How to use data to make a hit TV show
-
0:01 - 0:05Roy Price is a man that most of you
have probably never heard about, -
0:05 - 0:08even though he may have been responsible
-
0:08 - 0:15for 22 somewhat mediocre
minutes of your life on April 19, 2013. -
0:15 - 0:18He may have also been responsible
for 22 very entertaining minutes, -
0:18 - 0:20but not very many of you.
-
0:20 - 0:22And all of that goes back to a decision
-
0:22 - 0:24that Roy had to make
about three years ago. -
0:24 - 0:29So you see, Roy Price
is a senior executive with Amazon Studios. -
0:29 - 0:32That's the TV production
company of Amazon. -
0:32 - 0:35He's 47 years old, slim, spiky hair,
-
0:35 - 0:40describes himself on Twitter
as "movies, TV, technology, tacos." -
0:40 - 0:45And Roy Price has a very responsible job,
because it's his responsibility -
0:45 - 0:49to pick the shows, the original content
that Amazon is going to make. -
0:49 - 0:52And of course that's
a highly competitive space. -
0:52 - 0:54I mean, there are so many
TV shows already out there, -
0:54 - 0:57that Roy can't just choose any show.
-
0:57 - 1:01He has to find shows
that are really, really great. -
1:01 - 1:04So in other words, he has to find shows
-
1:04 - 1:06that are on the very right end
of this curve here. -
1:06 - 1:09So this curve here
is the rating distribution -
1:09 - 1:13of about 2,500 TV shows
on the website IMDB, -
1:13 - 1:16and the rating goes from one to 10,
-
1:16 - 1:19and the height here shows you
how many shows get that rating. -
1:19 - 1:24So if your show gets a rating
of nine points or higher, that's a winner. -
1:24 - 1:25Then you have a top two percent show.
-
1:26 - 1:29That's shows like "Breaking Bad,"
"Game of Thrones," "The Wire," -
1:29 - 1:32so all of these shows that are addictive,
-
1:32 - 1:35whereafter you've watched a season,
your brain is basically like, -
1:35 - 1:37"Where can I get more of these episodes?"
-
1:37 - 1:38That kind of show.
-
1:39 - 1:41On the left side, just for clarity,
here on that end, -
1:41 - 1:45you have a show called
"Toddlers and Tiaras" -- -
1:45 - 1:47(Laughter)
-
1:47 - 1:49-- which should tell you enough
-
1:49 - 1:51about what's going on
on that end of the curve. -
1:51 - 1:55Now, Roy Price is not worried about
getting on the left end of the curve, -
1:55 - 1:58because I think you would have to have
some serious brainpower -
1:58 - 2:00to undercut "Toddlers and Tiaras."
-
2:00 - 2:04So what he's worried about
is this middle bulge here, -
2:04 - 2:06the bulge of average TV,
-
2:06 - 2:09you know, those shows
that aren't really good or really bad, -
2:09 - 2:10they don't really get you excited.
-
2:10 - 2:15So he needs to make sure
that he's really on the right end of this. -
2:15 - 2:17So the pressure is on,
-
2:17 - 2:19and of course it's also the first time
-
2:19 - 2:21that Amazon is even
doing something like this, -
2:21 - 2:25so Roy Price does not want
to take any chances. -
2:25 - 2:27He wants to engineer success.
-
2:27 - 2:29He needs a guaranteed success,
-
2:29 - 2:31and so what he does is,
he holds a competition. -
2:31 - 2:35So he takes a bunch of ideas for TV shows,
-
2:35 - 2:37and from those ideas,
through an evaluation, -
2:37 - 2:41they select eight candidates for TV shows,
-
2:41 - 2:44and then he just makes the first episode
of each one of these shows -
2:44 - 2:47and puts them online for free
for everyone to watch. -
2:47 - 2:50And so when Amazon
is giving out free stuff, -
2:50 - 2:51you're going to take it, right?
-
2:51 - 2:56So millions of viewers
are watching those episodes. -
2:56 - 3:00What they don't realize is that,
while they're watching their shows, -
3:00 - 3:02actually, they are being watched.
-
3:02 - 3:04They are being watched
by Roy Price and his team, -
3:04 - 3:06who record everything.
-
3:06 - 3:09They record when somebody presses play,
when somebody presses pause, -
3:09 - 3:12what parts they skip,
what parts they watch again. -
3:12 - 3:14So they collect millions of data points,
-
3:14 - 3:16because they want
to have those data points -
3:16 - 3:19to then decide
which show they should make. -
3:19 - 3:21And sure enough,
so they collect all the data, -
3:21 - 3:24they do all the data crunching,
and an answer emerges, -
3:24 - 3:25and the answer is,
-
3:25 - 3:30"Amazon should do a sitcom
about four Republican US Senators." -
3:30 - 3:32They did that show.
