Agent Copilot AI
← Blog

guide

How to Run a CMA: A Working Agent's Method for Pricing From Closed Sales

By Christiano Sampaio Β· November 2, 2026

A comparative market analysis is the most common piece of work a listing agent produces and one of the easiest to do badly. Done well it is a short, defensible argument about what a specific house should sell for, built from what comparable houses actually sold for. Done badly it is a stack of listings printed off the MLS with a number on the front page that the seller already wanted to hear.

This is the method, in the order you should actually do it.

First, be clear about what a CMA is not

A CMA is an informal market study prepared by a licensed real estate agent from recent comparable sales. An appraisal is a formal valuation performed by a licensed appraiser under the Uniform Standards of Professional Appraisal Practice, signed under that license, and used where an official value is required - lending, taxation, estate settlement, court. Only licensed appraisers can conduct appraisals, and preparing a CMA requires no license beyond a real estate license.

The practical consequence matters more than the definition. You must not present a CMA as carrying the weight of an appraisal. Rules vary by state, and several are explicit about it: in California only appraisers regulated by the Bureau of Real Estate Appraisers may conduct appraisals and agents must not imply appraiser credentials; Texas restricts appraisal to TALCB-licensed appraisers and requires that CMAs be clearly labelled as such; Florida places appraisals outside agent services under Chapter 475. An agent who also holds an appraiser license may appraise, but has to do it under the appraiser designation rather than the agent one.

So label the document, say plainly in the cover conversation that it is a pricing opinion rather than an appraisal, and check your own state's wording. None of this weakens the CMA. It is what keeps it credible.

Start with closed sales, not active listings

Active listings tell you what sellers hope for. Closed sales tell you what buyers paid. A CMA is built on closed sales; actives and pendings belong in it, but as context about current competition, not as the basis for the number.

Work outward in this order, and stop as soon as you have enough:

  1. Closed sales in the same building, tract, or street, most recent first.
  2. Closed sales in the same immediate neighbourhood with a comparable lot position and the same basic product type.
  3. Closed sales in a directly competing neighbourhood, if and only if you can say why a buyer would genuinely cross-shop the two.

The third step is where most CMAs go wrong. Two areas that look adjacent on a map are frequently not one market: a different school attendance area, a hill, a main road, or a different jurisdiction with a different building department can all split a market that looks continuous. If you cannot explain in one sentence why a buyer would consider both, it is not a comp.

How recent is recent

There is no universal answer and you should distrust anyone who gives you one, because it depends on turnover. In a dense condo building with frequent sales, a sale from three months ago is stale. On a street that trades twice a year, an eighteen-month-old sale may be the best evidence available.

The honest approach is to use the most recent sales you have, and then state the date range you used on the document. A seller who can see that your comps run from March to September reads the number differently from one handed an undated list, and they should.

Adjust for differences, and show your work

A comp is never identical, so every comp needs adjusting. The adjustments that usually carry the most weight:

  • Living area, which is the obvious one, and the one most often misapplied. Price per square foot is a result, not an input. Do not average the price per foot of a comp set and multiply it by the subject's square footage - on any street with a wide size range that method produces a number that is wrong in both directions.
  • Lot size, grade, and position, which on hillside and view property frequently matters more than the house.
  • The view itself, and whether it is protected. An unobstructed sightline and a partially screened one are different products even on the same street.
  • Condition and date of the last meaningful renovation, separating cosmetic work from systems and structure.
  • Beds, baths, and how they are arranged. A fourth bedroom in a converted garage is not a fourth bedroom.
  • Parking, especially in urban buildings where a deeded space can be a material part of the value.
  • Terms of the sale. A cash sale with a short close, a seller credit, or an off-market trade between known parties is not a clean arm's-length data point. Note it or drop it.

Write the adjustments down. The number at the end of a CMA is only as persuasive as the reasoning a seller can follow, and the adjustment list is the reasoning.

What to do when there are no comps

Sometimes there are genuinely only two usable sales. The wrong responses are to widen the geography until you have six, or to pad the set with actives. Both produce a false sense of precision.

The right response is to say so. Present the two real comps, explain what makes the subject unusual, give a range rather than a point estimate, and set out what the first two weeks on the market will tell you. A CMA that says "there is limited evidence, here is the range, and here is how we will read early showing activity" is more useful than one that manufactures confidence.

Present a range, then a recommendation

Give the seller three things: the range the evidence supports, your recommended list price inside that range, and the reasoning for where in the range you landed. Then pair it with a net sheet, because the seller's real question is not what the house is worth, it is what they will walk away with. A pricing conversation held without a net sheet tends to become an argument about the list price; held with one, it becomes a conversation about outcomes.

And be specific about what happens if the market disagrees. Name the review date, name the showing and inquiry thresholds you will treat as a signal, and agree in advance what a price adjustment would look like. Sellers accept a lower number far more readily when it arrives with a plan attached.

Where the tooling helps, and where it does not

The mechanical half of a CMA - pulling sales, assembling them, laying out a clean document a seller can read - is work that software should be doing. That is what the AI CMA Generator is for, and pairing it with the Seller's Net Sheet covers both halves of the listing conversation in one sitting.

What no tool can do is choose the comp set. Deciding that a sale three streets over belongs in the analysis and one next door does not, judging whether a view is protected, knowing which building has a pending special assessment - that is local knowledge, and it is the part of a CMA that actually determines whether the number is right. Use the tooling to remove the typing. Do not use it to remove the judgement.


This article is general information for real estate professionals and is not legal advice. A comparative market analysis is not an appraisal. Licensing rules for valuation work vary by state - confirm the current requirements in your own jurisdiction before describing or presenting pricing work to a client. Sources checked 2 November 2026: CMA vs. Appraisal, LicensePrep.