00:00:02
We're here today with Ken Moon, a
00:00:03
Wharton professor of operations,
00:00:04
information, and decision, and he's here
00:00:06
to talk to us about some of his recent
00:00:07
research. Ken, thanks for being with us.
00:00:09
Oh, thank you so much. So, first of all,
00:00:10
could you give us kind of a short
00:00:11
summary of what you looked at, what
00:00:13
you're trying to find out? Um, so I I do
00:00:16
research in empirical operations. So,
00:00:18
um, basically that means two things.
00:00:20
One, I I work with data. Um, typically
00:00:22
sometimes in collaboration with uh
00:00:24
companies, hospitals, marketplaces. Um,
00:00:28
and second, I'm really trying to be uh
00:00:30
prescriptive about decision- making in
00:00:32
my research. Um, in this particular
00:00:35
project, uh, we work with an a retailer
00:00:37
that's active online. Um, it was a very,
00:00:41
uh, detailed customer level data set.
00:00:43
Uh, what's kind of interesting about it
00:00:44
is that you can track a single customer
00:00:47
um, both online and offline. So, for
00:00:48
instance, if the customer today were to
00:00:51
um go on their phone, look at a product,
00:00:54
go to their computer tomorrow, look at
00:00:55
that, and then walk into a store on
00:00:57
another day and actually buy it, we
00:00:59
would be able to track all of those
00:01:00
things. Um, so it really opened up a lot
00:01:03
of avenues to explore there. And um I
00:01:06
think uh one thing that's very
00:01:08
interesting about uh this particular
00:01:10
project was that we were able to look at
00:01:11
a very information-rich environment in a
00:01:14
way that I think um we see in our
00:01:16
everyday lives. So, not only can we
00:01:18
browse in that way and have a lot more
00:01:19
information, say we think a product that
00:01:22
we're interested in might drop in price,
00:01:24
we can check our smartphone. Um, it's
00:01:26
also the case now that uh these
00:01:28
companies can actually track all of this
00:01:31
information um at a very individual
00:01:33
level. Um, so for both sides, it's a
00:01:35
very information-rich environment. And
00:01:38
so it's very interesting um to think
00:01:40
about how does that affect uh the
00:01:42
decisions of say companies that are
00:01:44
active in this sort of space but also um
00:01:46
how does that affect outcomes for for
00:01:48
customers and consumers right I mean
00:01:50
price monitoring has really become kind
00:01:52
of a daily part at least of my life I
00:01:54
know so when you were looking at this
00:01:55
data set what did you find about price
00:01:58
monitoring
00:01:59
um I think one uh sort of broad takeaway
00:02:02
is that information seems to matter um
00:02:05
so you do have customers who are very
00:02:07
intensive in their monitoring. Um
00:02:10
they're typically the more price
00:02:11
sensitive customers actually, but also
00:02:13
their opportunity cost to be doing this
00:02:15
sort of monitoring is very low. So
00:02:17
they're going to be checking very often.
00:02:19
Um and your most price insensitive
00:02:21
customers actually uh they're not
00:02:23
checking very often. It's about every 20
00:02:25
days on average between uh visits. Um so
00:02:27
it's a very big difference in terms of
00:02:29
how these uh consumers are able to
00:02:32
access information even from this very
00:02:34
ubiquitous channel. Um, and it makes a
00:02:37
big difference in terms of outcomes as
00:02:39
well. Now, is there a clear indication
00:02:41
of how companies should be doing this?
