Tip:
Highlight text to annotate it
X
ERIC BRADLOW: Most of us know about the World Cup, the most
watched sporting event in the world.
ESPN obviously provides many different platforms to watch
the World Cup.
Obviously you can follow it on the Web.
You can also video stream things on ESPN 360.
And, of course, newly you can now watch things on your
mobile device.
And so as someone that's a portal, that provides value to
advertisers, they want to understand how much
consumption is going to happen across these multiple
platforms.
And so through the Wharton Interactive Media Initiative,
and we formed what I'll call a research lab around this, with
my colleague Pete Fader and I as kind of the heads of that
research lab.
Dr. Elea Feit, who's someone we just hired
as a research director.
We have a number of doctoral students working on the
project as well.
We're going to take the first two weeks of
the World Cup data.
We're going to take about a 100,000 people in their
consumption behavior across, again, the Web, mobile, and
video streaming, and we're going to forecast their
consumption across the remainder of the World Cup.
And so the way we're describing this project, and
the way it was announced in the press release, is that the
Wharton Interactive Media Initiative is going to develop
a platform agnostic forecasting tool.
So we're going to help them forecast reach and frequency
of the World Cup in and out of sample period, which they can
then go to advertisers about and say, hey look.
This is the value of advertising.
If you want to reach this demographic, you should go
towards this channel versus another channel.
And so it's really taking what I've done over the last 20
years as a researcher, which is build mathematical
forecasting models, but applying it to a very rich and
unique data set.
So we've actually gotten some early data from ESPN on the
NCAA tournament.
And so we're actually going to pretend like the NCAA
tournament was actually the World Cup, do our forecasting
model on the NCAA data, see how accurate we are.
Tweak the model, make it so that it actually does a good
job of forecasting.
And then once the World Cup comes around, we'll be really
ready to go.
It'll just be be data in, apply the model, generate the
forecast, and see how we're doing.
I think we're expecting a lot of things here.
I mean we're really looking at US consumption of World Cup,
so one of the things we expect to see, not
surprisingly, is a US effect.
If the US gets eliminated, that's actually likely to
severely damage the World Cup.
As a matter of fact, the group that the US is in will
probably get heightened interest in the US, because
whether they can advance or not.
And so what's interesting about this is not that we're
into gambling, but in some sense we have to incorporate
the betting lines into our forecast. Because we have to
understand who's likely to be around in later rounds
of the World Cup.
So it's an interesting marketing problem, where, in
some sense the competitors over time are changing.
In this case the competitors aren't Coke, Pepsi, Sprite, et
cetera, the competitors are the US, Brazil, Great Britain,
et cetera, et cetera.
So it's really a challenging problem.
Our early results from the NCAA tournament suggest that
we can predict things extremely accurately.
However there a lot of challenges there.
There's day of the week effects, there's timezone
effects, there's effects due to which
competitors are playing.
Actually the same model that we've applied to the NCAA
we'll apply to the World Cup.
It's a general class of models called Item Response Theory
Models where in some sense, teams have attractiveness
parameters.
And so, for example, the US will have a massive
attractiveness parameter.
Brazil will likely in the World Cup have a massive
attractiveness parameter.
And so countries where either they're historically good in
soccer, and/or countries that are of interest to the US,
possibly because of the racial and ethnic population in the
US, will actually get estimated as
having larger impact.
And obviously as they go on in the tournament, that will lead
to greater viewership across the different platforms.