Turn a seller thesis into a data-pull recipe
Rung 4 of 10 · Recipe pattern
1 · Learn the move · Recipe pattern
Every agent farms the same neighborhood off the same platform and mails the same 'thinking of selling?' postcard. The edge is the QUERY: the filter combination that isolates the owners most likely to list before the rest of the market pulls them. The recipe pattern turns a plain-English thesis, say, long-tenure absentee owners with high equity, into the exact filter steps for your data platform (PropStream, your MLS's public-record search, a county export), so an assistant can pull an expert-grade list. Two things keep it clean. The output is FILTER CRITERIA, not a list of people and not a fact about any owner, so there's nothing for the AI to fabricate, it translates your thesis into clicks, never invents a name, an equity figure, or a phone number. And every filter targets owner BEHAVIOR or STATUS, tenure, absentee, equity band, pre-foreclosure status, never a protected characteristic (race, religion, national origin, familial status, disability) or a proxy; where a thesis drifts near that line the recipe flags it for counsel instead of encoding it.
Turn this seller thesis into the exact filters for my platform: long-tenure, high-equity, absentee owners in [farm area]. Output filter CRITERIA only, never an owner fact; target behavior/status, never a protected class. Name my platform.
2 · Your turn. You write the prompt
You farm a zip code and want the owners most likely to list, not the whole tax roll. Your thesis: owners who've held 12+ years, have high equity, and live out of the area, because long-tenure absentee owners with equity are the likeliest to sell. You use PropStream. You want the exact filter recipe your assistant can follow, without the AI inventing owners or drifting into targeting you can't defend. Write that prompt.
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.