00:00:01
[Music]
00:00:08
[Music]
00:00:20
Uh Susan Spencer, thank you so much for
00:00:23
joining us today. Great pleasure. Thank
00:00:24
you. So Spencer, I wonder if I could
00:00:26
start with you. U we are starting to
00:00:28
hear that the housing market is
00:00:30
recovering. Uh based on the data you see
00:00:33
at Zillow. uh what do you see and um how
00:00:37
even is the recovery across different
00:00:38
regions? Well um housing is back. So in
00:00:41
2012 home values were up 6%
00:00:44
year-over-year and in 2013 Zilla is
00:00:47
forecasting 3% appreciation
00:00:48
year-over-year which is a little bit on
00:00:49
the conservative side as compared with
00:00:51
other economists. But we've had 16
00:00:52
months in a row of appreciating home
00:00:54
values which is great. Um there's very
00:00:57
tight inventory across the country and
00:00:58
one of the reasons for that is so many
00:01:00
homeowners still have negative equity
00:01:02
and so it's difficult for them to sell.
00:01:04
So there's not that much inventory and
00:01:05
that's because of low mortgage rates.
00:01:07
That's creating bidding wars in a lot of
00:01:08
areas. And so we're seeing some parts of
00:01:10
the country appreciating at
00:01:12
unsustainably high rates. Phoenix is
00:01:14
over 15% year-over-year. South Florida
00:01:16
is appreciating very quickly. Northern
00:01:17
California is appreciating very quickly.
00:01:19
And that's a little worrisome. I think
00:01:20
those those rates of appreciation are
00:01:21
unsustainable. Um but overall, this is a
00:01:25
much much better housing market than
00:01:27
really it has been in the last several
00:01:29
years. What do you think, Susan? Well, I
00:01:31
agree. And I think that the months of
00:01:33
appreciation, 16 months and and ongoing
00:01:37
restores confidence and the negatives in
00:01:39
the market have been a lack of
00:01:41
confidence, an unwillingness to put your
00:01:43
money into housing when it's been such a
00:01:47
threatening position to have been most
00:01:49
people know people who've lost $40,000.
00:01:52
And but this is a different world and
00:01:54
this is a world where housing price
00:01:55
rises are in fact sustainable. They will
00:01:58
continue. I also concur and a 3% fact
00:02:01
I'm forecasting 5% price rises this
00:02:04
coming year. That's interesting. Well,
00:02:06
uh let let's go back Spencer to to uh
00:02:09
you know before Zillow was launched uh
00:02:11
you founded Hotwire which you sold to
00:02:14
Expedia and then you also worked for
00:02:16
Expedia as vice president of lodging.
00:02:19
How did your background in technology
00:02:21
and marketing uh inform the the way
00:02:25
Zillow was shaped? Um when we sold
00:02:28
Hotwire to Expedia, uh I joined and my
00:02:32
management team from Hotwire joined the
00:02:33
folks that had created Expedia within
00:02:35
Microsoft about 10 years earlier. And
00:02:37
the whole um thesis behind creating
00:02:39
Expedia was to provide information
00:02:41
transparency to consumers. Prior to
00:02:43
Expedia, as you probably recall, you'd
00:02:45
have to speak on the phone to a travel
00:02:47
agent and you could hear them type, you
00:02:50
know, over the phone and all you'd want
00:02:52
to do is jump through the phone and see
00:02:53
for yourself what secret database is
00:02:55
that travel agent looking at that I
00:02:57
don't have access to. And that's what
00:02:58
Expedia did. It turned around the screen
00:02:59
so that you as a consumer could have
00:03:01
access to the secret database. When we
00:03:03
were at Expedia, we started looking at
00:03:04
other industries that hadn't been
00:03:06
revolutionized by the internet yet. And
00:03:07
it was obvious to us that the real
00:03:08
estate industry was among them that
00:03:10
there were these secret databases that
00:03:12
only professionals had access to and
00:03:14
there was all this great data locked up
00:03:15
in these these databases but also in
00:03:17
county courouses. And so at Zillow we
00:03:19
aim to unlock that information and make
00:03:21
it readily available to consumers. The
00:03:22
difference though is in the real estate
00:03:24
space unlike the travel space there will
00:03:26
always be an intermediary uh somebody in
00:03:28
the transaction and that's that's fine
00:03:30
by us. We actually think that real
00:03:32
estate transactions are so infrequent,
00:03:34
so expensive, so highly emotional, so
00:03:35
complex that there will always and
00:03:36
should always be a professional in the
00:03:38
transaction. So, we have a very
00:03:39
different business model at Zillow. We
00:03:41
sell advertising and software tools to
00:03:43
professionals. Um, that's how we make
00:03:45
money rather than being in the
00:03:46
transaction, which is how Expedia makes
00:03:48
money by being an online travel agent.
