00:00:02
today we're joined by Wharton marketing
00:00:03
Professor Jonah berer to talk about his
00:00:05
latest research on how language impacts
00:00:08
Word of Mouth thanks for joining us
00:00:09
Jonah thanks for having me so your
00:00:11
previous work has looked at uh how
00:00:12
things go viral and your latest book uh
00:00:15
invisible influence looks at how in
00:00:18
Hidden influences affect our
00:00:19
decision-making processes yeah could you
00:00:21
summarize what you and your co-author
00:00:22
looked at in this paper we've all seen
00:00:24
the power of Word of Mouth uh whether
00:00:26
we're making a simple decision like what
00:00:28
breakfast cereal to buy or more
00:00:29
important one like which house to choose
00:00:31
we use online reviews and word of mouth
00:00:33
all the time to help us make those
00:00:34
decisions but is that word of mouth
00:00:36
always helpful uh and that's really what
00:00:38
this with this research looks at so
00:00:39
imagine you're at a party for example or
00:00:41
at a conference uh and you're talking to
00:00:42
people about movies to two people you
00:00:44
haven't really met before one person
00:00:46
says they like movie A and another
00:00:48
person says they recommend movie B which
00:00:50
of those movies are you more likely to
00:00:51
see as a result and are you going to be
00:00:53
happy with your choice and what we find
00:00:55
I don't think so surprisingly uh is that
00:00:57
people are more likely to follow
00:00:58
recommendations you're more likely to
00:01:00
movie be cuz you think the other person
00:01:01
liked it more and it's a better movie
00:01:03
but actually you might end up not liking
00:01:05
that movie so much you might end up
00:01:06
making a worse choice because the type
00:01:08
of people that tend to use the word I
00:01:10
recommend it's a a language uh device I
00:01:13
recommend something that suggests not
00:01:14
only you like something but you're
00:01:16
making inference about what someone else
00:01:17
likes and it turns out that people that
00:01:19
say I recommend something whether we've
00:01:21
looked at books or wine or hotels across
00:01:23
range of domains it ends up that novices
00:01:25
are more likely to say that they
00:01:26
recommend something than experts experts
00:01:29
aren't as willing to use use that
00:01:30
recommend language they're more willing
00:01:31
to say I like something sure I like this
00:01:33
thing but they're less willing to make a
00:01:34
guess about what you're going to like
00:01:36
and so as a result if people end up
00:01:37
listening to recommendations as they
00:01:39
often do we might sometimes end up
00:01:41
making worse choices so what were the
00:01:43
key takeaways from from your study yeah
00:01:45
so we look at we look at two things one
00:01:47
how people endorse things so sometimes
00:01:49
people say I like something sometimes
00:01:51
say I recommend something those might
00:01:52
seem like really subtle differences in
00:01:54
language but they have a big impact on
00:01:56
two things first of all whether we're
00:01:57
persuaded by that language do we take
00:01:59
that person's endorsement and end up
00:02:01
going to see that movie or that
00:02:02
restaurant uh and also whether we end up
00:02:04
making a good choice as a result lots of
00:02:06
research shows that word of mouth is
00:02:08
really helpful and indeed often it is
00:02:10
but in some cases where we don't know if
00:02:11
someone's an expert or not like two
00:02:13
people we meet at a conference that
00:02:14
we've never met before should we take
00:02:16
their advice or not and when we can't
00:02:17
tell if someone's an expert or they're
00:02:19
not our best friend who knows a lot
00:02:20
about movies sometimes we use what they
00:02:22
say and the way they say it as a CU to
00:02:25
whether they're an expert or not so we
00:02:26
assume that if someone says I recommend
00:02:28
something they actually know a lot about
00:02:29
movies for example where if they just
00:02:31
say I like it we assume they don't know
00:02:33
as much what's dangerous there though is
00:02:35
that the opposite is actually true
00:02:36
novices people that don't know a lot
00:02:38
about movies are more willing to say
00:02:40
they recommend something same thing
00:02:42
happens with restaurants or other
00:02:43
domains because they don't think about
00:02:44
the fact that others may have different
