The science of storytelling | Prasad Setty, Google People Analytics


[MUSIC PLAYING] I’m Prasad Setty. I lead, among other groups
and people operations at Google, our people analytics
and communications groups as well. Now that we’re
back from lunch, I know exactly what’s going
on in all your minds. You’re thinking, there
are those napping parts that we hear are
all over Google? We’ve hidden them away. But I am going to indulge
you for a little minute. Close your eyes, everyone. We’re going to do a little
bit of a thought experiment to begin with. I want you to think about your
most favorite piece of artwork. Some of you might think
about a masterpiece from one of your most favorite
post-impressionist artists. Others might think
about the dinosaur their kid drew in first
grade and is still on their refrigerator. Mine is this charcoal
piece of work that my wife did and was her
very first present to me. What emotions come
into your mind when you think
about this artwork? What meaning does it
have in your life? Keep your eyes closed
this a little longer. Now I want you to think
about something different. I want you to think about
the most compelling piece of science or analytical work
that registers in your life; again, something that has
a deep, personal meaning. And for all the academics
in the audience, you can’t think about
your own research. That’s would be too easy. No thinking about
your own research. Give it a couple more seconds
and now open your eyes. Wasn’t the second exercise
much, much more difficult? I see a lot of heads nodding. We spend an inordinate
amount of time doing hardcode
science and analytics. But how do we ensure
that it’s memorable? How do we ensure that
we can communicate better so that our messages
resonate and stick? Over the next 30 minutes,
that’s exactly what we’re going to explore. The speakers that follow me
Michelle Gielan and Christine O’Connell, have the answers. I, on the other hand,
get to play executive. So I’m going to vent– there’s
going to be a lot of venting. And I’m just going to leave
you with a lot of problems to solve. Tim Chatwin, who leads
communications and public relations for Google in
our Asia Pacific region, and who used to be the
speechwriter for David Cameron, the prime
minister of the UK before he joined Google–
that is Tim joined Google, not David Cameron– when you
ask him what he thinks about as good communications
he says there are three things, It all boils down
to three things– what do you want your
audience to know, how do you want them to
feel, and what do you want them to do. And when it comes to
communicating science and analytics, we typically fall
short on all three questions. Instead of telling people
what they should know, we like to tell
them what we did. We like to use a lot
of highfalutin jargon in all of our work. It takes a PhD typically
to understand the work that another PhD does. Of course there’s
the age-old question of if a tree falls in a
forest does it make a sound. The philosophers can
duke that one out. But I have a follow-on
question for you. If a tree falls in a forest and
we use the pulp up the print and publish a prestigious
academic journal, did we suck out even one
more sound out of it? [LAUGHTER] And I think we know
the answer to that one. So I’m really glad
that Christine is going to come up on
stage soon and tell us how to communicate science
in the comprehensible manner. Tim’s second question is
even more of a problem. We don’t even think
about emotions when we communicate
science and analytics. It’s as if in our
quest for objectivity and rational thinking
we try to strip away all emotion from our speech. And that becomes less memorable. So personally,
for instance, I’ve been at Google for
a few years now. And I have a pretty vivid
memory of everything that we’ve done
in analytics here. I spent five years at
my previous employer. And if you ask me what I
recollect from that time, I can really think of one
vivid analytical example. The organization was going
through a troubled phase and we needed to lay off
a significant fraction of the workforce. And we had develop the
right severance packages for these people. The analytical team
developed a simple visual that showed what happens under
the existing severance policy. Executives were going to
make much, much more money than your typical rank
and file employee. And we shared that
with the CEO, who also happened to be the
founder of the organization and though of every
employee as a family member. He had such a visceral
reaction to seeing that visual, and immediately
made the decision to double the
severance for all rank and file employees,
while keeping executives exactly where they were. It was going to cost a ton
of money, but in his mind it was absolutely the
right thing to do. Going into this
piece of work, I just treated it as just
another piece of analysis. But coming out of the
meeting with the CEO, I could see the difference
that it was going to make in people’s lives. And as I think about
what we did there, it was quite by
accident that we were able to induce that emotion. I’m really glad
Michelle is going to tell us soon about
what it means to induce an emotion intentionally. That gets us to
Tim’s third question about how we get people
to act on information. And we have a long way
to go there as well. But this is an area
that Google has invested a tremendous amount of
effort in to try and improve. We experiment a lot with
all the communications that we do to see what
influences behavior. And just to look at
one particular example, we’ve been thinking about all
the advances in the literature around framing. As you know, how
you frame a message has an impact on what happens. To summarize from some of
the prominent researchers on the framing field,
here’s what they’re saying. There’s an understandable
but misguided tendency to try to mobilize action
against socially disapproved conduct by depicting it
as regrettably frequent, thereby inadvertently installing
a counterproductive descriptive norm in the minds
of their audiences. You got that? I put into to Google Translate. [LAUGHTER] And our machine
learning algorithms have still not caught up. So I asked Jessie Wisdom, who
is one of the PhD’s in our team, and speaks all
this research juju, and she told me what
this actually means. She said basically if you frame
something in a positive light it leads to better outcomes. So what is an example of that? So let’s say you’re
talking about ground beef– pretend you were. If you framed it
in a positive light and said that it was 75% lean
instead of negative light saying it was 25%
fat, apparently it tastes less greasy and it’s
going to be registering better. It’s going to sell better. The only people this kind
of framing doesn’t work on is the vegetarians, like me,
but who cares about them. So we’ve done similar types of
randomized controlled trials at Google as well. And what we’ve
found is that when you use social norms
to nudge people we actually do find
changes in behavior. We have fewer people
cancel interviews that they’re scheduled for,
fewer people cancel training sessions that they’re
registered for. Behavior change is possible. And later this
afternoon, we’re going to hear about a lot of
nudges that are good. So in summary, what am I saying? Just in case I haven’t
been here so far all, let me go meta and tell you
the answers to Tim’s three questions in my own talk. What I want to
know is that we are pretty terrible at communicating
science and analytics. And I want us to feel–
how do I put it gently– I want us to feel
really crappy about it. But fear not– all
the evidence suggests that 100% of people who listen
to Michelle and Christine walk away as better
communicators. So let’s pay
attention to them, OK? All right, thank you. [APPLAUSE] [MUSIC PLAYING]

Leave a Reply