The Zestimate on my third house was $428,000. My listing agent’s comp-based price was $412,000. I wanted to list at $435,000 because I had put a new roof on it the previous fall and I was convinced the roof should add to the price. I was wrong about the roof. My agent was right about the price. The house sold in nine days at $411,500, which is the kind of outcome that looks like luck and was actually just listening to someone who had run the numbers more carefully than I had.
This is the article about how that agent ran those numbers, and what you should do when your Zestimate and your listing agent hand you two different numbers and neither of them is obviously right.
Three numbers, three different things
When you start thinking about selling, you usually end up with three price estimates in front of you, and it helps to know what each one is actually measuring.
The Zestimate is an algorithm that uses public records, recent sales in your ZIP code, and a model that Zillow refines over time. Zillow publishes its own accuracy data, and the numbers are more interesting than most sellers realize. The median error rate for homes currently listed for sale is about 1.83 percent. For homes not on the market, the median error jumps to around 7 percent. That is a big gap, and the reason is that Zillow has access to much more signal on an active listing than on a house that has not changed hands in years. The error on any specific house can also be much higher than either median, especially if your house has features the algorithm cannot see from public records. A finished basement that was never permitted. A gutted-and-rebuilt kitchen a previous owner did without pulling permits. A bad neighbor across the street that no model can detect.
A comparative market analysis, or CMA, is what a listing agent builds by hand. The good ones pull the last six months of sold comps within a specific radius, filter for similar square footage and bedroom count, adjust for differences like lot size and condition, and walk you through their reasoning. The bad ones show you a printout with five random comps and a recommended list price circled in marker. The quality of a CMA is the single biggest reason agents are worth talking to before you list, even if you think you already know your price.
An appraisal is what a licensed appraiser produces for a buyer’s lender after you have accepted an offer. It is the one estimate you cannot see in advance, and it is the one that can blow up your deal if it comes in below the contract price. Appraisers use their own comp methodology, and good ones often converge on a number close to the CMA.
Those three estimates rarely agree. My job as a seller was to understand why they did not.
The overpricing penalty nobody tells you about
The intuition most sellers bring to pricing is “list high, come down if I have to.” This feels like it preserves your upside. The problem is that the research keeps finding the opposite pattern. A Zillow Research study on overpricing documented that listings which linger end up selling for meaningfully less than their asking price, with the penalty getting worse the longer the house sits. That is the quantified version of something every experienced listing agent will tell you for free.
There are a few reasons the penalty shows up. The first is that the strongest buyer demand for your house arrives in the first two or three weeks. If your price is too high during that window, the buyers who would have been most motivated to offer fast move on. When you reduce later, you are fishing in the pool of buyers who are still looking after the first wave has cleared, which is a smaller and more cautious pool.
The second is days-on-market. Every listing platform shows how long a house has been for sale. Past a few weeks, buyers and their agents start assuming there is a reason it is still sitting, and they price that assumption into their offers whether the assumption is fair or not. A price reduction does not fully reset that clock.
The third is that sellers who overprice tend to also be sellers who negotiate emotionally later. Once you have anchored yourself to “the price I should be getting,” every buyer offer feels like an insult, and you end up giving concessions in the closing costs column because you could not budge on the headline number. The final net price is often worse, and the path to get there is longer and more stressful.
What to actually do when Zillow and your agent disagree
Here is the part where most articles give you a formula. I am not going to, because there is not one, and the sellers I have watched do this well all followed the same three-step process instead of a formula.
Ask the agent to show their work. If the CMA is a printout with no walk-through, push for the walk-through. Which six comps did they pick, and why those six? Which ones did they adjust up or down, and by how much? If they cannot answer confidently in five minutes, their number is not worth more than the Zestimate.
Pull the comps yourself. Zillow and Redfin both let you filter recent sold comps by neighborhood, size, and bedroom count. Spend twenty minutes looking at the last six months. You are not trying to out-analyze the agent. You are trying to see the distribution, so when the agent tells you “the band is $405 to $420,” you can see that band with your own eyes and trust it.
Price at the band, not above it. If your CMA band is $405 to $420, list somewhere inside that range. If you feel emotionally that your house deserves $435 because of the roof, the kitchen, or the memory of what you paid in 2018, that emotion is real and it is also irrelevant to the 2026 buyer. The buyer is comparing your house to the other ones on Zillow that cost the same, and nothing about your memory of the house is in their comparison.
What the agent was right about
The pricing question ends up being less about pricing and more about whether you can tell yourself the truth about what your house is worth to someone who is not you. My agent on my third sale did not tell me the roof was worth nothing. He told me the roof was worth about $3,000 in the comp math, because a two-year-old roof and a twelve-year-old roof are different numbers, and the buyer adjusts for one but not the other. He was right, and I was annoyed, and the house sold in nine days. That is usually how it goes.