How Accurate Are Zillow Zestimates? I Tested the Home Valuation Tool Post-Upgrade

We all want a quick, one-stop shop to value properties these days, especially when investing out-of-state or when buying in volume, where it’s hard to look at dozens of properties every single week. Conveniently, many websites have popped up with handy algorithms to provide us the answer to the rather subjective question: How much is this property worth?

Such sites include Trulia, Redfin, Realtor.com, and eAppraisal, but the most famous of all is Zillow, with its tried-and-sort-of-true Zestimate.

Of course, this begs the question: How much can Zillow (or any of the others, for that matter) be trusted?

Should You Trust Zillow’s Zestimate?

The short version is that Zillow is a good place to start but should never be relied upon for a final determination of value.

This is much different than my answer five years ago when I would have said that Zestimates are pretty close to worthless. But Zillow has continuously been updating and improving their software. The newest version even sounds slightly Orwellian. (1)

“The Zestimate revolutionized real estate when it launched in 2006, using facts from public records to estimate a home’s value. With today’s update, it can now, in a sense, ‘see’ in photographs features that humans would instantly understand, such as curb appeal and natural light. The new Zestimate uses neural networks and computer vision to distinguish between high- and low-end finishes and to incorporate the value of features like updated bathroom fixtures, fireplaces, and remodeled kitchens. The Zestimate also now uses real-time data from for-sale homes, including list price and how long a home has been on the market.”

How Accurate Are Zillow’s Estimates After Updating the Platform?

Zillow also publishes a list of how its Zestimates compare to actual sales prices based on county, state, and metropolitan areas for both on-market and off-market (for sale by owner) sales. For the nation on the whole, here’s what their data says (2):

For states, the median error ranged from 1.4 percent for on-market sales to 11.1 percent for off-market sales.

However, I am a bit skeptical about the “neural networks” ability to evaluate listing pictures with a high degree of accuracy and believe that, at least in part, the list price itself affects the Zestimate. To illustrate this, here’s the last 12-months of the original listing price versus the price sold in Kansas City, Mo.:

Pretty close indeed.

In August 2019, the average sale price in Kansas City was $208,198 and the average list price was $212,167. This is a 1.9 percent difference, which is exactly the same as the median error for Zestimates versus on-market properties sold in the United States last year.

As an example, we just sold a small two-bedroom home. We ended up going over budget on the rehab and therefore failed to “BRRRR out.” The Zestimate was at $83,500, but my valuation of the comparables made me believe it was worth more.

We listed it for $94,900 and immediately the Zestimate jumped up to $96,100. Then, we sold it for $93,000.

Now, perhaps those “neural networks” saw that our listing pictures had very nice finishes in them and our property was in really good condition. Sure, but it was originally intended to be a rental, so while the work was well-done, it was nothing fancy. So, while the Zestimate is only off by 2.3 percent now, it was off by 11.4 percent before we listed it.

Thereby, a better way to evaluate the quality of Zestimates may be appraisals, which would eliminate the listing price as a confounding variable. That being said, in Zillow’s defense, without the listing pictures, it can’t evaluate the property’s condition very well and appraisals are by no means perfect either.

Appraisals vs. Zestimates: Which Are More Accurate?

We just got appraisals on 12 properties we are refinancing, and here’s how they compare to the Zestimate:

Admittedly, some of these appraisals I disagree with to an extent—but not necessarily in the direction that would help Zillow. Overall, there’s quite a range with the Zestimate, including half being over 10 percent off the appraisal and two being over 20 percent.

Therefore, I think it’s fair to say that Zestimates are not that great when it comes to off-market properties and are probably, at least in part, only so accurate on listed properties because the list price is also accounted for. In that way, they are riding off of the work of real estate agents, investors, and homeowners, who valued the properties themselves before listing them.

Despite advances in technology that allow for photo recognition and the like, there is only so much these programs can do. They can’t evaluate the condition of properties with no pictures. They can’t take into account major structural issues or recent upgrades.

It’s also all but impossible to differentiate between two things that are listed the same way but are realistically of very different value. For example, take a finished basement. One house may just have a basement with a painted floor and walls that gets water in it every fall. Another may be a walkout basement with plenty of sunlight that is fully furnished and decorated. It’s very difficult for an algorithm to tell the difference.

The Bottom Line

In summary, Zillow is a good place to start, as are the other websites with similar tools. If you don’t know the value of properties in a particular area, then check the Zestimate to get a rough idea. Then, you’ll at least have some idea whether a property is worth pursuing or not.

But that’s about where its use ends. Never rely on the Zestimate for what your offer or strike price should be. For that, you’ll need to do a good old-fashioned comparative market analysis.

References

How do you feel about Zillow? Have you had any good or bad experiences with Zestimates?

Share in the comment section below.

Note By BiggerPockets: These are opinions written by the author and do not necessarily represent the opinions of BiggerPockets.