Value raw land from your comps. A range, never a number
Rung 2 of 10 · Comparison matrix
1 · Learn the move · Comparison matrix
Raw land is the hardest thing to value, because the comps are scarce and no two parcels match on the four things that move price: access, acreage band, road frontage, and topography or view. A residential AVM built on rooftops fails on dirt. The move is a comparison matrix off the sold-land comps YOU paste: one row per comp, columns for each price driver, and a plain-language adjustment against the subject, up or down, with the reason. The output is a defensible RANGE with a confidence read, never a single confident number, because a point value on thin comps is a fiction. Two rules carry the lesson: no comp gets used that you did not paste and verify (an invented comp is the fastest way to misprice), and any parcel where the paste is silent, unknown access or an unconfirmed acreage, is marked UNKNOWN rather than guessed. This is an opinion of value for your own pricing, not an appraisal, and the strategy of listing just under the competing inventory is a way to sell faster, never a promise that it sells.
Here are five sold-land comps and my subject parcel. What's it worth? [paste the comps and the subject]
2 · Your turn. You write the prompt
You're pricing a 10-acre rural parcel you're under contract to flip. You paste five sold-land comps (each with acreage, sold price, price/acre, access type, road frontage, and a topo note) and the subject's specs: 10 acres, paved-road frontage, gently rolling, county-maintained access. Two comps are larger with worse access; two are similar; one is a 40-acre outlier at a much lower price/acre. You want a defensible value range and a list-to-sell price, without the AI inventing a comp or handing you one magic number. 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.