How to Write a CMA Narrative with AI in 2026

The CMA narrative is the part of a listing presentation that sellers actually read. The spreadsheet of comparable sales tells them what the data is. The narrative tells them what the data means — and why your suggested list price is the right call. Writing the narrative by hand for every listing presentation is what most agents do badly because they're tired of rewriting the same paragraphs from scratch. AI handles this in under three minutes.

This is a practical walkthrough for writing a CMA narrative with AI that closes listing appointments.

What a great CMA narrative contains

Before you can use AI well, you need to know what good looks like:

• Subject property recap — the basics, no surprises
• Comparable sales summary — the 3-5 comps you used and why
• Adjustments explanation — why the comps are close but not identical
• Market conditions context — current trends in the area (DOM, list-to-sale ratio, inventory)
• Suggested list price recommendation — your number with the reasoning
• What to expect — likely buyer pool, timeline, negotiation expectations

The agents who win listing appointments are the ones whose CMA narrative makes the seller feel informed, not sold to. AI is excellent at producing the informational layer.

The right prompt structure

The mistake most agents make on first try is pasting MLS data and asking for "a CMA writeup." The prompt that actually works gives the AI the structured comp data and your recommended price:

<task>Write a CMA narrative for a listing presentation.</task>

<context>
Subject: 3BR/2BA ranch, 1,800 sqft, updated kitchen, original baths,
corner lot, Maplewood NJ, built 1968

Comp 1: 22 Elm St, sold $485,000, 1,750 sqft full reno, 45 days ago
Comp 2: 18 Oak St, sold $462,000, 1,900 sqft original condition, 30 days ago
Comp 3: 7 Pine St, sold $478,000, 1,820 sqft updated kitchen+baths, 60 days ago

Market context: median DOM 28 days, list-to-sale 97.2%, inventory 2.1 months,
prices stable with slight upward pressure

Recommended list: $469,000
</context>

<instructions>
- Tone: warm, informative, not salesy
- Three sections: comparables analysis, market context, recommended pricing strategy
- Explain why subject is between Comp 1 (renovated) and Comp 2 (original)
- Include expected DOM and likely buyer pool
- 400 words max
</instructions>

<avoid>
- Inventing comp data I didn't provide
- Promising specific outcomes (sale price, days on market)
- Generic real estate clichés
</avoid>

Notice the structure: facts, your recommended price, and explicit instructions about what NOT to invent. The AI produces the narrative; you provide the comps and the price.

What to never let AI do

Pull comps for you. AI tools do not have access to current MLS data. Any comps the AI "finds" are fabricated. Always pull comps yourself and paste them into the prompt.

Recommend a price. Pricing is your judgment work — your knowledge of the market, the seller, the property condition. The AI scaffolds the narrative around your number; it does not pick the number.

Promise specific outcomes. "This will sell in 14 days" is the kind of claim that creates problems if it doesn't. Stick to "we expect typical market timing" or use ranges.

Make adjustments without your input. The AI doesn't know that the kitchen update added $25K of value. You do. Tell it.

Common mistakes

Skipping the market context. Comps without context are just numbers. The current market conditions are what make the recommendation defensible.

Over-explaining the math. Sellers want the conclusion, not the methodology. Keep adjustments brief.

Forgetting the buyer pool. "Who is going to buy this house" is the question every seller is secretly asking. Answer it.

Using AI clichés. "Stunning," "must-see," "won't last" — strip them out of the narrative. CMAs are not marketing copy.

Want the whole system, not just this one workflow?

The Real Estate Agent Claude Vault includes 50 real-estate-specific Claude prompts (listing descriptions, client emails, CMAs, social posts, transaction management), a complete Claude Project setup file with fair housing guardrails, and three fair-housing-compliant listing templates. $29, one-time. Get the vault →

The free tool that handles this for you

If you don't want to engineer the prompt every time, the CMA Narrative Generator on AI Career Lab is pre-configured for the format that wins listing presentations. It produces structured narratives with the elements above, in the warm-but-informative tone sellers actually respond to.

Pair it with the Listing Description Generator for the post-listing marketing copy and the Client Email Generator for the follow-up communication that turns the listing presentation into a signed agreement.

Free with an AI Career Lab account, capped at five runs per day on the free tier.

Try it on your next listing presentation

Pick a property you're presenting on this week. Pull your comps. Decide on your recommended price. Run the data through the tool above. See how close the output is to what you would have written by hand. The narrative is the part of the listing presentation that turns "thinking about it" into "signed agreement." Worth the two minutes.

Create your free AI Career Lab account and try the real estate tools today. No credit card.