Accuracy of Zillow in predicting home sale prices
Saw an interesting article in the Wall St Journal this week. How Good Are Zillow’s Estimates? discusses, surprisingly, the accuracy of Zillow’s estimates. Bet you never saw that coming.
In any case, the article pulls out a few interesting data of interest to Zillow users, based on research performed by the Journal:
- * Median difference between Zillow’s ‘Zestimate’ and a property’s actual sale price was +/- 7.8%.
- * Zillow was within 5% on about 1/3 of properties.
- * Zillow was more than 25% off on about 11%.
- * Zillow tends to perform best in neighborhoods where houses are of similar quality and price range. It underperforms on very high end and very low end properties.
- * Zillow’s estimate can only be as accurate as the information available, which may not take into account, for example, that the house recently burned out or was flooded.
At this point in its existence, it should be clear to all reasonable observers that Zillow presents a data point. The numbers it provides should not be the sole basis on which you make a pricing or negotiating decision. However, the accuracy is reasonable, especially in the middle-market. And the ‘implied’ accuracy (Zillow presents its estimate of a house’s value right down to the dollar, ie. $157,243) can be a powerful psychological tool in real estate negotiating, where so many decisions are made solely on emotional, and non-data-driven bases.
Over time, we can expect Zillow, or competing sites with the same goals, to continue to refine their algorithms and improve their accuracy as more of a track record is established. It’s a powerful trend to watch, and another threat to the hegemony of real estate agents over the buying and selling process.
