Using Zillow To Efficiently Estimate The Value Of A Client's Personal Residence

Posted by Michael Kitces on Tuesday, June 12th, 11:01 am, 2012 in General Planning

While most aspects of the financial planning update are getting easier, especially as account aggregation software becomes more prevalent - allowing the planner to automatically get regular updates from all client financial accounts, including those not under the planner's direct investment purview - the value of real estate continues to be a sticking point for many planning firms. How do you update the client's net worth statement without slowing the process down by waiting for the client to provide an estimate? Will the client's estimate really even be accurate, anyway? And the process can be even more problematic for firms that set fees based not on assets under management but net worth. How do you get a fair estimate of the value of property that relies on the client when the client's fee is impacted by that estimate? In an increasingly technology-driven world, there's now an answer: with real estate valuation services on the web, like Zillow.

The inspiration for today's blog post was a recent conversation I had with another planner about the challenges of trying to update a client's financial plan, especially regarding the difficulties of updating values for the client's personal real estate. "We just use Zillow," said the planner, "it's made the process a whole lot easier!"

What Is Zillow?

Zillow is a website that provides information and tools around the real estate market, including most notably their Zestimate service, which seeks to estimate the value of any piece of real estate using Zillow's proprietary formula. 

The Zestimate is based initially on publicly available data, including the dimensions of your home and lot as documented in public tax records, the tax assessments themselves, and comparable sales in your neighborhood (as well as prior sales of the home itself). Zillow uses all of this information in an attempt to extrapolate the value of land, square footage, number of rooms/bedrooms/bathrooms, etc., to determine an overall value, not unlike how a real estate agent might evaluate the home and look at comparable sales to estimate a price.

Of course, it's difficult to estimate a price for some properties, especially if there are not a lot of comparable sales in the area (and/or the sales are foreclosures, which the Zestimate ignores), so in reality Zillow provides not only a Zestimate price but also a Zestimate range (which represents a 70% confidence interval around the Zestimated price). A wider range indicates more uncertainty either due to more volatile comparable sales or a lack of enough comparable sales for a good baseline; a narrower range indicates that Zillow had a strong base of information upon which to provide the estimate.

In addition to the Zestimate provided initially, the client can create an account on Zillow and "claim" his/her property to correct or provide further supplemental information. For instance, if the home has had significant interior upgrades that weren't reflected in public records, Zillow won't know until the user supplements the information. Similarly, a remodeling/expansion that is not yet captured in public records might not be used in the Zestimate but could be updated by the client/user.

Zestimate Accuracy

So how good are the Zestimates that Zillow provides? Zillow does provide information on their data coverage and accuracy statistics (by comparing Zestimate prices with actual transactions that occurred for those properties). Notably, for some states where there are relatively few sales and/or homes that have Zestimates, the data coverage is limited; in addition, some states provide far less information in their public records. As a result, some states like Alaska, Kansas, Idaho, and Maine have almost no coverage at all, while other states may have Zestimate coverage in metropolitan areas but not in more rural counties.

Where Zillow does have coverage, though, the data suggests that Zillow provides at least "reasonable" estimates; in most areas where Zestimates are provided at all, the median error is between about 6% and 10%. And of course, this is across all sales of properties that had Zestimates; if the property information has been updated by the owner to account for details not evident in the public records, the error rate ostensibly would be even lower.

In fact, in situations where the client firmly believes that his/her property is worth more or less than Zillow, it's not always clear whether the client is right and the impersonal computer is wrong, or if Zillow's more data-intensive analysis is correct and the client is simply providing a poor estimate! 

Using Zillow Zestimates in Financial Plans

Notably, Zillow still does not carry the weight of a formal appraisal. It cannot be used as a basis for substantiating the value of a home for a mortgage. And even Zillow encourages users to have a real estate agent do a local comparative market analysis and/or get an appraisal from a professional appraiser before setting an actual price to sell a house or make an offer to buy one.

Nonetheless, as a general estimate of the value of a home, arguably a Zestimate is a pretty good starting point; to say the least, Zillow likely takes into account more information about the activity of home sales in the region than the client, who may simply guess at a value based on the assessed value for tax purposes (which isn't always representative of a fair market value price) or a single property sale up the street (which may or may not have truly been a comparable property). 

Clients who think the Zestimate for their property may be too high or too low can always log into the site, "claim" their house, and supplement with additional information that can make the Zestimate even more accurate. And at a minimum, a conversation about why the Zestimate seems higher or lower than the client expected may itself be a fruitful conversation about the value of the real estate, potential planning issues or concerns, and the client's longer term intentions for the property.