00:02:44
Like how they should be playing with
00:02:45
price based on how someone's looking at
00:02:48
and monitoring price? Should we be doing
00:02:49
one thing or does it depend on the
00:02:51
customer or No, that's it's it's very
00:02:53
interesting. It was it was that's
00:02:54
exactly uh those are exactly some of the
00:02:56
issues that we wanted to explore in this
00:02:58
research. And um uh one of the
00:03:01
interesting things that we found is that
00:03:03
even in this very uh rich
00:03:05
informationally rich space um
00:03:09
sometimes very simple uh policies and
00:03:12
very simple decisions can be very
00:03:14
effective. You can capture most of the
00:03:15
value as a firm. Um so to give you two
00:03:18
examples um the retailer that we worked
00:03:20
with follows a very simple pricing
00:03:22
policy for each product. You're going to
00:03:24
start at a certain price, a list price,
00:03:26
and then at a certain point in time
00:03:28
during its season, you drop uh the price
00:03:31
down to its sale price, which is a very
00:03:33
predictable percentage of that initial
00:03:35
list price. And then finally, you move
00:03:36
to another predictable price, a
00:03:38
clearance price, where you're trying to
00:03:39
just get the products out of the off the
00:03:41
shelves. And um what's interesting there
00:03:44
is that the consumers understand what
00:03:47
prices they'll see. Um it's very
00:03:48
predictable. Um but all that the
00:03:52
retailer did is to make the timing of
00:03:54
those markdowns unpredictable.
00:03:57
And by doing something very simple like
00:03:59
that um it really uh uh uh exacerbated
00:04:04
theformational
00:04:06
uh asymmetry in terms of the cost of
00:04:08
monitoring. So those customers who were
00:04:10
price insensitive who um it was very
00:04:13
costly for them to be monitoring often
00:04:15
they were the ones who couldn't take
00:04:16
advantage of a markdown when it happened
00:04:19
and they understood that. So then um
00:04:21
they would buy earlier. So there's an
00:04:23
interesting aspect there where um this
00:04:25
sort of pricing has an allocative role.
00:04:27
You're deciding who buys at what price
00:04:29
because it seems like I mean it seems
00:04:31
like more companies and maybe correct me
00:04:33
if I'm wrong that more companies have
00:04:34
moved towards more unpredictable
00:04:37
pricing. I mean, it used to be it seems
00:04:38
like that, you know, there was there was
00:04:39
the price, then it went on sale, then it
00:04:41
went on clearance, and now it seems to
00:04:42
be like, you know, one day it could be
00:04:43
50% off, one day it could be 30, the
00:04:45
next day it could be full price, and
00:04:46
then it could be 50 again. I mean, it
00:04:48
seems like companies are actually moving
00:04:49
towards that as opposed to
00:04:50
predictability. And is that hurting
00:04:53
them? I think it depends on the market.
00:04:55
And an interesting thing is that um in
00:04:57
this sort of setting, we find that being
00:04:59
predictable, being simple, but also
00:05:01
having this um uh some degree of
00:05:04
flexibility is actually the right way to
00:05:06
go. So um you are capturing from the
00:05:08
firm's perspective uh most of that
00:05:10
value. An interesting thing there I
00:05:11
think that you're sort of mentioning is
00:05:12
that um uh if you think about an
00:05:15
industry where that sort of uh quickly
00:05:18
changing pricing has been very
00:05:19
successful. An example would be the
00:05:20
airline industry where you might be on a
00:05:23
plane and you sit next to someone who's
00:05:25
paid a very different price for for the
00:05:27
same ticket. U for in I'm pretty price
00:05:29
sensitive so I might have bought a
00:05:31
cheaper ticket. Um and the other thing
00:05:33
in that setting is when they do that
00:05:35
very successfully um the plane tends to
00:05:38
be full. So they tend to be able to
00:05:39
allocate all of um the seats that they
00:05:42
have. So that's sort of the price I
00:05:43
think you pay for having that cheaper
00:05:45
ticket. But the same message carries
00:05:47
over into this setting. We find that
00:05:49
when you do this sort of pricing
00:05:50
correctly um with these um simple sort
00:05:53
of policies, you're actually able to
00:05:56
sell a lot more units. you're actually
00:05:58
able to put more products profitably in
00:06:00
the hands of more people who want uh
00:06:03
those products and also um with these
00:06:06
simple uh policies you're actually able
00:06:08
to get more of those products into the
00:06:10
hands of uh the consumers who want them
00:06:12
the most. Um so there's an allocative uh
00:06:15
role there. So uh I think an important
00:06:17
message here is that um from a consumer
00:06:19
welfare standpoint uh this sort of uh
00:06:22
pricing um can have um ripple effects
00:06:25
that have positive implications. So if
00:06:28
I'm a retailer and looking at this
00:06:30
research, what are some ways that
00:06:31
practical ways that retailers could kind
00:06:32
of apply this or some sort of advice or
00:06:34
takeaways that they could have from it
00:06:35
that they could use in their business?
