If I had to write a listing description in the next 90 seconds and I’d never seen the property before, I would open ChatGPT, paste my exact prompt below, drop in 5 inputs, and hit enter. I’d get a 220-word, fair-housing-compliant description that uses Zillow’s data-backed power words, hooks the buyer in the first sentence, and ends with a call-to-tour line — every single time.
That’s the whole tip. The rest of this post is the prompt, the 5 inputs, the edits I always make before pasting it into the MLS, and the 4 traps that get agents in trouble with fair housing.
Before I show you the prompt, we need to establish the rules
Before you copy a single AI-generated word into your MLS, there are three non-negotiables. Skip these and the prompt won’t save you — it’ll just generate problems faster.
First, every MLS has a character limit on public remarks, and they vary wildly. Bright MLS (Mid-Atlantic) caps you at 1,000 characters. CRMLS (California) gives you 1,500. Most boards land somewhere between 800 and 1,500. Know your board’s limit before you write a 4,000-character novel ChatGPT will happily produce if you don’t constrain it.
Second, the data-proven sweet spot for converting listing descriptions is around 250 words — long enough to be thorough, short enough to actually get read. The average MLS description today clocks in around 60 words, which is a wasted opportunity. NAR’s 2025 Profile of Home Buyers and Sellers found 78% of buyers read listing descriptions “carefully” or “very carefully” before requesting a showing. That’s the highest-leverage 250 words you’ll write all month.
Third — and this is the one that gets agents fined — the Fair Housing Act prohibits any preference, limitation, or discrimination based on race, color, national origin, religion, sex, disability, or familial status. HUD doesn’t keep a strict banned-words list anymore. They look at the whole ad in context. But there’s still a list of phrases that reliably trip the wire — “family home,” “perfect for couples,” “near St. Mary’s,” “safe neighborhood,” “walking distance,” “master bedroom” in some states. The prompt below has fair housing rules baked in. Don’t strip them out.
78% of buyers read listing descriptions “carefully” or “very carefully” before booking a showing (NAR 2025 Profile of Home Buyers and Sellers). The optimal length is 150–300 words. The average is 60. That’s the gap.
Shame on me — how I used to write listing descriptions
For my first 18 months as an agent I wrote every listing description from a blank Word doc at 10 PM the night before the photos went live. I’d stare at a feature sheet, type “Welcome home to this charming 3-bedroom…” and immediately want to throw my laptop. The descriptions were generic. They opened with “Welcome home” or worse, with a square-footage number. I used “family room” without thinking. I once wrote “great for empty nesters” — which is a familial-status fair-housing violation that a compliance officer caught before it went live. Shame on me. The prompt I’m about to give you is the system I built so that never happens again.
I know what you’re thinking
I know a lot of you reading this think AI listing descriptions sound like a robot wrote them, and you’d rather just write your own — which is totally fine. But here’s the really big problem: the agents I see writing their own descriptions are spending 30–45 minutes per listing and producing 60-word remarks that sound exactly like every other description on the MLS. The point of the prompt isn’t to remove you from the process. It’s to get you a strong 80% draft in 90 seconds so you can spend 5 minutes adding the human voice — instead of 45 minutes building from scratch and giving up at “charming.”
The 90-Second Listing Description Prompt (My Exact Setup)
Here’s the exact 5-step workflow I run for every listing. It takes 90 seconds if you have your inputs ready.
Step 1: Gather your 5 inputs before you open ChatGPT
The prompt only works if you feed it real, specific information. Open the property’s feature sheet or your seller’s questionnaire and pull these 5 things:
- Address + basic stats: beds, baths, sq ft, lot size, year built, school district (if it’s a selling point — and not a fair-housing trip wire in your market).
- The 3 hero features: the things that would make a buyer drive across town. Renovated kitchen with quartz counters? Pool? View? Workshop? ADU? Pick 3.
- The lifestyle hook: who’s the dream buyer and what does life look like in this house? “Coffee on the back deck watching the sunrise over the cascade foothills.” Not a buyer profile — a moment.
- Neighborhood or location anchor: the walkable coffee shop, the trail access, the boat launch. One concrete proof point that makes the location real.
- The sentence you don’t want left out: the seller’s favorite thing about the house, the one feature you noticed on your walkthrough that no listing photo can capture.