00:03:50
Uh, so what do you think of the business
00:03:53
model that uh Spencer just described?
00:03:55
Zillow is a game changer. I can't talk
00:03:57
about the uh profitability of the of the
00:04:00
company but the what it has brought to
00:04:03
the consumer to the home owner in terms
00:04:05
of valuation of their home is extremely
00:04:08
important especially in these years of
00:04:10
extraordinary volatility when people
00:04:12
want to know what is their home worth
00:04:14
has it in fact fallen by 50% which homes
00:04:17
in many parts of the country have uh do
00:04:20
you have to actually go to someone to an
00:04:22
expert get several different views no
00:04:24
this is getting the power of all all the
00:04:26
data bringing it together in very
00:04:28
sophisticated analytic models and it's
00:04:30
there for you to test and by the way if
00:04:32
it appears to be off in some ways there
00:04:34
is now a way of going back. So this is
00:04:37
this is really extraordinarily
00:04:38
important. Um you need to know the value
00:04:41
of your home today for many reasons. Are
00:04:43
you going to sell as compared to your
00:04:46
mortgage amount? Uh this democratizes
00:04:48
data and it makes the market more
00:04:50
transparent and more efficient. All of
00:04:53
which is very important to this market
00:04:55
which has seen its troubles. This helps
00:04:58
bring liquidity back. That that's very
00:05:00
interesting Susan that you said that
00:05:02
because uh one of the things that I
00:05:04
found most interesting on Zillow was the
00:05:06
was the Z estimate. Yes. Uh and I
00:05:09
promptly of course went and looked up
00:05:11
the value of my own house. Uh now what I
00:05:14
was very curious is how do you figure
00:05:16
out what the house is worth? How does
00:05:18
your algorithm work? Well, the the short
00:05:20
answer is the the data base look the the
00:05:24
the software looks at the property
00:05:26
attributes of your home and compares it
00:05:27
with like homes. Um the much longer
00:05:30
answer is at each census tract level.
00:05:33
Census track is basically a couple
00:05:34
neighborhood blocks. We have um a meta
00:05:37
model that sits on top of many other
00:05:38
subm models. And so for example, a
00:05:40
submodel might look at tax assessment.
00:05:43
Another might look solely at uh distance
00:05:44
to an arterial. Another might look at
00:05:46
dollars per square foot. Uh, another
00:05:48
might look at um just regressions
00:05:50
against number of bedrooms for example.
00:05:52
Um, and the meta model at the census
00:05:55
track level tries different weightings
00:05:57
of each of the subm models going back 15
00:06:00
years and is constantly improving itself
00:06:03
by changing the allocations the relative
00:06:04
weighting of the different submodels to
00:06:06
figure out what's the right waiting
00:06:08
across each submodel to be as accurate
00:06:10
as possible. So as time passes as more
00:06:12
and more data comes into the system it
00:06:14
becomes more accurate over time. And
00:06:16
point of fact, about a third of homes in
00:06:18
the US now have had their property
00:06:19
attributes changed on Zillow by
00:06:21
homeowners or their agents. So much as
00:06:24
Wikipedia is a living database of all
00:06:26
the world's knowledge, Zillow has become
00:06:27
a living database of all the property
00:06:29
parcel information in the US. And that
00:06:31
allows this estimate and other data on
00:06:32
Zilla to become more accurate over time.