00:06:38
Uh I I think one is um to uh to be able
00:06:42
to understand why certain price uh
00:06:45
policies might work including ones that
00:06:46
you're using already. So in this case um
00:06:49
our retailer one thing that was
00:06:50
interesting is we asked them why are you
00:06:52
using this type of policy and they
00:06:54
almost think of the customer sort of
00:06:56
like a pet or a dog where if you train
00:06:59
them the wrong way they'll just start
00:07:01
expecting to wait for a markdown. Um so
00:07:03
this was their way actually um
00:07:05
heristically of uh sort of uh not
00:07:08
training the customer by introducing
00:07:10
this uncertainty making them unsure but
00:07:12
actually what we found is you have
00:07:13
different types of customers who have
00:07:15
these different costs of monitoring this
00:07:17
channel and that's really what's driving
00:07:19
um what was good about this way of
00:07:21
pricing. Um so one is to um if you have
00:07:24
a lot of data be able to understand even
00:07:26
with a simple policy why is it
00:07:27
effective. Um and the second message
00:07:30
there was that um again going to the
00:07:32
sort of uh simplicity
00:07:35
um that simple the message that simple
00:07:36
works um in this setting where you were
00:07:38
trying to say give coupons to your most
00:07:41
price sensitive uh customers identify
00:07:43
who they are and you have this mountain
00:07:45
of data recording all of their behavior
00:07:47
online. Um what we find is that uh
00:07:50
there's some very strong signals from
00:07:52
that data. So you only need a few
00:07:53
things. Um if you look at how people
00:07:56
monitor online the frequency with with
00:07:58
which they monitor that's a very strong
00:08:00
signal of their price elasticity. So um
00:08:03
you actually don't need to take you
00:08:05
don't need to always be using all of
00:08:08
that information. Tracking something
00:08:09
very simple like the ratio of purchases
00:08:12
to visits online is actually a very
00:08:14
strong signal and captures almost all of
00:08:16
that value um that you would have from
00:08:19
uh sophist a sophisticated analysis of
00:08:21
all the data.
00:08:23
And so what's next for this research or
00:08:25
what are you planning on looking at
00:08:26
next? Um I I think directly um there uh
00:08:31
the most related thing would be looking
00:08:33
at these sort of informationational
00:08:34
costs, these frictions um in a number of
00:08:36
other settings. Um uh I'm I'm doing some
00:08:39
work in online marketplaces and and
00:08:41
other places where you can get very
00:08:42
interesting data um at the sort of
00:08:45
granular level. Um but more broadly I
00:08:48
think there are a lot of settings that
00:08:49
are becoming much more informationally
00:08:51
rich whether it's um firms that are able
00:08:54
to track you online or um uh as a
00:08:57
patient or in a marketplace or or even
00:09:00
in the workplace. And I think um an
00:09:03
important aspect of these changes is to
00:09:06
be able to understand how does it affect
00:09:08
um firms who are sort of experimenting
00:09:11
to see what can they do with this sort
00:09:12
of data but also how should consumers
00:09:15
and workers feel about um uh how
00:09:18
comfortable should they feel about these
00:09:19
changes. So I think it's a very
00:09:21
interesting space from a research
00:09:23
standpoint. You get lots of data so it's
00:09:24
very interesting as well and uh um I'm
00:09:27
I'm excited about it. Great. Thanks Ken.
00:09:29
Thanks so much for being here. Oh, thank
00:09:30
you.
00:09:42
[Music]