Step 2: Paste this prompt into ChatGPT
This is the prompt I’ve been refining for two years. Copy it word for word. The constraints at the bottom are what keep it out of fair-housing trouble.
PROPERTY INPUTS
– Address & basics: [PASTE]
– 3 hero features: [PASTE]
– Lifestyle hook (the moment in the house): [PASTE]
– Neighborhood / location anchor: [PASTE]
– Don’t-leave-out sentence: [PASTE]
TARGET
– Length: 220–250 words. Hard ceiling: 1,300 characters.
– Reading level: 6th grade. Short sentences. Active voice.
– Tone: warm, confident, specific. No “welcome home,” no “nestled,” no “boasts,” no “must see.”
STRUCTURE
1. Opening line: a sensory or lifestyle hook tied to the strongest hero feature — never a square footage stat. 12–16 words max.
2. Body: 3–4 short paragraphs. Lead with the strongest hero feature. Use 3 of these Zillow-data words only if they’re true: luxurious, captivating, impeccable, stainless, granite, landscaped, remodeled, upgraded, beautiful.
3. Neighborhood paragraph: name 1 specific nearby anchor. No school names. No religious institutions. No demographic language.
4. Closing line: a clear call to schedule a private tour.
FAIR-HOUSING GUARDRAILS (HARD RULES)
– Never use: family, families, kids, children, bachelor, empty nester, couples, singles, seniors, retirees, walking distance, safe, exclusive, private community, near [house of worship], perfect for [demographic].
– Replace “master bedroom” with “primary bedroom.”
– Describe the property, never the buyer.
OUTPUT
– Plain text only, no headings, no bullets, no emojis.
– After the description, list any phrase you considered cutting for fair-housing reasons and why.
Step 3: Review the fair-housing audit ChatGPT outputs
The last instruction in the prompt — “list any phrase you considered cutting for fair-housing reasons” — is the part most agents skip. Don’t. It surfaces the borderline language ChatGPT almost included so you can make the final call. Nine times out of ten it’ll flag something like “family-friendly layout” or “near the elementary school” that I would’ve missed on a tired night.
Step 4: Run the 4-edit pass (60 seconds)
The first draft from ChatGPT is an 80% draft. Here are the four edits I make every single time:
- Replace one generic word with a specific one. ChatGPT loves “spacious.” Swap it for “11-foot ceilings” or “1,200 sq ft of finished basement.”
- Add one human detail. Something only you know from walking the home. “The seller’s favorite morning spot is the bench under the pergola.” Buyers feel that. ChatGPT can’t.
- Cut every adverb. If a sentence has “really,” “very,” or “incredibly,” delete it. The sentence will be stronger.
- Read the first line out loud. If it’s boring, rewrite it. The first 12 words decide whether the buyer keeps reading or scrolls.
Step 5: Character-count check before you paste
Before pasting into the MLS, paste it into a free character counter (or use the word count built into Google Docs). If you’re over your board’s character limit, cut from the neighborhood paragraph first — it’s the easiest to trim without losing punch. Never cut the opening line or the call-to-tour close.
Take Your Real Estate Business to the Next Level
The 4 traps this prompt avoids (and most agents fall into)
| Trap | Why It Hurts | What to Do Instead |
|---|---|---|
| Opening with “Welcome home” or square footage | It’s the most-skipped line on the MLS. Buyers tune out instantly. | Lead with a sensory or lifestyle hook tied to the strongest feature. |
| Using “family home,” “kids,” or “empty nester” | Familial-status fair housing violation. Real fines, real complaints. | Describe the property’s features. Let the buyer self-select. |
| Naming schools, churches, or “safe” neighborhoods | Religious/national-origin/race steering risk under HUD’s contextual standard. | Name a coffee shop, trail, or transit anchor instead. |
| A 60-word description (the average) | You’re leaving the most-read 200 words on the table. | Aim for 220–250 words inside your MLS character cap. |
The Zillow words that actually move price
Zillow ran the data on which adjectives correlate with above-expected sale prices. Two stand out. Bottom-tier homes (lowest third of their market) described as “luxurious” beat their expected sale price by 8.2%. Top-tier homes described as “captivating” beat theirs by 6.5%. Other words that correlated with stronger pricing: “stainless,” “granite,” “landscaped,” “remodel,” “upgraded,” “impeccable,” “beautiful,” “pergola.” The prompt above is wired to use these — but only when they’re actually true. If you call a 1962 kitchen “luxurious” and the photos show a Formica counter, buyers will spot it instantly and trust drops to zero.