00:06:34
And does it apply only to existing home
00:06:36
sales or also new homes? We also come up
00:06:38
with estimates with new homes based on
00:06:39
the property records that either come to
00:06:41
us from the county or come to us from
00:06:42
the listings feeds from new constru from
00:06:45
home builders who send us feeds. So um
00:06:47
also when we get new information about a
00:06:49
home it this estimate will change. So
00:06:52
for example if county records have three
00:06:53
bedrooms, three baths, 3,000 ft² and
00:06:55
then we become aware either because the
00:06:57
listing information says so or because a
00:06:59
homeowner tells us that the property has
00:07:01
four bedrooms and four baths and 4,000
00:07:03
feet, we recalculate this estimate at
00:07:05
that point. So, what is your view of the
00:07:07
process that uh uh you Spencer just
00:07:10
described is how sound from a pricing
00:07:12
standpoint is it? It's a great process.
00:07:15
It's using what we call hyonic analysis
00:07:18
that is taking the attributes of the
00:07:20
property and waiting them to efficiently
00:07:23
use them statistically to come out with
00:07:24
the best estimate. There are other ways
00:07:26
of doing this. You could do repeat sales
00:07:28
and that avoids the structure of the
00:07:30
home and the differences in the
00:07:31
structure of the home. What's
00:07:32
interesting about this is it
00:07:34
incorporates not just the value changes
00:07:36
in the neighborhood the or the nearest
00:07:39
neighbors but also the attributes of
00:07:40
your own home and that's an advantage.
00:07:44
Now, uh question for both of you is that
00:07:46
I've heard some criticism especially
00:07:48
from realtors uh that it's um that that
00:07:52
uh this estimate or or the pricing model
00:07:54
that Zillow uses may have some flaws. Uh
00:07:57
and because there are some factors that
00:07:59
just can't be captured and automated. Uh
00:08:03
do you agree with that criticism? And if
00:08:05
so, what do you what's your response? I
00:08:08
I do agree with that. I mean, it's
00:08:09
computerenerated on every home in the
00:08:11
country three times a week. So, uh, it
00:08:14
we Zillow, no employee of Zillow has
00:08:16
ever been in your home. Uh, so, um, for
00:08:19
a computerenerated valuation, we believe
00:08:21
it's quite accurate and it's a good
00:08:22
starting point. We call it a Zestimate,
00:08:23
not as appraisal. It's meant to be a
00:08:25
starting point. And if you want to
00:08:28
figure out a more accurate determination
00:08:29
of your home's value, then we encourage
00:08:31
you to talk to a real estate agent or an
00:08:33
appraiser who will come to your home and
00:08:34
refine the valuation. Absolutely. If you
00:08:37
have something special like um uh a um a
00:08:42
pool for koi fish, let's say, they'll
00:08:44
never know and you'll know. It's very
00:08:46
important to you and might be important
00:08:47
to some buyer. Uh so that's why it's a
00:08:51
basic estimate and there may be some
00:08:53
negatives as well that may not be
00:08:55
incorporated. That's right. So u if you
00:08:58
if you take some of the other companies
00:09:01
that are active in the same space as
00:09:03
Zillow uh say realtor.com or Trullia uh
00:09:08
everyone has access to the same basic
00:09:10
data which is from the multiple listing
00:09:12
system or the MLS uh what is your
00:09:15
strategy to contain to to create a
00:09:18
sustainable advantage uh for Zillow
00:09:21
based on the fact that you all start
00:09:23
with the same you're you're right that
00:09:26
at the core it's there's fundamentally
00:09:28
commoditized listings information which
00:09:30
is why we don't focus exclusively on
00:09:32
that. So if you think of um the entire
00:09:34
housing stock of the United States 110
00:09:36
million homes about 3% of them are for
00:09:39
sale in the MLS at any point in time
00:09:41
everybody has that information um and
00:09:43
that's on tens of thousands of other
00:09:44
websites. So we try to innovate on top
00:09:47
of that. For example, by having
00:09:48
estimates, by showing historicals
00:09:50
estimates, by showing sale data going
00:09:52
back 15 years, tax information, making
00:09:54
predictions of what's going to happen to
00:09:55
that home's value going forward. But
00:09:57
fundamentally, the the price, the
00:09:59
photos, the square footage, that's on
00:10:01
every website. And so that's not enough.
