Write listing descriptions that don't sound like every other listing

Open Zillow. Read 10 listings in a row. Notice they all sound identical — “charming”, “cozy”, “must see!”, “won’t last!”. Buyers’ eyes glaze over within three listings. AI can write you something different in under five minutes per listing — but only if you brief it like a copywriter, not a feed.

What MLS descriptions are actually for

The description isn’t there to describe — that’s what the photos and specs do. The description’s real jobs:

1. Surface buying triggers that don’t fit in specs (the morning light, the walking neighborhood, the school district nuance, the unique architectural detail).
2. Pre-qualify buyers so showings are with people likely to make offers.
3. Differentiate the listing in a feed of 50 similar properties.

Generic descriptions do none of these. AI-with-the-right-brief does all three.

The gather step (the one most agents skip)

Before AI can write anything good, you need 90 seconds of pre-work while you’re at the property. Voice memo or note these:

• The single best feature (the thing that closed YOU when you toured it)
• One quirk that’s actually a feature (“the kitchen is small, but there’s a giant pantry off the side most agents miss”)
• One quirk that’s actually a quirk (“low ceilings on the second floor”)
• The neighborhood feeling at 7am, 7pm, weekends (3 separate observations)
• The “who is this for” — the actual buyer profile
• 2–3 sensory details (light, sound, materials, views)

This 90 seconds is the difference between AI writing well and AI writing soup.

The listing description prompt

Common AI failures to watch for

Even with this prompt, AI will sometimes:

• Sneak in Fair Housing violations. “Family-friendly”, “ideal for active buyers”, and “great for young professionals” are all problems. Strip them.
• Invent features. If you didn’t say there’s a wine cellar, but AI thinks this kind of house has one, it might add it. Read every word.
• Default to “perfect for”. This phrasing is both weak copy AND a Fair Housing risk. Banish it.

The Fair Housing checklist

Before you submit, scan for:

• No language about race, color, religion, sex, national origin, familial status, or disability.
• No “perfect for” / “ideal for” + person-type language.
• Neighborhood described by amenities, not demographics (“near the farmers’ market” yes; “great schools, low crime” no — those imply demographic filtering).
• Pets language (“dog-friendly yard”) is fine; “kid-friendly” is not.

When in doubt, describe what’s there, not who it’s for.

What to expect

Listings written this way:

• Get 20–40% more saves in the first week vs. generic descriptions on comparable inventory.
• Pre-qualify buyers, so showings are higher-intent (fewer drive-bys, more offers per showing).
• Let your agency or brokerage stand out without you having to be a great writer yourself.

The AI is the writer. You’re the editor. The 90 seconds you spent gathering on-site is what makes both jobs possible.