I have seen from time to time people, sometimes even Realtors, include in a blog or forum post that they use the Zillow "Zestimate" as a baseline of value, "and then do their own analysis".

Very little makes me more frustrated than statements like these, for a number of reasons. First, because the Zillows/Trulias/Cyberhomes of the world do not exist to provide relevant and accurate valuation models, they exist solely to aggregate consumer information by providing wildly inaccurate estimates and calling them "data". This is at best, misleading, and at worst, detrimental to the industry.

However, it is often difficult to find a concrete example of Zillow's inadequacies to prove to their users how wrong the Zestimates are, again for a couple reasons. First, because valuations are subjective and even those trained in valuing property rarely come up with the exact same valuation for a particular property. Second, because the type of people that would think to use Zillow for a valuation are by definition not sophisticated enough to understand why they are wrong, and thus it is difficult to make logical points to them.

During my bi-weekly run along the beach, I usually take a glance at the current housing inventory, to see which (if any) of the oceanfront homes have come on the market, had their price reduced, etc. Yesterday, it hit me that there was a vacant lot on this path, and that I should look up the Zestimate on a vacant oceanfront lot, just to see how far off they were.

Here is the link:

http://www.zillow.com/homedetails/614-W-Oceanfront-Newport-Beach-CA-92661/25134676_zpid/#{scid=hdp-site-map-bubble-address}

What you'll notice is that while Zillow is smart enough to pinpoint the exact address, it is unable to determine whether there is actually a structure on the land. Thus, it has valued the vacant land as a 3/1/1500, and gives it a value of $1.075 M.

Is that the kind of AVM you want doing your ARV calculations for you?