00:10:03
What we then do is we have information
00:10:05
on hundreds of thousands of make move
00:10:07
listings when an owner posts a dream
00:10:08
price for their home, new construction
00:10:10
listings, for sale by owner listings.
00:10:11
We're the only website with over a
00:10:12
million foreclosure and pre-forclosure
00:10:14
listings um posted for free for
00:10:16
consumers. And then of course we have
00:10:18
market context. We have information on
00:10:19
every home in the country. And then we
00:10:21
have ratings and reviews, 250,000
00:10:23
reviews of real estate agents that
00:10:24
consumers have left on Zillow. So we do
00:10:26
our best to innovate not on that
00:10:28
commoditized information but on all the
00:10:30
other stuff which we think is what
00:10:32
differentiates us from competitors. Do
00:10:34
you think this is enough to create a
00:10:36
sustainable competitive advantage for
00:10:37
Zillow or can others catch up? The key
00:10:40
is the
00:10:42
algorithms which one can improve over
00:10:44
time and which become self-improving
00:10:46
with additional data and uh adding data
00:10:51
particularly individual input that
00:10:53
others don't have or can't organize and
00:10:56
some of this data you can't just
00:10:57
replicate back and forth individual uh
00:11:00
reestimates of the property I think is a
00:11:03
sustainable model. There's also of
00:11:05
course the name Zillow and uh there
00:11:08
aren't there isn't going to be room
00:11:09
here. I don't think for many firms with
00:11:11
many names, people are going to
00:11:13
gravitate to whoever has the most
00:11:14
information and then that firm will have
00:11:16
the most information going forward. The
00:11:19
the platform shift from the desktop to
00:11:21
mobile has also been a great boon to our
00:11:24
business because um we're a software
00:11:26
company and we can devote massive
00:11:28
amounts of resources to create software
00:11:30
especially on mobile where it's very
00:11:32
complex to write software across
00:11:34
multiple mobile platforms. So for
00:11:35
example, we went public about a year and
00:11:36
a half ago. When we went public, 20
00:11:39
homes were viewed every second on Zillow
00:11:41
on mobile. A year and a half later, it's
00:11:42
75 homes viewed every second. A year and
00:11:45
a half ago, we had five mobile apps.
00:11:46
Today, we have 24. Just to give you a
00:11:48
sense of just how quickly our our usage
00:11:50
is shifting to mobile. More homes are
00:11:51
now viewed on Zillow on mobile than on
00:11:53
the desktop. So, as more and more usage
00:11:56
shifts towards mobile, it's difficult
00:11:57
for individual brokerage websites at at
00:11:59
a local level to compete because it
00:12:01
requires so much software to create
00:12:03
great mobile app great mobile
00:12:05
experiences and great mobile apps. Um,
00:12:06
and it's difficult for even other
00:12:08
national players to compete. So, we we
00:12:10
definitely benefit from that. Um, I I
00:12:12
just saw some interesting data from
00:12:13
Domino's Pizza, believe it or not. Um,
00:12:16
at their investor day, they put up a
00:12:17
slide that showed national uh market
00:12:20
share of Domino's and Pizza Hut over the
00:12:22
last 10 years versus local pizza places.
