Screen a county in fifteen minutes
Lesson 1 of 6 · Pick the county before the parcel
1 · Learn the move · Specific task + easy yes
Counties get chosen by feel: a friend did well there, a forum thread said it was hot. Ask a model whether a county is a good market and it will answer fluently from nothing - it has no current sales, no delinquent list, and no idea how many other buyers mail there. The fix is to shrink the task until the answer is checkable. Write down your own gates first (records online, delinquent list obtainable, enough parcels in your size band, sold examples you found, recording by mail), paste what you pulled, and ask for one YES, NO or [NEEDS] per gate. Then end on one easy yes: keep researching this county this week, or drop it. That is a screen. It is not a ranking, and it never says a market is good, because nothing you pasted could support that sentence.
Screen one county against my five gates below, using only what I pasted. For each gate: YES, NO, or [NEEDS: the record]. Then one line: KEEP RESEARCHING or DROP. Do not rate the market, compare it to other counties, or tell me what land sells for there. [my gates] [what I pulled]
2 · Your turn. You write the prompt
You have fifteen minutes before your weekly call with your VA, and a county the members' forum keeps naming, Pellam County. Before the VA spends a week building a mailing list there, you have the treasurer's web page text, a row count from the assessor export you filtered to vacant 5 to 40 acres, and two sold parcels you found on a land marketplace. Write the prompt that screens it against your own gates and ends in one decision.
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.