Between January 16, 2019, and June 14, 2019, we recorded the prices of 532 home sales across the Tri-Cities. We also recorded the public Zestimates® of those homes as shown on Zillow.com the day of each home’s closing. Here are the key takeaways from our research:
For more detail on each of those, read on.
Zillow uses a metric called “median error rate” to gauge the accuracy of Zestimates®. In late January 2019, Zillow said its median error rate nationwide is 4.5 percent. That means half of all Zestimates® are within 4.5 percent of a home’s real sales price.
According to our research, the median error rate of Zestimates® in the Tri-Cities is 5.6 percent. In other words, half of local Zestimates® are wrong by more than 5.6 percent of the actual selling price and half are wrong by less than 5.6 percent. That’s worse than the national median error rate, and also worse than our big city neighbors in Seattle (2.2 percent) and Portland (1.5 percent). Across Washington state, the median error rate is 2 percent, according to Zillow’s own tracking.
Zestimates® in each city we studied — with the exception of Pasco — are worse than the national 4.5 median error rate. Pasco has the lowest median error rate locally at just 2 percent. At 8.7 percent, West Richland has the highest median error rate in the Tri-Cities area.
The median error rate tells us how wrong Zestimates® are, but it doesn’t indicate whether the Zestimates® are higher or lower than the real sales price of homes.
Based on our research, a significant majority of Zestimates® were higher than the homes’ sales price in all cities we studied.
Across the Tri-Cities, almost four of every five Zestimates® was higher than the home’s price — 79 percent, to be precise (far right column). This was most pronounced in Kennewick, where 87 percent of Zestimates® were too high, and only 13 percent were too low.
Our data makes it clear that Zestimates® in Pasco are more accurate than anywhere else in the Tri-Cities.
In addition to the charts above, let’s look at how many Zestimates® were wrong by more than 5 percent in each city. Across the Tri-Cities, 52 percent of Zestimates® were at least 5 percent wrong. In Kennewick and Richland, more than 60 percent of Zestimates® were at least 5 percent wrong. But in Pasco, that figure is less than 20 percent.
In addition to studying accuracy based on what percentage a Zestimate® was compared to a real sales price, we can also study how close Zestimates® and sales prices were in terms of raw dollars. This view again shows that Pasco has the most accurate Zestimates® in the Tri-Cities.
Half of all Pasco Zestimates® were within $5,000 of the home’s sales price. In the three other cities we studied, only about 10 percent of Zestimates® were within $5,000 of the sales price.
Determining which city has the least accurate Zestimates® isn’t as clear-cut as seeing that Pasco’s are most accurate. In the second chart above, we show West Richland as having an 8.7 percent median error rate — the highest in the area. That means half of West Richland Zestimates® are more than 8.7 percent wrong.
But a couple charts later, we show West Richland with only 58 percent of its Zestimates® being more than 5 percent wrong — that’s better than both Kennewick and Richland.
On that chart, Richland has the highest percentage of Zestimates® (68%) that were more than 5 percent wrong. As the next two charts show, Richland also has the highest percentage of Zestimates® that are 10 percent and 20 percent wrong.
The answer to which part of the Tri-Cities area has the worst (most inaccurate Zestimates®) may depend on how you choose to define “worst.”
According to our research, 16 percent of Tri-Cities homes — about one of every six homes sold — had a Zestimate® that was at least $50,000 wrong.
Our research also suggests that the accuracy of Zestimates® can be connected to a home’s sale price. According to our research, homes that sold in the $225,000 to $299,999 range had significantly more accurate Zestimates® than any other price range.
Tri-Cities Zestimates® got slightly more accurate as our study progressed. Each measurement point — 5 percent, 10 percent and 20 percent wrong — showed improvement between February and May. (Note: We don’t include January and June in monthly trends since we don’t have a full month of data for those.)
We’ve helped hundreds of families and individuals buy and sell homes across the Tri-Cities. We’re ready to help you, too.
Tell Us How We Can Help!