00:12:24
And it was basically flat between
00:12:25
national and local for the last 10 plus
00:12:27
years. And then about two years ago,
00:12:28
something interesting happened. The
00:12:30
national guys gained a couple points of
00:12:32
share really out of nowhere versus the
00:12:33
local pizza um pizza stores. And the
00:12:36
reason is the smartphone because all of
00:12:37
a sudden people now are ordering pizzas
00:12:39
from Domino's and Pizza Hut's mobile
00:12:41
apps, but the local pizza guy can't
00:12:42
really compete with that. And you know,
00:12:44
a big shareholder of ours showed that to
00:12:46
me and said, you know, I'd bet you never
00:12:47
thought you'd have something in common
00:12:48
with Domino's Pizza, but the way you're
00:12:50
dominating on mobile is very similar.
00:12:52
And you're seeing this in other
00:12:53
verticals as well. That's really
00:12:55
fascinating. Um, one more question about
00:12:58
how do you ensure the accuracy of your
00:13:00
data? Uh I was talking to some real
00:13:03
estate developers and they said that for
00:13:05
some markets your data is absolutely
00:13:08
spot-on accurate and other markets maybe
00:13:11
not so much. Do you think that the
00:13:13
quality of your data is sort of uneven
00:13:14
and if so what are you doing to address
00:13:16
that issue? Well, this is an
00:13:17
industrywide problem which as the
00:13:19
largest player in the industry we um we
00:13:21
definitely have and it's something that
00:13:23
we focus on and that we apply a lot of
00:13:25
resources to. Um, the data on Zillow
00:13:27
comes partly from county records and
00:13:29
there it can be corrected by owners or
00:13:31
agents, but it also comes to us by
00:13:32
agents. And so, for example, when a
00:13:34
listing agent posts a home for sale and
00:13:37
if the home sells, he doesn't take it
00:13:38
down, well, we have a stale listing. So,
00:13:40
we've deployed a lot of software, a lot
00:13:42
of business expertise, and a lot of
00:13:45
money towards improving this problem.
00:13:47
And we've come a long way, but it's an
00:13:48
industrywide problem, which
00:13:49
unfortunately I think will will continue
00:13:51
to be something that we that we work on
00:13:53
for for years. It is a problem. It's a
00:13:56
problem in part because our sales
00:13:58
records are are are county level. Uh
00:14:01
it's actually in some ways stunning that
00:14:03
we as a country do not collect the data
00:14:05
and don't verify the data. Uh and some
00:14:08
states have really poor and some cities
00:14:10
have really poor collection
00:14:12
methodologies. This is a concern on a
00:14:14
wider scale. Uh for example, Fanny
00:14:17
Freddy uh which are national players now
00:14:21
taxpayer owned. uh they have a price
00:14:23
index which uh arguably is far behind
00:14:27
the Zillow price index. So this tells us
00:14:29
we should be putting more public
00:14:30
resources I think in this direction as
00:14:32
well. Uh you're also doing some
00:14:35
interesting work in the area of uh
00:14:37
rental housing. Could you tell us a
00:14:39
little bit about what are some of the
00:14:40
challenges involved in that? Sure. So
00:14:42
rentals is a very big area of investment
00:14:44
for us. We bought three companies in the
00:14:45
last 18 months uh there for over $60
00:14:47
million. We have about 100 people out of
00:14:50
600 people at the company focused
00:14:51
exclusively on the rentals opportunity.
00:14:54
And the reason that we're so excited
00:14:55
about it is it's a $5 billion market.
00:14:57
Property managers and landlords spend $5
00:14:59
billion advertising their listings to
00:15:01
consumers. And the structure of the
00:15:03
industry is quite different than the
00:15:04
structure of the forale industry.
