Zillow Is Out of the Home Buying Business, and We Are Not Surprised
Zillow is exiting the home buying market. The stated reason is that they were not able to predict house prices and were therefore are losing money. I believe their problem goes far beyond this.
Zillow made several fundamental mistakes. Below are just a few:
- They did not select their target buyer and work backward to the properties they should purchase. This is a fatal error. Every property targets a narrow pool of people. They should ensure that every property they buy conforms to a specific pool of people who are willing and able to pay premium prices.
- From what I can see, they based their offers on their Zestimates, which are highly inaccurate. They should have looked at a property, decided what it would sell for after it was market-ready, and derived the offer price from that. I saw no evidence they were doing this. If they did it the right way, they would buy fewer properties, but they would have consistently made money.
- Another important factor they did not consider is that the renovation must match the specific buyer segment you are targeting. They seemed to be renovating every property the same way, which is an obvious error.
- They seemed to use "national" data. To be successful with real estate, you need to base decisions on hyper-local data. For example, I looked at a Zestimate recently for a 380,000 Las Vegas home, and below is how their data compared to actual data:
| Cost Item | Zillow | Actual/Comment | Zillow Error |
|---|---|---|---|
| Principal and interest | $1,151/Mo. | The payment is correct based on a 3.5% down FHA type loan, 30 year, 3.5% | $0 |
| Mortgage Insurance | $0 | Invalid. Unless you put down 20% or more, you will have to pay mortgage insurance. According to one website, the mortgage insurance would be about $180/Mo. | $180/Mo. |
| Property Taxes | $269/Mo. | Actual taxes for this property are $123/Mo. I believe they were using some sort of national average. | $146/Mo. |
| Home Insurance | $122/Mo. | We just purchased a more expensive property ($420,000) and the insurance was $550/Yr or $42/Mo. Again, I believe they were using some sort of national average | $80/Mo. |
| Total error by Zillow | $406/Mo. |
With such inaccurate data, how could they hope to provide accurate Zestimates. This same inattention to accuracy was no doubt why they failed to price properties correctly when they were flipping properties.
When I look at the overall state of their software and processes, they are about at the same point we were 13 or 14 years ago. At that point, we realized that you must use hyper-local information if you want to produce meaningful information. Without all the hyper-local data we've accumulated over the years and our software being optimized for Las Vegas, we would be in the same state as Zillow.
The takeaway for Zillow and others should be that you need to make investment decisions based on a specific buyer (or tenant) pool and hyper-local data. National date and "average" buyers are not relevant.