This weekend I was asked to lend on a file in Colorado.  The borrower mentioned his zestimate was 250k.  I was intrigued because the house was not in the nicest area.  His zestimate value sucked…. Zillow also thinks it sucks since they are offering $1m to anyone who can help improve the algorithm.  Why is the zestimate so wrong? How can you know if your zestimate sucks?

According to a Washington post article that examined zestimates to actual sales, some zestimates were as low as 64% of sale price and some were 256% above sales price.  I would be hesitant to bet my house on the zestimate value!

Let’s talk first about why the zestimate above was so wrong. The zestimate was 250k, yet when I pulled comparables the house has sold twice in the last year (no change in square footage or major remodels).  It sold 9 months ago for 90k and recently for 150k. There were 3 houses on the same street or within a block or so of the subject that averaged 150-170.  There is no way this house was worth 250k.  Zillow is also convinced it’s zestimate isn’t that great as well since it is offering 1m to anyone who can help improve the accuracy of the zestimate (see article).

Whenever I ask a borrower what their property is worth, more often than not they respond: “The Zillow value is X.” This got me thinking: How accurate are online realty sites (Zillow,, Trulia, etc.) at valuing a property? For this article, I will focus on Zillow, since it is the largest and most well-established of the online real estate valuation/information companies.

Should you use Zillow to value a property?   To evaluate the accuracy of Zillow, I pulled a Zillow report (along with other comparable data from a lender program we subscribe to that pulls directly from the assessors) for each residential property I inspected.  Since we are a direct privately funded lender and do not rely on outside appraisers, I have driven almost 50,000 miles in Colorado the last 18 months personally inspecting each property (both residential investment properties and commercial properties) before we make a loan.  I was curious to see how accurate Zillow was.  Could Zillow or other automated method be used to replace some physical inspections? I quickly noticed some very stark trends.

First Zillow (Trulia/, etc. use similar methodology) has a proprietary algorithm that estimates the present market value of a property.  Zillow calculates this by comparing the subject property with similar properties close by.  On paper, this sounds very plausible and theoretically should give an accurate assessment on value. How does this pan out in real life?

The good news is that Zillow is pretty accurate in certain circumstances.  Through my research, I have found that if a property is located in a very uniform high density area developed with similar houses on similar lots all of a similar age, Zillow is close on its valuation.  An example would be certain subdivisions in Highlands Ranch that were developed in the 80s and 90s in a master planned community where many of the houses look very like each other.  The valuation isn’t 100 percent since one house could have a brand-new kitchen and new windows, etc. versus a neighbor that has no updates, but it is typically in the range of the correct value (within 15 percent this could be high or low).

The reality check: Zillow, or any other automated method, should not be relied on in most instances. I’ve seen Zillow off on their value by over 50 percent (higher or lower than market value). I’ll provide two examples (I could list hundreds).  First, I recently inspected a property in the foothills of Denver. I looked at the subject property that I was going to lend on as well as a number of comparables.  One of the properties I looked at for a comparable had a Zillow value of $315,000.  When I drove by the comparable, it looked about the same size as the property that I was inspecting that was selling for $190,000.

I was curious why a house of similar construction that looked about the same size as the subject would have a Zillow value of 50 percent more.  From the interior pictures I saw online, they both looked very comparable.  When I looked at the square footage on Zillow the house down the street ($315,000 house) showed nearly 1,500 more square feet in Zillow.  I was intrigued, since the two houses looked virtually identical.  I checked with the county to confirm the square footage.  The houses were within 100 square feet of each other, but Zillow had counted the entire basement (both houses had finished basements) as above grade square footage.  What does this mean?  According to Zillow, the $315,000 house was 30 percent larger than the house down the street selling for $200,000.  In reality, the houses were almost identical, and Zillow was considerably over-valuing the $315,000 home.

The second example was in a nice area of Colorado Springs.  The Zillow value of the subject was $160,000 and the borrowers needed only $90,000.  On paper, everything added up, so I set up an inspection.

Unfortunately, when I got there, I discovered the house was not worth even close to the Zillow value.  The house had an odd corner lot surrounded by houses that were not very nice; one block up the neighborhood was totally different and considerably nicer.  After looking at the comparables (ones that I found that were similar location, build quality), the property was worth only $100,000 (I had found five comparable sales in the last 120 days very close to the subject that led to the much lower value).  In this case, Zillow was off by nearly 60 percent!

What does all this mean? Should you bet your wallet on Zillow or other automated valuation?  In certain very uniform neighborhoods, an automated valuation should be pretty close (not 100 percent, since one house could be updated).  Unfortunately, any properties outside of this uniform box would be ill-advised to rely on an automated valuation.  So what is the secret to accurately value a property?

The answer is simple: DO NOT RELY SOLELY ON ZILLOW!  Get out of your car and walk the neighborhood and physically look at the comparables yourself to see which ones are most closely like the subject property: size (verify with assessor to make sure comparing above grade), updates, view, neighborhood, lot, location, curb appeal.  This simple step will ensure you are accurately comparing apples to apples and not just blindly trusting your wallet to Zillow or other automated search to make the best decision.