How to Turn Online Data into a Pricing Strategy That Works
o
If you’re a
frequent online shopper, you already know you’re being watched. Companies
collect reams of data based on your browsing and purchasing patterns. There’s
so much data, in fact, that it’s easy to get lost in the numbers. That’s
where Ken Moon comes in. A professor of operations, information and decisions
at Wharton, Moon is researching how retailers can use online data to create
effective pricing policies. He visited with Knowledge@Wharton to discuss his
paper, “Randomized Markdowns and Online Monitoring,” which was co-authored with
Kostas Bimpikis and Haim Mendelson of Stanford University’s Graduate School of
Business.
An edited transcript of the
conversation follows.
Knowledge@Wharton: Could you give us a summary of your research?
Ken Moon: I do research in empirical operations. That means
two things: I work with data, sometimes in collaboration with companies,
hospitals or marketplaces. And I’m really trying to be prescriptive about
decision-making in my research. In this particular project, we worked with a
retailer that is active online. It was a very detailed, customer-level data
set. What’s interesting about it is that you can track a single customer both
online and offline. For instance, if the customer today were to go on his or
her phone, look at a product, go to a computer tomorrow, look at it, and then
walk into a store another day and buy it, we would be able to track all of
those things. It really opened up a lot of avenues to explore.
One thing that’s very interesting about this
particular project was that we were able to look at a very information-rich
environment in a way that we see in our everyday lives. If we think a product
that we’re interested in might drop in price, we can check our smartphone. It’s
also the case now that these companies can track all of this information at a
very individual level. For both sides, it’s a very information-rich
environment. It’s very interesting to think about how that affects the
decisions of companies that are active in this space and how it affects
outcomes for customers and consumers.
Knowledge@Wharton: When you were looking at this data set, what did
you find about price monitoring?
Moon: I think one broad takeaway is that information
seems to matter. You do have customers who are very intensive in their
monitoring. They’re typically the more price-sensitive customers. But also, their
opportunity cost to be doing this sort of monitoring is very low, so they are
going to be checking very often. Your most price-insensitive customers are not
checking very often. It’s about every 20 days, on average, between visits. It’s
a very big difference in terms of how these consumers are able to access
information, even from this very ubiquitous channel. And it makes a big
difference in terms of outcomes as well.
Knowledge@Wharton: Is there a clear indication of how companies should
be playing with price based on how someone is monitoring price?
Moon: Those are exactly some of the issues that we wanted
to explore in this research. One of the interesting things that we found is
that, even in this very informationally rich space, sometimes very simple
policies and very simple decisions can be very effective. You can capture most
of the value as a firm.
To give an example, the retailer that we
worked with follows a very simple pricing policy. For each product, you’re
going to start at a certain price, a list price. Then at a certain point in
time during a season, you drop the price down to its sale price, which is a
very predictable percentage of that initial list price. Finally, you move to
another predictable price, a clearance price, where you’re trying to just get
the products off the shelves.
What’s interesting there is that the
consumers understand what prices they’ll see. It’s very predictable. But all
the retailer did was to make the timing of those markdowns unpredictable. By
doing something very simple like that, it really exacerbated the informational
asymmetry in terms of the cost of monitoring. Those customers who were
price-insensitive, for whom it was very costly to be monitoring often, were the
ones who couldn’t take advantage of a markdown when it happened. They
understood that, so they would buy earlier. There is an interesting aspect
there where this sort of pricing has an allocative role. You’re deciding who
buys at what price.
Knowledge@Wharton: It seems like more companies have moved toward
unpredictable pricing. There used to be the price, then it went on sale, then
it went on clearance. Now it seems that items could be 50% off one day, 30% off
the next day, or it could be full price and then 50% again. Is that
unpredictable strategy hurting retailers?
Moon: I think it depends on the market. In this sort of
setting, we find that being predictable, being simple but also having some
degree of flexibility, is actually the right way to go. You are capturing, from
the first perspective, most of that value. If you think about an industry where
that sort of quickly changing pricing has been very successful, an example
might be the airline industry. You might be on a plane, and you sit next to
someone who’s paid a very different price for the same ticket. I’m pretty
price-sensitive, so I might have bought a cheaper ticket.
The other thing in that setting is that when
they do that very successfully, the plane tends to be full. They tend to be
able to allocate all the seats that they have. That’s sort of the price I think
you pay for having that cheaper ticket. But the same message carries over into
this setting. We find that when you do this sort of pricing correctly, with
these simple sort of policies, you’re able to sell a lot more units. You’re
able to put more products profitably in the hands of more people who want those
products. There’s an allocative role there. I think an important message here
is that, from a consumer welfare standpoint, this sort of pricing can have
ripple effects that have positive implications.
Knowledge@Wharton: What are some practical ways that retailers can apply
this research?
Moon: I think one is to be able to understand why certain
price policies might work, including ones that you’re using already. In the
case of our retailer, we asked them, “Why are you using this type of policy?”
They almost think of the customer sort of like a pet or a dog, where if you
train them the wrong way, they’ll just start expecting to wait for a markdown.
This was their way, heuristically, of not training the customer by introducing
this uncertainty, making them unsure. But what we found was that you have
different types of customers who have these different costs of monitoring this
channel, and that’s really what’s driving what was good about this way of
pricing.
If you have a lot of data, be able to
understand, even with a simple policy, why it is effective. The second message
there was that simple works. In this setting where you were trying to, say,
give coupons to your most price-sensitive customers and identify who they are,
you have this mountain of data, recording all of their behavior online. What we
find is that there are some very strong signals from that data. You only need a
few things. If you look at how people monitor online, the frequency with which
they monitor, that’s a very strong signal of their price elasticity. You don’t
need to always be using all of that information. Tracking something very
simple, like the ratio of purchases to visits online, is a very strong signal
and captures almost all of that value that you would have from a sophisticated
analysis of all the data.
Knowledge@Wharton: What are you planning on looking at next?
Moon: The most related thing would be looking at these
informational costs, these frictions, in a number of other settings. I’m doing
some work in online marketplaces and other places where you can get very
interesting data at the granular level. But more broadly, there are a lot of
settings that are becoming much more informationally rich, whether it’s firms
that are able to track you online, as a patient, in the marketplace or even in
the workplace. An important aspect of these changes is to be able to understand
how it affects firms who are experimenting to see what they can do with this sort
of data, and how comfortable consumers and workers should feel about these
changes. It’s a very interesting space from a research standpoint. I’m excited about
it.
http://knowledge.wharton.upenn.edu/article/creating-retail-pricing-strategy/?utm_source=kw_newsletter&utm_medium=email&utm_campaign=2017-06-06
No comments:
Post a Comment