AI listing marketing: what the data says it does, and what it doesn’t
A working definition, the three layers AI can operate on, the two things it should never be trusted with, and five questions that separate a production system from a prompt box.
We publish what we can source on how listing marketing actually gets produced — what the industry spends on it, what AI has and hasn't changed, and how the work is done well.
Categories: AnalysisMethodField notes
Arguments built on published industry data — NAR surveys, brokerage filings, transaction records. Every figure cites a public source, and we say where the record is thin.
How to produce a specific piece of listing marketing, with the reasoning made explicit: what belongs in it, what order to build it in, and the failure modes to design against.
What we observed building and shipping Mareto — the decisions that turned out to matter, including the ones we got wrong the first time.
Brokerage AI use has gone from novelty to default in two years, yet 46% of agents report no noticeable effect on their business. The evidence suggests the industry automated the cheapest step and left the expensive one untouched.
A working definition, the three layers AI can operate on, the two things it should never be trusted with, and five questions that separate a production system from a prompt box.
The five components a single-property site needs, the order to build them in, and the three failure modes — latency, staleness, and template drift — that make a listing site worse than none.
Sellers rank marketing as their first criterion when choosing an agent. Here is what belongs in a pre-listing package, the four tool families used to produce one, and how to choose by volume.
Build your first pre-listing package free — no card required.
Tell us a little about your business and we'll set up a demo on one of your own listings.