-
3:32 - 3:34So does anyone know the name of the show?
-
3:35 - 3:36(Audience: "Alpha House.")
-
3:36 - 3:37Yes, "Alpha House,"
-
3:38 - 3:42but it seems like not too many of you here
remember that show, actually, -
3:42 - 3:43because it didn't turn out that great.
-
3:44 - 3:45It's actually just an average show,
-
3:45 - 3:50actually -- literally, in fact, because
the average of this curve here is at 7.4, -
3:50 - 3:52and "Alpha House" lands at 7.5,
-
3:52 - 3:54so a slightly above average show,
-
3:54 - 3:57but certainly not what Roy Price
and his team were aiming for. -
3:58 - 4:01Meanwhile, however,
at about the same time, -
4:01 - 4:03at another company,
-
4:03 - 4:07another executive did manage
to land a top show using data analysis, -
4:07 - 4:09and his name is Ted,
-
4:09 - 4:12Ted Sarandos, who is
the Chief Content Officer of Netflix, -
4:12 - 4:14and just like Roy,
he's on a constant mission -
4:14 - 4:16to find that great TV show,
-
4:16 - 4:18and he uses data as well to do that,
-
4:18 - 4:20except he does it
a little bit differently. -
4:20 - 4:24So instead of holding a competition,
what he did -- and his team of course -- -
4:24 - 4:27was they looked at all the data
they already had about Netflix viewers, -
4:27 - 4:29you know, the ratings
they give their shows, -
4:29 - 4:32the viewing histories,
what shows people like, and so on. -
4:32 - 4:34And then they use that data to discover
-
4:34 - 4:37all of these little bits and pieces
about the audience: -
4:37 - 4:38what kinds of shows they like,
-
4:38 - 4:40what kind of producers,
what kind of actors. -
4:40 - 4:43And once they had
all of these pieces together, -
4:43 - 4:44they took a leap of faith,
-
4:44 - 4:47and they decided to license
-
4:47 - 4:49not a sitcom about four Senators
-
4:49 - 4:52but a drama series about a single Senator.
-
4:53 - 4:54You guys know the show?
-
4:54 - 4:56(Laughter)
-
4:56 - 4:59Yes, "House of Cards," and Netflix
of course, nailed it with that show, -
5:00 - 5:02at least for the first two seasons.
-
5:02 - 5:06(Laughter) (Applause)
-
5:06 - 5:09"House of Cards" gets
a 9.1 rating on this curve, -
5:09 - 5:12so it's exactly
where they wanted it to be. -
5:12 - 5:14Now, the question of course is,
what happened here? -
5:15 - 5:17So you have two very competitive,
data-savvy companies. -
5:17 - 5:20They connect all of these
millions of data points, -
5:20 - 5:22and then it works
beautifully for one of them, -
5:22 - 5:24and it doesn't work for the other one.
-
5:24 - 5:26So why?
-
5:26 - 5:29Because logic kind of tells you
that this should be working all the time. -
5:29 - 5:32I mean, if you're collecting
millions of data points -
5:32 - 5:33on a decision you're going to make,
-
5:33 - 5:36then you should be able
to make a pretty good decision. -
5:36 - 5:38You have 200 years
of statistics to rely on. -
5:38 - 5:41You're amplifying it
with very powerful computers. -
5:41 - 5:45The least you could expect
is good TV, right? -
5:46 - 5:49And if data analysis
does not work that way, -
5:50 - 5:52then it actually gets a little scary,
-
5:52 - 5:55because we live in a time
where we're turning to data more and more -
5:55 - 6:00to make very serious decisions
that go far beyond TV. -
6:01 - 6:04Does anyone here know the company
Multi-Health Systems? -
6:05 - 6:07No one. OK, that's good actually.