00:15:06
Notably, it's it's disorganized. There
00:15:08
are no in most markets outside of
00:15:10
Manhattan, for example, it they're not
00:15:12
broker controlled markets. So property
00:15:14
managers and real estate developers
00:15:15
market their listings directly on sort
00:15:18
of a many to many approach rather than
00:15:20
going through a centralized MLS or
00:15:21
centralized brokerage system. And that
00:15:24
allows us, we hope, the opportunity to
00:15:27
create the marketplace to basically
00:15:28
superset Craigslist. So to give you some
00:15:31
numbers, there are 4 million rental
00:15:33
units available for rent in the US at
00:15:34
any point in time. Craigslist has about
00:15:36
750,000 of them. We have about 600,000
00:15:39
and a year ago we had 300,000. Most
00:15:41
other sites have about 100,000. So the
00:15:44
strategy is to give away software, give
00:15:46
away property management software to the
00:15:48
industry for free. And one of the
00:15:50
acquisitions that I mentioned is a
00:15:51
company that that built that type of
00:15:52
software. Give it away for free in order
00:15:54
to out Craigslist. Craigslist to end up
00:15:56
with many more listings than Craigslist
00:15:58
and then hopefully we'll become the
00:15:59
marketplace and then we can monetize it.
00:16:01
So today we don't monetize rentals at
00:16:03
all or hardly at all. Um, and we're
00:16:05
delaying that gratification until we are
00:16:08
the marketplace and then we can focus on
00:16:10
monetizing it. What do you think of uh
00:16:13
Zillow's strategy, Susan? I think the
00:16:15
rental market is growing in the US. So,
00:16:17
it's important to have that coverage.
00:16:18
It's less organized as we said. So,
00:16:20
that's a a real strategic move. But even
00:16:23
more so going forward, there's a lot of
00:16:26
information in the comparisons of rents
00:16:27
to values. And to be able to get that
00:16:30
accurately, that too will be a gamecher.
00:16:33
So let me ask end with one last
00:16:35
question. So in so many ways Zillow has
00:16:37
been highly successful and to take just
00:16:40
one metric uh since your IPO in 2011
00:16:45
your stock stock price has gone up to
00:16:47
$55 and I just saw a report from Goldman
00:16:50
Sachs setting a $60 target for it. Uh
00:16:55
but regardless of what happens to the
00:16:57
stock, what is your definition of
00:17:00
success for Zillow? Um my definition of
00:17:03
success is about creating a ever
00:17:06
everlasting and enduring brand that my
00:17:09
children or maybe someday grandchildren
00:17:10
will know. Um there are very few
00:17:13
technology companies that have created
00:17:15
brands that endure for 10, 20, 30 years.
00:17:19
Um you know Microsoft is really one of
00:17:21
the only ones and I suppose Apple that
00:17:23
has endured for a long long period of
00:17:25
time. And to me it feels like we're just
00:17:28
getting started. We have less than 1%
00:17:30
revenue share across the 35 billion
00:17:33
dollars in advertising in our space. Um,
00:17:35
and we have a relatively little known
00:17:38
brand relative to the size of the
00:17:40
opportunity. Only about 12% of
00:17:41
Americans, believe it or not, have heard
00:17:43
of Zillow. Uh, if you ask Americans, and
00:17:45
yet we're the leading real estate
00:17:46
website. If you look at other
00:17:48
categories, you know, name a finance
00:17:50
website, 90% of people will say Erade or
00:17:52
CNBC.com. Name a travel website, 90% of
00:17:54
people will say Expedia. Name a
00:17:55
healthcare website, 90% of people say
00:17:57
WebMD. name a real estate website, 12%
00:17:59
will say Zillow. That to me, you know,
00:18:02
you might think as the CEO of Zillow,
00:18:04
that would be depressing, but actually I
00:18:06
get very excited about that opportunity.
00:18:08
You know, think of all the potential.
00:18:09
Um, and to me, that's success is closing
00:18:11
that gap. And if we do that, then a lot
00:18:13
of other great things will follow.
00:18:14
Revenue, profit, market cap, etc. Great.
00:18:17
Susan, you get the last words.
00:18:20
Well, market information in real estate
00:18:22
is almost contradictory in in very
00:18:25
sense, but in fact, this is this
00:18:28
contradiction is being undone and um
00:18:31
we're we're bringing transparency and
00:18:33
Zillow is one of the ways that's
00:18:34
happening. So, I congratulate the firm.
00:18:36
Good luck and thank you to both of you.
00:18:38
Really glad you joined knowledge at
00:18:40
Wharton today. Thank you.
00:18:45
[Music]