00:02:46
preferences than them so if you're an
00:02:47
expert you're not really willing to say
00:02:49
I recommend something because if we
00:02:50
don't know each other well I don't know
00:02:51
your taste I don't know if you'd like
00:02:53
the same movies that I liked so I'm not
00:02:54
as willing to recommend it for you but
00:02:56
if I'm a novice I'm very willing to use
00:02:57
that strong recommendation to say I
00:02:59
recommend this movie whereas you as a
00:03:01
listener might end up seeing it and be
00:03:03
dissatisfied as a result so so my saying
00:03:06
I recommend this might actually mislead
00:03:08
some mislead somebody exactly yeah and
00:03:11
and um you know in general if we follow
00:03:13
experts that's a good thing right when
00:03:14
we look online we follow the wine
00:03:16
experts when we're picking wine we look
00:03:17
for reviewer badges to figure out who
00:03:19
knows a lot but there are many cases in
00:03:21
our lives where we don't know whether
00:03:22
someone knows a lot or not and so we use
00:03:24
their language as a cue to whether they
00:03:27
have expertise but that cue May
00:03:29
sometimes lead us to
00:03:30
did the finding surprise you definitely
00:03:32
yeah so uh we sort of thought uh that
00:03:34
recommendations would be more impactful
00:03:36
than likes uh that someone saying I
00:03:38
recommend this movie it seems stronger
00:03:40
uh it suggests one that they know a lot
00:03:42
about the domain but also they just
00:03:44
plain liked it more um and so that
00:03:46
didn't surprise us so much what did
00:03:47
surprise us a little bit is the type of
00:03:49
people that use these different types of
00:03:50
language and I've done a lot of work on
00:03:52
Word of Mouth and in general I think
00:03:53
word of mouth is a good thing but but in
00:03:55
this case sometimes Word of Mouth can
00:03:56
actually lead us astray because
00:03:58
different people tend to use different
00:03:59
types of language right and what do you
00:04:02
think the key takeaways of your study uh
00:04:04
which you co-authored with um with Grant
00:04:06
Packard yeah uh what do you think the
00:04:08
key takeaways would be for marketers for
00:04:10
marketers I think it's interesting you
00:04:11
know the first thing that comes to mind
00:04:13
probably when people listen to this when
00:04:14
they think of likes versus
00:04:15
recommendations is they think of
00:04:16
Facebook right where we say hey I like
00:04:18
this well if they instead change it to I
00:04:20
recommend this it'd probably have more
00:04:21
impact right knowing that a given movie
00:04:23
or a given brand has a certain number of
00:04:25
likes isn't as impactful as knowing it
00:04:27
has a certain amount of recommendations
00:04:29
so by subt changing the language you as
00:04:31
a marketer use you can impact whether
00:04:33
people follow that language or not that
00:04:35
said you need to be careful right
00:04:36
because what you don't want to encourage
00:04:38
people to follow the wrong information
00:04:40
we actually did a study where we show
00:04:41
that uh novices end up picking worse
00:04:43
wine for example end up being more
00:04:45
likely to recommend that wine and then
00:04:47
following their Word of Mouth leads
00:04:48
people to be more likely to choose bad
00:04:50
wine over good wine if they don't know
00:04:52
the difference um and so to be careful
00:04:54
there you know marketers need to think
00:04:55
about well are we allowing people to use
00:04:57
other cues as well are we just put
00:04:59
putting the language out there are we
00:05:00
giving information about how many books
00:05:02
that person has reviewed how many uh
00:05:04
other bottles of wine they've talked
00:05:05
about before uh or potentially other
00:05:07
cues that will allow people to get a
00:05:09
better sense is this good for this
00:05:11
person or might it be particularly good
00:05:12
for me okay and what um what about the
00:05:16
consumer side what are the implications
00:05:17
for people who are trying to get
00:05:19
information about what they like or what
00:05:20
they might want to buy y uh how do they
00:05:22
how can they judge a source next time
00:05:24
someone tells you they recommend
00:05:25
something be very very careful uh first
00:05:27
thing think about you know do you know
00:05:29
them or or not uh if it's someone you
00:05:30
