00:00:05
in this particular study we're
00:00:06
interested in the relationship between
00:00:08
television viewership and online sales
00:00:11
and asking the simple question
00:00:13
of whether their substitutes or
00:00:15
complements
00:00:16
the way that we studied this is we
00:00:18
gathered a pretty large comprehensive
00:00:20
data set
00:00:21
on viewers in germany
00:00:24
where we got tv viewership data at the
00:00:27
national level
00:00:29
hourly tv viewership data and then we
00:00:31
also got sales data from a large online
00:00:34
company
00:00:35
a company similar to ebay so it was a
00:00:37
large auction site
00:00:38
we performed the study over four years
00:00:41
or we got data for four years and
00:00:44
in addition to the tv viewership and the
00:00:46
online sales that we were most
00:00:48
interested in we also collected a lot of
00:00:50
other data that we used as controls
00:00:52
so we collected weather data because
00:00:54
weather has been shown in the past to
00:00:57
impact online sales we also
00:01:00
from a large bank in germany collected
00:01:03
average bank account to get a sense for
00:01:07
the potential for the for the country at
00:01:09
any given on any given day
00:01:11
and then in addition for ebay in germany
00:01:14
we got online sales as well
00:01:15
adver as well as their advertising spend
00:01:17
to get a sense for
00:01:19
um the competition for this particular
00:01:22
company so we use those as controls
00:01:24
to basically control for anything else
00:01:27
that could possibly explain the online
00:01:29
sales
00:01:29
and we found that tv viewership in this
00:01:32
case
00:01:33
is in fact a substitute for online sales
00:01:36
of the particular online retailer
00:01:42
so we found that there's this negative
00:01:45
relationship between tv viewership and
00:01:47
online sales and one might think that's
00:01:49
not surprising
00:01:50
however there were two competing
00:01:52
hypotheses going into the study
00:01:54
on the one hand you might think well you
00:01:56
have an attention budget
00:01:58
so you can't watch television and
00:02:01
use the internet at the same time and
00:02:04
therefore
00:02:04
one should be a substitute for the other
00:02:07
however
00:02:08
before we performed this study we
00:02:10
actually ran a large survey
00:02:12
where we asked a large number of people
00:02:15
in germany
00:02:16
about their online behavior specifically
00:02:19
whether they use the internet while
00:02:21
they're watching tv
00:02:23
so a competing hypothesis would be well
00:02:25
since there are more people online and
00:02:26
at home while watching television
00:02:28
online sales might actually go up so we
00:02:31
were able to
00:02:32
show at least in this context that tv
00:02:35
viewership has a negative relationship
00:02:37
with online sales and therefore
00:02:38
they're substitutes
00:02:44
well the first one of course is that
00:02:47
tv viewership at least on days where
00:02:50
there are
00:02:50
huge numbers of viewers watching
00:02:54
shows should be factored into models of
00:02:57
predicting online sales
00:02:58
so when companies predict online sales
00:03:01
they factor in weather like i mentioned
00:03:03
among other things and viewership should
00:03:05
be accounted for
00:03:07
and then for auction sites in particular
00:03:10
which was
00:03:10
sort of the context of this study
00:03:14
sellers on auction sites may want to pay
00:03:16
attention to
00:03:17
the closing date and time of their
00:03:20
auctions
00:03:20
and not have them coincide with times
00:03:23
when popular television shows are on
00:03:30
so i think there are a lot of potential
00:03:32
implications the first one being that
00:03:34
if in fact this is an attention story
00:03:37
where people can only pay attention to a
00:03:39
limited number of
00:03:40
activities at any given time then this
00:03:42
would have implications for tv shows
00:03:44
that are prompting
00:03:46
viewers to participate in the
00:03:49
in tv show activity while the tv shows
00:03:52
are going on for example
00:03:53
asking viewers to participate in polls
00:03:56
to tweet to post on facebook
00:03:58
all of these activities potentially have
00:04:01
the
00:04:04
unintended consequence of impacting
00:04:06
online sales and the interesting thing
00:04:07
about that
00:04:08
is the online sellers or companies in
00:04:11
general
00:04:11
are the customers of tv shows they buy
00:04:13
advertising and ultimately don't want
00:04:15
their sales impacted so i think this is
00:04:17
just
00:04:18
an eye opener perhaps for tv shows to
00:04:20
think about the
00:04:21
relationship between viewership and
00:04:26
sales
00:04:28
there are a lot of stories in the news
00:04:29
that are relevant to my research
00:04:31
um the tv industry and tv measurement
00:04:35
industry
00:04:35
is really overnight becoming
00:04:38
revolutionized in
00:04:39
the amount of data that we or that tv
00:04:42
shows and networks can collect
00:04:44
as well as tv measurement companies and
00:04:47
so
00:04:47
pretty much anyone involved in
00:04:49
television tv producers tv networks
00:04:52
tv advertisers are interested in the
00:04:55
relationship between
00:04:57
um what's happening on television and
00:04:59
what's happening online so specifically
00:05:02
people tend to focus on engagement so
00:05:05
how are people responding online as it
00:05:07
relates to social media
00:05:10
but also there are new ways to measure
00:05:12
audiences in terms of their demographics
00:05:14
and what they're doing so
00:05:16
if they're buying but in all of this the
00:05:19
nice thing is that we have all of this
00:05:21
data now that we can link in new and
00:05:23
interesting ways
00:05:24
that can help us better understand the
00:05:26
impact of tv and view
00:05:27
on viewership and sales and social media
00:05:29
in ways that we couldn't even imagine
00:05:31
before
00:05:36
so as i just mentioned there's a lot of
00:05:39
hype about
00:05:40
um sort of second screening and what
00:05:43
people are doing
00:05:44
while they're watching television and i
00:05:47
think there's this
00:05:48
general belief that it's actually a good
00:05:51
thing for television shows and for
00:05:53
um and for online sellers and for
00:05:56
sellers in general
00:05:57
um so here for the first time it's kind
00:05:59
of saying well maybe we should take a
00:06:01
step back and think about what the
00:06:02
impact of
00:06:03
television is on online sales so maybe
00:06:06
second screening isn't
00:06:08
positive for sellers just yet
00:06:15
it's really the data so we have data on
00:06:18
viewership
00:06:18
at the hourly level as well as all of
00:06:21
the sales data
00:06:22
from this online company and so the data
00:06:25
enables us to answer these questions
00:06:27
in the first place and
00:06:30
at least we think it's for the first
00:06:32
time
00:06:37
we have a planned project for bringing
00:06:39
this study to the u.s context
00:06:42
so extending it both because we're going
00:06:44
to use u.s viewership data
00:06:46
and then also because we're going to
00:06:48
look across different online retailers
00:06:50
so
00:06:50
in the first paper we focused on a
00:06:52
particular online retailer and auction
00:06:54
site
00:06:55
and in the second study we're going to
00:06:56
focus more on which types of online
00:06:59
firms are impacted the most so in
00:07:01
preliminary analysis what we found was
00:07:04
in the u.s context on days like the
00:07:06
super bowl and mba finales where people
00:07:08
are watching television in large numbers
00:07:11
while on average sales go down sales for
00:07:14
pizza go up
00:07:15
so it's possible that while
00:07:18
on average high levels of television
00:07:21
viewership
00:07:22
impact sales in a negative way that for
00:07:24
some categories of products
00:07:26
sales might actually go up so the plan
00:07:29
is to see
00:07:30
if there are variants across categories
00:07:32
as it relates to television viewership
00:07:33
and online sales
00:07:57
you
