I get the sentiment, but having spent the last six years rethinking customer experiences for companies from Disney to Starbucks, a better estimate sounds like they are optimizing one specific thing without taking a step back to look at the end-to-end customer experience.
Zillow (all of Zillow Group) would be better off putting themselves in the shoes of the end user. Netflix (which ran a similar algorithm challenge but never adopted the winning result) has taught us that our tastes are unique to us. They tell me that because I liked Star Wars and Pulp Fiction that I will like Let the Right One In. Highly personalized recommendations based on actual history of what I have watched and liked. So when I go to a place like Amazon and they say “people who like books by Po Bronson also like books by Martin Seligman” I tend to ignore them, even though Amazon has a lot more of my purchase history than Netflix – because I am not “other people” – I am me – and Netflix has reinforced that.
That’s why the Zestimate doesn’t serve the buyer as well as it could.
It’s too focused on old school two-dimensional comparable listings. What they can do, and should do, is collect more information about what people value, and based on that – help people find the best fit for them, based on their values and wallet.
As with Starbucks customers, I expect there are are Zillow Group users that have different visit frequency – some go every day, some go once a week, and some go once a month – and based on a variety of factors, usage might even vary from month-to-month.
In this day and age, a lot of users, not everyone – but a lot, will be comfortable saying where they live now – address or even ZIP code, where they work, the ages of their kids, and what parts of a house are most valuable to them. Some will even say what they make and what their current rent/mortgage is. And for high density and expensive places, like Seattle and San Francisco, people probably have a pretty specific tolerance for commute time. Zillow already captures a lot of these things in the filter on a fairly binary basis but doesn’t weigh the inputs. In my case, I would be willing to live on a golf course, but I don’t value it. Being a Seattle area person, I value living close to the salt water much more than being close to fresh water. A modern kitchen is a must for me, and I have a pretty high tolerance for a long commute.
With that information, Zillow could flip the information and make it less about comparable in the area and make it much more about the personal preferences of the customer. I use Zillow about once a month, and if it had all of the information, including my wallet, and it told me what I could afford that lined up with my values – I would probably find all sorts if places I would not be looking for today, but I would feel like Zillow was really paying attention to what matters to me – not the “other people” Amazon keeps telling me about.
So here’s what I would do:
- Start with the customer and customer segments,
- Break down their experience journeys by segment,
- Figure out what information they are willing to share, and with that,
- Weigh the filter so it’s much more personalized
- Notifications are now highly personalized and are no longer using the same comparables that real estate agents use
- Zestimates can then be broken down by values – still do standard comparables, price per square foot in a given area, but then show it based on values – so if you value living on a golf course (or salt water, or a modern kitchen, or whatever), a home might be a relative bargain with that feature – so for you – that house would be worth X but for me it would be worth a lot less. The Zestimate is still proprietary because you need to handle the weightings intelligently, but that makes Zillow Group much more valuable.
Because of the “voyeuristic” aspect of Zillow that founder Rich Barton has talked about – when people have entered their wallet/income information, people could also forecast based on getting married or getting a promotion or new job, what addresses would come up that hadn’t come up before. Anticipate the question “what can we afford after we get married, or after that promotion/new job” – that would be fun and people could model all sorts of scenarios no matter how realistic they are.
You could even go one step further and have the preferences of everyone in the household and merge them to see which houses are the best fit for the things everyone value (my wife would never live in a home without a bath tub – not something I value).
That’s my $.02 on the direction Zillow should take with the Zestimate. That would be a ZOW improvement.
And don’t miss the TEDx talk about how technology is going to improve your quality of life – this fits squarely in that.