-
6:07 - 6:10OK, so Multi-Health Systems
is a software company, -
6:10 - 6:13and I hope that nobody here in this room
-
6:13 - 6:16ever comes into contact
with that software, -
6:16 - 6:18because if you do,
it means you're in prison. -
6:18 - 6:19(Laughter)
-
6:19 - 6:23If someone here in the US is in prison,
and they apply for parole, -
6:23 - 6:27then it's very likely that
data analysis software from that company -
6:27 - 6:31will be used in determining
whether to grant that parole. -
6:31 - 6:33So it's the same principle
as Amazon and Netflix, -
6:33 - 6:38but now instead of deciding whether
a TV show is going to be good or bad, -
6:38 - 6:41you're deciding whether a person
is going to be good or bad. -
6:41 - 6:47And mediocre TV, 22 minutes,
that can be pretty bad, -
6:47 - 6:49but more years in prison,
I guess, even worse. -
6:50 - 6:54And unfortunately, there is actually
some evidence that this data analysis, -
6:55 - 6:59despite having lots of data,
does not always produce optimum results. -
6:59 - 7:01And that's not because a company
like Multi-Health Systems -
7:02 - 7:03doesn't know what to do with data.
-
7:03 - 7:05Even the most data-savvy
companies get it wrong. -
7:05 - 7:08Yes, even Google gets it wrong sometimes.
-
7:09 - 7:13In 2009, Google announced
that they were able, with data analysis, -
7:13 - 7:17to predict outbreaks of influenza,
the nasty kind of flu, -
7:17 - 7:21by doing data analysis
on their Google searches. -
7:21 - 7:25And it worked beautifully,
and it made a big splash in the news, -
7:25 - 7:27including the pinnacle
of scientific success: -
7:27 - 7:30a publication in the journal "Nature."
-
7:30 - 7:33It worked beautifully
for year after year after year, -
7:33 - 7:35until one year it failed.
-
7:35 - 7:37And nobody could even tell exactly why.
-
7:37 - 7:39It just didn't work that year,
-
7:39 - 7:41and of course that again made big news,
-
7:41 - 7:43including now a retraction
-
7:43 - 7:45of a publication
from the journal "Nature." -
7:46 - 7:50So even the most data-savvy companies,
Amazon and Google, -
7:50 - 7:52they sometimes get it wrong.
-
7:52 - 7:55And despite all those failures,
-
7:55 - 7:59data is moving rapidly
into real-life decision-making -- -
7:59 - 8:01into the workplace,
-
8:01 - 8:02law enforcement,
-
8:03 - 8:04medicine.
-
8:04 - 8:08So we should better make sure
that data is helping. -
8:08 - 8:11Now, personally I've seen
a lot of this struggle with data myself, -
8:11 - 8:13because I work in computational genetics,
-
8:13 - 8:15which is also a field
where lots of very smart people -
8:15 - 8:19are using unimaginable amounts of data
to make pretty serious decisions -
8:19 - 8:23like deciding on a cancer therapy
or developing a drug. -
8:24 - 8:26And over the years,
I've noticed a sort of pattern -
8:26 - 8:28or kind of rule, if you will,
about the difference -
8:28 - 8:31between successful
decision-making with data -
8:31 - 8:33and unsuccessful decision-making,
-
8:33 - 8:37and I find this a pattern worth sharing,
and it goes something like this. -
8:39 - 8:41So whenever you're
solving a complex problem, -
8:41 - 8:42you're doing essentially two things.
-
8:42 - 8:46The first one is, you take that problem
apart into its bits and pieces -
8:46 - 8:48so that you can deeply analyze
those bits and pieces, -
8:48 - 8:50and then of course
you do the second part. -
8:50 - 8:53You put all of these bits and pieces
back together again -
8:53 - 8:54to come to your conclusion.
-
8:54 - 8:57And sometimes you
have to do it over again, -
8:57 - 8:58but it's always those two things:
-
8:58 - 9:01taking apart and putting
back together again. -
9:02 - 9:04And now the crucial thing is
-
9:04 - 9:07that data and data analysis
-
9:07 - 9:09is only good for the first part.
-
9:09 - 9:12Data and data analysis,
no matter how powerful, -
9:12 - 9:16can only help you taking a problem apart
and understanding its pieces. -
9:16 - 9:20It's not suited to put those pieces
back together again -
9:20 - 9:21and then to come to a conclusion.
-
9:22 - 9:24There's another tool that can do that,
and we all have it, -
9:24 - 9:26and that tool is the brain.
-
9:26 - 9:28If there's one thing a brain is good at,
-
9:28 - 9:30it's taking bits and pieces
back together again, -
9:30 - 9:32even when you have incomplete information,
-
9:32 - 9:33and coming to a good conclusion,
-
9:33 - 9:36especially if it's the brain of an expert.
