via Data Coverage and Zestimate Accuracy – Zillow.

Oy!  Good information is hard to come by.  That’s why zillow holds so much promise.  Alas, Zillow comes up very short.

I have a client that would like to refinance.  The current rate is 6.375%, the rate that we want is 5.25%, which works beautifully today.  The only remaining question is, will the property appraise?

Let’s turn to zillow to find out.  We type in the address an zillow gives us their zestimate of $332,000 down from the $337,000 that the property was purchased for, but still high enough for the deal to work.

Just to make sure we look at the properties zillow used as an input.  This is where things get messy.  Appraisers are limited to using recent sales.  The sales zillow includes are over a year old.

Appraisers must also use the same subdivision.   Zillow gives us properties well outside the subdivision.

Now the appraiser has discretion, not much these days, to step outside these bounds, but zillow turns the discretion into a rule.  The zestimate is, as a result, useless.

Zillow is aware of their limitations.  They publish a chart of how good their zestimates are.   Zillows chart shows us that within the Denver region the zillow zestimate is within 5% of the eventual sale price 30% of the time.

What does that mean?  It means that there is a 30% chance that this property that needs to appraise for no less than $337,000 is somewhere between $320,200 and $353,800.   Five-percent accuracy isn’t very accurate.  Five-percent accuracy only 30% of the time is beyond hand grenade range.

Zillow has such promise.  I was at a convention six months prior to zillow’s launch where Rich Barton coyly spoke of his creation.   He was onto something, but several years later they are still not very close.

They do get traffic though, and are probably profitable as  a result.  At this juncture I’m wondering if zillow has simply decided that, that is good enough.

Don’t even get me started on their mortgage efforts, which is still a bait-and-switch frenzy.