But the bottom line is that services like Zillow (or some of their competitors, including Trulia or the National Association of Realtors' own Realtor.com) seem like a good way to get at least an initial estimate of the client's real estate, especially in metropolitan areas and/or other regions where Zillow's Zestimates have a good track record and a reasonably narrow range. At the least, it's probably a more data-based and rigorous process than simply asking the client for a guesstimate, and may even lead to a more fruitful conversation.

So try it out the next time you're updating a plan; or better yet, pull up Zillow on the screen in your conference room during the client review, type in the client's address, and use it as an opportunity to frame a discussion around the client's real estate and long-term plans.

So what do you think? Have you ever used a site like Zillow to estimate home values for a client's financial plan? Do you find it useful, or too inaccurate to be relevant? Is it more useful for some clients and/or in some areas than others? Is it useful as a conversation tool as well?


  • Bill Winterberg

    Better yet, Zillow has an API, so financial planning software developers can automatically aggregate Zillow value information into their program.

    Give the advisor the choice to use Zillow data or enter a value manually. That should save a few steps each time the advisor refreshes a client's plan.

    For some reason I think FinanceLogix may be doing some like this today. It might be built in to their RetireLogix app. Don't quote me on that!

    • http://www.zillow.com sara bonert

      Great article Michael. And good point about the API Bill. Thought I would include a link to it, to help show people the data which is available. http://www.zillow.com/howto/api/APIOverview.htm
      Another example of someone using the data like you illustrate is mint.com.
      - Sara from Zillow

  • Jean Fullerton

    I have used Zillow estimates in net worth statements when assessments are clearly outdated, and also used them to start discussions with clients who have an outdated price stuck in their head.
    However,I question how accurate the estimate can be when I assume that Zillow does not account for significant items such as a view, condition of the home, if its in a flood plain, and energy star rating.
    I just had an appraisal done on my own home. It came back as $375K. Zillow says $405K, and Trulia says $489K. Big differences.

    • Michael Kitces

      Jean,
      Information like location, in a flood plain, etc., is publicly available data (and would also be recognized in sales of comparable homes also in that location and in the flood plain).

      While condition and other details aren't public domain information, they CAN be updated on Zillow directly - which of course produces a more accurate estimate. But to say Zillow is bad because it's price is inaccurate because we didn't give it updated data seems odd to me; if that's a criticism, we can update the data, and get a better estimate!

      As for the difference between the Zillow and Trulia estimates... sounds like Zillow is the right platform after all? :)
      - Michael

  • http://www.mayflowercapital.com Don Martin, CFP

    Good article, however I feel if a client is paying thousands for a financial plan he ought to consider paying for a manually done appraisal. Automated appraisals fail to account for very important things like school districts (extremely important), views, street noise, proximity to unpleasant things. I have seen them in error by a factor of 20%. An alternative to paying for an appraisal would be to use Realtor's websites lowest asking prices for comparable sales that are in escrow but have not yet closed escrow after having the client carefully check that the comps were in the same school district, similar views, etc. But clients are usually not motivated to really assist in the research needed for a plan so I would not expect them to do a fair job appraising their own home.

    • Michael Kitces

      Don,
      You assume that "automated" appraisals can't take into account many of these factors, but I'm not certain that's correct.

      Zillow already has information on school districts (it's public record), and comparable sales for that school district (again, public record). Which means in practice, Zillow already does EXACTLY what your real estate appraiser does - adjust the price of your home based on the way prices go up or down depending on your school district (compared to other school districts).

      Similarly, location relative to major highways (and their noise), proximity to other pleasant or unpleasant things, etc., is all available to be tracked in the data. Again, that's how the appraiser or real estate agent makes the same adjustments - take the same publicly available location information and publicly available sales data, and adjust accordingly.

      In both cases - whether Zillow, or a real estate agent - areas with fewer comparable sales at all are harder to price accurately. We may lambast Zillow because it's prices aren't always "right" - but then again, when real estate agents list properties for 6-9 months and drop the price 1-5 times over that interval (which is incredibly common around here, even though it's a dense metropolitan area that should be "easier" to price), it's not clear to me whether the real estate agent prices are "better" or simply "more optimistic".

      But the bottom line is that the overwhelming number of factors you're citing here are still in the price data, whether it's Zillow or a realtor estimating a price using that data. That isn't to say that Zillow is perfect. But the point is, I'm not certain there's any reason to say they're systematically LESS accurate than having a subjective human being look at the same data and same information to make their own subjective decision. In fact, if there's anything we see in the research of other markets, it's that human beings have a tendency to factor information in an incomplete manner and make inaccurate estimates of price and value, as witnessed by the average client's stock market investing behavior.
      - Michael

  • Chuck Donalies

    Another search option is http://www.redfin.com. I've spoken with a few realtors that prefer to use this site over zillow. It's got some of the same problems Michael brings up in his post, but it's another tool you can use to help clients.

