learn the move · Cognitive verifier
The weekly review is where a land business finds out whether it is working, and it is where a model does its most confident damage. Hand it a week of tallies and ask how you did, and it computes a rate slightly wrong, compares it to a typical figure it made up, and tells you the campaign is performing well. The cognitive verifier pattern breaks that into checkable steps. First the model lists the questions the review must answer, and the ones these tallies cannot. Then it computes each figure from your pasted tallies only, with the arithmetic written out. Then it re-checks every figure against the source rows and flags anything that does not tie, including a ledger that does not add up. What it never does is supply a benchmark, call a number good or bad, or project next month. The only comparison is your own last week, and the meaning of the number stays with you.
Here are this week's tallies, last week's, and this week's ledger. Step 1: list the questions a weekly review should answer from these numbers, and the ones it cannot. Step 2: compute each figure from my tallies only, arithmetic shown. Step 3: re-check every figure against the source and flag any that do not tie. No benchmarks, no 'good' or 'bad', no forecast.
2 · Your turn. You write the prompt
Every Friday you look at the week and get a feeling. This week in Tallow County: 2,400 letters mailed, 31 inbound calls, 9 of them from owners on the list, 3 offers sent, 1 accepted, 1 closing funded. Last week: 2,100 letters, 24 calls, 8 from owners, 2 offers, none accepted. This week's ledger has three lines, $1,512 printing and postage, $180 list data, $45 phone line, and a total you wrote as $1,757. Write the prompt that gives you the same review every week, with numbers you can trust.
Write the prompt for this scenario as given, or swap in a real one from your own work. The score works the same either way.
Remember: the AI sees only your prompt, not this page. If the situation isn't in your prompt, it doesn't exist.
Optional. These shape the output when you run your prompt below, not your score.