Spencer Rascoff’s Seattle home sold for $1.05 million on February 29, 2016. The next day, Zillow showed a Zestimate of $1.75 million—about 40% above the sale price. Zillow’s 2016 explanation pointed to property features its model handled poorly, including the home’s triangular lot and position on a busy arterial. That account helps explain the mismatch, but it does not prove that $1.75 million was the home’s true market value.
What happened to Rascoff’s Zestimate?
The sale price was a completed transaction; the Zestimate was an automated estimate. GeekWire reported that Zillow’s estimate reached $1.75 million on March 1, 2016, the day after the $1.05 million sale. By the time of GeekWire’s May 23, 2016 article, Zillow displayed a Zestimate of $1.575 million. The figures are snapshots from different dates, not competing appraisals of settled value. GeekWire’s May 2016 account covers the sale and estimate history.
Why did Zillow’s estimate miss?
The lot and street were unusual
Zillow chief analytics officer Stan Humphries said the model compared the home’s triangular corner parcel on a busy arterial with nearby homes on rectangular lots. He said the rectangular parcels offered more utility, while parcel shape and arterial placement were areas the company was still researching. Humphries called the case atypical and said, “The computer should have done better in this case.” These were Zillow’s explanations in 2016, not a published audit establishing a single cause or a description of today’s model. GeekWire
Some features are difficult to represent in data
Inman’s 2016 coverage also cited design and lighting—features that can affect how buyers value a home but are difficult to quantify from standard property records. Zillow senior economist Skylar Olsen said the “Zestimate algorithm can’t know” some of these characteristics. The comment was about the model’s limitations as described at that time. Inman’s May 2016 report
#1 Best Overall
- Used Book in Good Condition
Local sale data and estimate smoothing mattered
Inman reported that Zillow lacked access to Northwest MLS sold-listing data in Seattle at the time. It also described a smoothing function intended to keep the Zestimate from reacting too sharply to a recent sale, which contributed to the estimate’s slow adjustment after the transaction. Both points describe the reported 2016 system; they should not be assumed to describe Zillow’s current data access or methodology.
What the listing history adds—and what it does not
Inman reported that the home was listed and delisted several times after its July 7, 2015 listing, underwent price reductions, and eventually sold 19% below its July listing price. It also noted that the July 2015 Zestimate had been revised retrospectively on Zillow’s page. That means a historical estimate visible now may not match the figure a reader would have seen at the time; the listing history does not by itself establish the home’s value on any particular date. Inman
How much confidence should readers put in a Zestimate?
A Zestimate is an estimate, not an appraisal or guaranteed sale price. Inman reported that Zillow described it as a conversation starter rather than the final word on a home’s value. In 2016, GeekWire cited Zillow figures of a 7.9% national median Zestimate error rate and a 6.1% Seattle median error rate. A median error rate means half of estimates were closer than that percentage and half farther off; it does not mean every estimate is wrong by that amount. GeekWire also reported that Zillow covered 102.7 million homes and that about 20% of sales had a Zestimate miss greater than 20%. These are historical figures reported in 2016, not current accuracy claims. GeekWire
An undated Zillow explanatory PDF hosted by iHouseprd describes the Value Range as a 70% confidence interval, with a wider range signaling less data or greater volatility. Because the document’s underlying publication date is unclear, treat that description as a historical explanation rather than current product documentation. Zillow explanatory PDF hosted by iHouseprd
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why one sale does not settle which estimate was better
Inman reported that Redfin’s estimate was $1.1 million the day after the sale, closer to the $1.05 million transaction price, and that Redfin’s estimate later rose. That is a comparison of estimates from specific dates, not proof that one service is generally more accurate. A useful comparison would need to hold the date and property record constant and examine local closed-sale data, lot and road characteristics, difficult-to-measure features, uncertainty ranges, and how promptly each model incorporated the recorded sale. No single home can establish a universal winner.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What to do if you need a value for your own home
Use an automated estimate as a starting point, not as a substitute for an independent valuation when the decision has financial or legal consequences. For a current estimate grounded in your property’s condition and features, consult a licensed residential appraiser. The Rascoff episode is a reminder that unusual parcels, street exposure, and details absent from property data can make an automated number diverge from a real transaction.
Quick Recap
Rank #4
- 4-1/4" x 7-3/4"
- 2-part carbonless forms, white and pink
- Snap-out
- 25 per book
- Packaged 5 books per pack
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