  • Eric Bruskin

    In my experience Zillow has rarely been helpful in the reasonably mature suburban area where many of our clients live. I've seen cases where:

    1. Zillow's value is way off, by factors like 5 - clearly data glitches.

    2. Zillow's value fluctuates wildly from month to month - perhaps affected by other new data coming into their database, then being corrected, then who knows what ...

    3. Zillow systematically undervalues by perhaps 20% a type of home I'm very familiar with, ignoring recent sales.

    4. A home I just looked at of the same type as mentioned in #3 (plenty of comparable sales), offered at $350K, with a Zillow range of $170K to $525K. What's up with that? That's not even useful for VERY long-range planning.

    Maybe I've just been unlucky, frequently falling into the 25% that's not within 20% of the sales price according to their statistics.

    Also, in my experience real estate databases are egregiously error-prone. This is over and above the issues raised by comments #2 and #3 above.

    I look at Zillow mainly for fun. I wouldn't base a client report on their values.

    • Michael Kitces

      Eric,
      Bear in mind some of the criticisms you're levying here - like data glitches - can be fixed by the client proactively claiming the profile and entering more accurate data. That is an entirely controllable and solvable problem.

      As for the point you make in #3 and #4, the implication here is there's not actually nearly as much data available as you suggest. Whether that's because you're ultimately overweighting relatively few data points, or because Zillow isn't pulling all the data correctly, is difficult to judge. As I noted in the blog post, Zillow is clear that some properties have more comparable sales than others - in fact, the reality that they provided a wide range is their own open acknowledgement that the available data for the property is limited. I find that a plus to the transparency of their estimates, not a minus, especially given how many appraisals I've seen that were horribly wrong because they used far too few comparable sales and failed to disclose the error range of their own estimates.
      - Michael

      • Eric Bruskin

        With all due respect, Michael, I am intimately familiar with the situation in #3 and #4. It is a large development with about 20 sales (not just listings) per year in recent (i.e. slow) years including the past 6 months. There should be plenty of data available.

        As you said, it's "difficult to judge". Since I can't trace down the reasons for their -50%/+100% uncertainty limits, I find it unusable and lacking credibility that I could explain to a client.

        • Michael Kitces

          Eric,
          Almost by definition, the width of their error ranges MEANS they lack sufficient data to make a more accurate estimate. If the reality is that there's a data reporting problem, then indeed Zillow isn't taking into account all the information. On the plus side, they're also announcing that in the first place.

          Certainly, there are parts of the country where Zillow acknowledges they have so little data, they provide NO estimate at all. And some states and counties have far better or worse reporting and documentation than others, so clearly Zillow won't be a perfectly uniform tool for all clients and situations.

          But again, all this is disclosed by Zillow. Personally, if I saw a 50%+ error range, it's Zilllow announcing it has insufficient data for a good estimate, and not use their number - because wide error ranges are MEANT to disclose the number isn't reliable (that's the whole point of including confidence intervals).

          On the other hand, for many clients and areas, the confidence interval is much narrower and the outcome should be more accurate - again as Zillow's own data shows.

          I guess the bottom line to me is that the whole point of Zillow's estimates AND confidence intervals is to literally show how accurate the number is. Which means they're actually putting the tools in our hands to both shows us WHAT the estimate is, AND whether it's reasonable to rely on it. Of course we shouldn't put undue reliance into a number that by Zillow's own acknowledgement is not to be relied upon.

          The good news is that they disclose enough of their data to convey whether they believe it's a reliable estimate in the first place!
          - Michael

  • Tammy Prouty

    We have found that homes are selling at or below assessed value, so to be conservative or right on, I just pull the values from the county assessors office. of course, this probably won't always be the case, but it is definitley a conservative estimate.

  • Michael Kitces

    Tammy,
    Out of curiosity, how do those estimates compared with Zillow, which is also looking at comparable home sales in the area?
    - Michael

Blog Updates by Email

Nerd’s Eye View Praise

@MichaelKitces Twitter


Out and About

Thursday, November 6th, 2014

Cutting Edge Tax Planning Developments & Opportunities Future of Financial Planning in the Digital Age @ ACP 2014 Conference

Friday, November 7th, 2014

Setting a Proper Asset Allocation Glidepath in Retirement @ Charles Schwab IMPACT 2014

Monday, November 10th, 2014

Expanding the Framework of Safe Withdrawal Rates Future of Financial Planning in the Digital Age @ PICPA 2014 PFP Conference

VIEW FULL SCHEDULE