know and you know their preferences and
00:05:32
they know your preferences the fact that
00:05:34
they're recommending something is
00:05:35
probably a good signal but if you don't
00:05:36
know them if youve just met them if you
00:05:38
don't know whether they know a lot about
00:05:39
a domain a red flag should go off if
00:05:41
someone's saying they recommend
00:05:43
something be a little bit more careful
00:05:44
figure out do they actually know
00:05:46
something about that domain or not do
00:05:48
they know something about me or not um
00:05:50
and make sure they know enough about
00:05:51
your preferences before you actually go
00:05:53
ahead and take that recommendation I can
00:05:55
imag I mean I could see this playing out
00:05:56
all the time on Facebook for example
00:05:58
certainly and I guess when it's your
00:06:00
close circle of friends it makes perfect
00:06:02
sense in some ways to accept a re an
00:06:05
explicit recommendation as you describe
00:06:06
it but I guess the the caution comes in
00:06:09
when it's someone you may not know yeah
00:06:10
and I think uh you know a lot of times
00:06:12
when we're talking face to face it's
00:06:13
pretty clear online the boundaries blur
00:06:16
a little bit you know we we talk to our
00:06:17
good friends but we also talk to a lot
00:06:19
of people we don't know uh as well and
00:06:21
we often assume those people know a lot
00:06:23
about what they're talking about they
00:06:24
may not necessarily and so that I think
00:06:26
is is what's important here you know
00:06:27
lots of work has been done on online
00:06:29
reviews the power of online reviews
00:06:31
they're certainly a valuable tool to
00:06:33
often help us make better and faster and
00:06:35
easier decisions but on the margin
00:06:37
there's some cases where it can can lead
00:06:38
us astray and that's what I think this
00:06:40
this research points out great well what
00:06:42
is it that you want to look at next
00:06:43
research-wise yeah you know I think one
00:06:45
thing this paper starts to do um and
00:06:47
researchers starting to look in this
00:06:48
area more generally is the language
00:06:50
people use when they describe things you
00:06:52
would think in marketing we thought a
00:06:53
lot about language right uh the language
00:06:55
that works in advertisements or why
00:06:57
certain language would be more effective
00:06:58
in Word of Mouth and not really haven't
00:07:01
uh that much um and so there's a lot of
00:07:03
work now bridging uh marketing as well
00:07:05
as computer science called natural
00:07:06
language processing that's starting to
00:07:08
look at not just well did people
00:07:10
recommend something but what did they
00:07:11
say in that recommendation even saying I
00:07:14
recommend something I can say I strongly
00:07:15
recommend this I recommend this for
00:07:17
people who like this type of other thing
00:07:19
and the words I'm using there can be
00:07:21
quite important and so while a lot of
00:07:23
research just looks at the number of
00:07:24
stars for example saying you know a
00:07:26
five-star review on Amazon leads to
00:07:27
about 20 more books being sold what's
00:07:30
also important to say is well what's the
00:07:31
language being used in that review and
00:07:33
how does certain types of language
00:07:34
affect others uh more more importantly
00:07:37
so we're doing a bunch of research on
00:07:38
this we're looking at the language of
00:07:39
customer service calls looking at if you
00:07:41
call up uh let's say a retailer and you
00:07:43
say I'm really unhappy you know what
00:07:45
words do they use that might make you
00:07:47
happier or not as happy uh we're doing
00:07:50
some work on the language used in movies
00:07:52
uh for example might the certain words
00:07:53
people use over time in movies lead
00:07:55
those movies to be more or less
00:07:56
successful and can we predict that based
00:07:58
on the language
00:08:00
of those movies even song lyrics can we
00:08:02
predict how successful a song's going to
00:08:04
be by the lyrics contained in that song
00:08:06
There's a a lot of interesting data out
00:08:07
there that's texturally based and we're
00:08:09
trying to understand it better really
00:08:10
fascinating a lot to explore there yeah
00:08:12
yeah well thanks very much for giving us
00:08:13
your time no problem thank
00:08:26
[Music]
00:08:28
you
00:08:29
[Music]
