Your CMA vs. the Zestimate: The Algorithm Starts the Conversation. You End It.

Every seller opens the door already knowing their number. They did not get it from a CMA. They got it from Zillow, and they believe it.

That is the modern listing appointment. In the 2026 Survey of Best Practices, 63.7% of agents named pricing as the number one seller objection, up 11 points since 2020. The reason is not a mystery. Sellers now arrive anchored to a Zestimate and a screen full of public data. They have more ammunition than ever.

Here is the good news. The same data that arms the seller is exactly where you hold the edge.

Sellers trust the Zestimate. Agents trust the MLS.

When we asked 2,165 agents which data source they trust most, 90.8% said their MLS. Public portals like Zillow scored under 1%. Not close. The MLS is where the real, timely, verified sale data lives, and access to it is your professional edge. A consumer can pull a Zestimate in three seconds. Only you can pull the comps that actually explain a price.

That gap is the whole job. The Zestimate starts the conversation. The CMA ends it.

How accurate is the Zestimate, really?

Give Zillow credit. The Zestimate has improved. As of 2025, Zillow reports a national median error of 1.83% for homes that are actively on the market. That sounds tight. Then read one line down. For off-market homes, the ones not currently listed, the median error jumps to roughly 7%. When a home is not on the market, the algorithm is flying blind.

Two more things sellers never hear. First, "median" means half of all estimates miss by even more than that. Second, individual homes can be wildly off. One documented mountain property sold for more than 70% above its Zestimate. Academic research on New York City homes found individual errors of 17 to 18%.

Geography matters too. In newer, data-rich markets like Utah, Nevada, and Colorado, Zestimates land within roughly 1.3% of the sale price. In older, more varied markets like New Jersey, Connecticut, and New York, the error climbs past 3%. The algorithm is only as good as the data underneath it, and no two neighborhoods feed it the same data.

Why the gap exists

An algorithm cannot walk the home. It has not seen the renovated kitchen, the foundation crack, the lot that backs to a highway, or the primary suite added last spring. It cannot read the street. It cannot weigh the ten small things a local agent clocks in the first thirty seconds. As the book puts it, a Zestimate is no substitute for a site visit from an experienced human.

This is not anti-technology. It is the opposite. More public data makes the person who can interpret it more valuable, not less. Since Zillow launched the Zestimate in 2006, the CMA has only grown more relevant.

Do not fight the Zestimate. Use it.

The instinct is to argue with the number. Resist it. The book calls the better move Zestimate judo: use Zillow's own weight in your favor.

If the Zestimate happens to line up with your pricing, say so out loud. Two independent methods reaching the same answer builds a seller's confidence fast. When the Zestimate is off, do not just wave it away. Show the seller specifically why the algorithm is wrong for this property, this street, this condition. That is where a CMA earns its keep.

There is even a prospecting play. Hypothetically: scan your database for homes whose Zestimate reads well below true value. Print it, highlight the number, and add a short handwritten note offering a real valuation. It costs almost nothing and puts you top of mind in a way another email never will.

One more thing worth respecting. Part of the Zestimate's power is emotional. It is simple, instant, and reassuring. You are not just competing with a number. You are competing with that feeling of certainty. So match the clarity, then add the nuance only a professional can.

The survey and the book land in the same place. Local MLS data and a human-driven CMA are still the gold standard for pricing a home. The Zestimate is a great conversation starter. It is not a conclusion.

Let the seller bring their number. You bring the story behind the right one.

Key takeaways

• 90.8% of agents trust the MLS above all other sources. Public portals scored under 1%.
• Zillow's own 2025 figures: about 1.83% median error on-market, roughly 7% off-market. Median means half miss by more.
• Individual homes can be off by double digits, sometimes far more, and accuracy swings by region.
• Do not argue with the Zestimate. Confirm it when it agrees, and show exactly why it is wrong when it does not.
• The Zestimate starts the conversation. Your CMA ends it.

The chapter "Zestimate, Schmestimate?" in The Art of the CMA breaks down how to read AVM accuracy, handle the "Zillow says it's worth more" objection, and turn the algorithm into your best prospecting tool. Get your copy.