4 listings a month × 40 minutes saved per description = 160 minutes (2.7 hours) back. Over a year that’s 32 hours — a full work week — just from this one prompt. And that’s before you count the 8.2% lift Zillow attributes to using the right power words on the right tier of property.
How this fits into a bigger AI stack
This prompt is one of about 12 I run for every transaction — listing description, social caption, just-listed email, neighbor letter, open house recap, buyer welcome packet, and so on. They all live in my 2026 AI tools guide, which walks through every tool I use and which ones are worth paying for. If the listing prompt is the gateway drug, the full stack is what actually gets your weekends back.
If you’re running listings through a CRM that auto-launches a marketing sequence (just-listed email, social posts, drip), pair this prompt with a GoHighLevel setup and the description writes itself, then the marketing fires itself. That’s the move. And if you’re tired of paying a brokerage to do the marketing your CRM already does, here’s the full Real Broker breakdown on the brokerage I moved to so I could keep more of every commission and run the tech stack on my terms. For more on full automation, see the real estate automation masterclass.
Reassurance: imperfect descriptions still beat blank pages
If your first AI-drafted description sounds a little stiff, that’s totally fine. The point isn’t a perfect description on draft one. The point is getting you from blank-page-at-10-PM to a strong 80% draft you can polish in 5 minutes. After 10 listings using this prompt, you’ll start tweaking the prompt itself — adding constraints for your market, banning words your specific buyers hate, swapping in your signature closing line. That’s when it stops being “AI wrote my listing” and starts being “I wrote a system that writes my listings.” Big difference.
Your action step (do this in the next 5 minutes)
Open ChatGPT. Copy the full prompt above. Paste it into a new chat and save the chat with the name “Listing Description Engine.” Do not run it yet — just save it. Then on your next listing, the friction is zero: you open the saved chat, drop in your 5 inputs, and you’ve got a draft in 90 seconds. The agents who actually use systems are the ones who set them up before they need them.
FAQ
How long should a real estate listing description be in 2026?
The data-proven sweet spot is 150–300 words, with 220–250 words being the conversion peak. The average MLS description is around 60 words, which underuses the most-read 200 words on a listing. Always check your specific MLS board’s character cap — most fall between 800 and 1,500 characters.
Will MLS boards penalize me for AI-generated listing descriptions?
No MLS in the U.S. currently bans AI-assisted descriptions. What MLS boards do enforce is accuracy and fair housing compliance. As long as the description is truthful, within the character limit, and free of fair-housing violations, AI authorship is irrelevant.
What words are banned in real estate listings under fair housing?
HUD does not maintain a strict banned-words list anymore — they assess context. But high-risk phrases include “family home,” “kids,” “empty nester,” “couples,” “singles,” “bachelor,” “walking distance,” “safe neighborhood,” and naming schools or houses of worship. The prompt above has these wired in as hard rules.
Should I use ChatGPT or a custom GPT for listing descriptions?
Start with the prompt in plain ChatGPT. Once you’ve run it 10+ times and refined it for your market, build it into a Custom GPT so your team can use it without the prompt setup. Custom GPTs are the long-term play; the prompt is the fast start.
Does ChatGPT need access to MLS data to write the description?
No. ChatGPT does not need MLS access. You feed it the 5 inputs (address, hero features, lifestyle hook, location anchor, don’t-leave-out sentence) directly in the prompt. This keeps the workflow simple and avoids data-sharing concerns with the MLS.
How do I avoid the AI-generated “tone” in my listing description?
Run the 4-edit pass: replace one generic word with a specific number, add one human detail you noticed in person, cut every adverb, and rewrite the first line out loud. These four edits take 60 seconds and remove the “robot tone” 95% of the time.
Can I use this prompt for rental property listings?
Yes, with one swap. Change “private tour” in the closing line to “showing” or “application.” The rest of the prompt — fair housing rules, length, structure — applies identically to rentals. The fair-housing rules are actually more strictly enforced on rentals than sales.
Run your business on systems, not willpower.
The Automated Agent Workshop is the free training that walks through every prompt, automation, and AI workflow I use as a solo agent. No more 10 PM listing descriptions.