-
9:36 - 9:39And that's why I believe
that Netflix was so successful, -
9:39 - 9:43because they used data and brains
where they belong in the process. -
9:43 - 9:46They use data to first understand
lots of pieces about their audience -
9:46 - 9:50that they otherwise wouldn't have
been able to understand at that depth, -
9:50 - 9:52but then the decision
to take all these bits and pieces -
9:52 - 9:56and put them back together again
and make a show like "House of Cards," -
9:56 - 9:57that was nowhere in the data.
-
9:57 - 10:01Ted Sarandos and his team
made that decision to license that show, -
10:01 - 10:04which also meant, by the way,
that they were taking -
10:04 - 10:06a pretty big personal risk
with that decision. -
10:06 - 10:09And Amazon, on the other hand,
they did it the wrong way around. -
10:09 - 10:12They used data all the way
to drive their decision-making, -
10:12 - 10:15first when they held
their competition of TV ideas, -
10:15 - 10:18then when they selected "Alpha House"
to make as a show. -
10:18 - 10:21Which of course was
a very safe decision for them, -
10:21 - 10:23because they could always
point at the data, saying, -
10:23 - 10:25"This is what the data tells us."
-
10:25 - 10:29But it didn't lead to the exceptional
results that they were hoping for. -
10:30 - 10:35So data is of course a massively
useful tool to make better decisions, -
10:35 - 10:37but I believe that things go wrong
-
10:38 - 10:40when data is starting
to drive those decisions. -
10:40 - 10:44No matter how powerful,
data is just a tool, -
10:44 - 10:47and to keep that in mind,
I find this device here quite useful. -
10:47 - 10:48Many of you will ...
-
10:49 - 10:50(Laughter)
-
10:50 - 10:51Before there was data,
-
10:51 - 10:54this was the decision-making
device to use. -
10:54 - 10:55(Laughter)
-
10:55 - 10:56Many of you will know this.
-
10:57 - 10:58This toy here is called the Magic 8 Ball,
-
10:58 - 11:00and it's really amazing,
-
11:00 - 11:03because if you have a decision to make,
a yes or no question, -
11:03 - 11:06all you have to do is you shake the ball,
and then you get an answer -- -
11:06 - 11:09"Most Likely" -- right here
in this window in real time. -
11:09 - 11:11I'll have it out later for tech demos.
-
11:11 - 11:13(Laughter)
-
11:13 - 11:16Now, the thing is, of course --
so I've made some decisions in my life -
11:16 - 11:19where, in hindsight,
I should have just listened to the ball. -
11:19 - 11:22But, you know, of course,
if you have the data available, -
11:22 - 11:26you want to replace this with something
much more sophisticated, -
11:26 - 11:29like data analysis
to come to a better decision. -
11:29 - 11:32But that does not change the basic setup.
-
11:32 - 11:35So the ball may get smarter
and smarter and smarter, -
11:35 - 11:38but I believe it's still on us
to make the decisions -
11:38 - 11:41if we want to achieve
something extraordinary, -
11:41 - 11:43on the right end of the curve.
-
11:43 - 11:47And I find that a very encouraging
message, in fact, -
11:47 - 11:51that even in the face
of huge amounts of data, -
11:51 - 11:55it still pays off to make decisions,
-
11:56 - 11:58to be an expert in what you're doing
-
11:58 - 12:00and take risks.
-
12:00 - 12:03Because in the end, it's not data,
-
12:03 - 12:07it's risks that will land you
on the right end of the curve. -
12:08 - 12:09Thank you.
-
12:09 - 12:13(Applause)
- Title:
- How to use data to make a hit TV show
- Speaker:
- Sebastian Wernicke
- Description:
-
Does collecting more data lead to better decision-making? Competitive, data-savvy companies like Amazon, Google and Netflix have learned that data analysis alone doesn't always produce optimum results. In this talk, data scientist Sebastian Wernicke breaks down what goes wrong when we make decisions based purely on data -- and suggests a brainier way to use it.
- Video Language:
- English
- Team:
- closed TED
- Project:
- TEDTalks
- Duration:
- 12:25
Jenny Zurawell edited English subtitles for How to use data to make a hit TV show | ||
Jenny Zurawell edited English subtitles for How to use data to make a hit TV show | ||
Helene Batt edited English subtitles for How to use data to make a hit TV show | ||
Brian Greene edited English subtitles for How to use data to make a hit TV show | ||
Brian Greene approved English subtitles for How to use data to make a hit TV show | ||
Brian Greene edited English subtitles for How to use data to make a hit TV show | ||
Brian Greene edited English subtitles for How to use data to make a hit TV show | ||
Brian Greene edited English subtitles for How to use data to make a hit TV show |