The document discusses how Amazon and Netflix used data analysis to develop successful TV shows. Amazon held a competition to select TV show pilots, then analyzed viewer data like ratings and histories to develop shows. They concluded a sitcom about Republican senators would do well but "Alpha House" was only average. Netflix's Chief Content Officer looked at their viewer data to make "House of Cards", betting on a drama about a senator, which became a hit with a 9.1 rating. However, the document notes that while data analysis is useful, it does not always lead to optimal results, and following data alone can lead to wrong decisions. Complex problems require both deep analysis of parts and combining them insightfully.
3. Amazon Studios Vs. Netflix
● Ro y Price, senio r
executive wants to
engineer success
● Ho lds a co mpetitio n
– takes ideas fo r TV
sho ws and after
evaluatio n, selects 8 .
● Runs pilo t episo de o f
each o nline fo r free
● Ted Sarando s, Chief
Co ntent Officer
wants a hit sho w
● Lo o ked at allthe
data they already had
abo ut Netflix viewers
– audience ratings,
viewer histo ries, etc.
4. Amazon Studios Vs. Netflix
● Watches the viewers
when they watch the
sho ws.
● Co llect data, crunch
data and co nclude
that Amazo n sho uld
do a sitco m abo ut 4
Republican US
senato rs.
● And so , Amazo n
● Use that data to
figure o ut what
sho ws, pro ducers and
acto rs viewers like
● After this analysis,
to o k a leap o f faith
and made a drama
series abo ut a single
senato r
● And so , Netflix came
6. Amazon Studios Vs. Netflix
● “ Alpha Ho use”
turned o ut to be a
slightly abo ve
average sho w with
a rating o f 7 .5
( average = 7 .4)
● “ Ho use o f Cards”
go t a rating o f 9.1
which is exactly
where the team
wanted it to be
7.
8. should come out wit h TV shows wit h great
rat ings. Unf ort unat ely, t here is act ually
some evidence t hat t his dat a analysis,
despit e having lot s of dat a, does not
always produce opt imum result s.
Examples of cases where error occurs -
●
Sof t ware company Mult i-Healt h Syst ems
may not have decided opt imally who is
supposed t o get parole or not
●
From 2009, af t er years of correct ly
9. decision-making wit h dat a and
unsuccessf ul decision-making and t here' s
a pat t ern involved.
Whenever you' re solving a complex
problem, you' re doing essent ially t wo
t hings. The f irst one is, you t ake t hat
problem apart int o it s bit s and pieces so
t hat you can deeply analyze t hose bit s and
pieces, and t hen you put all of t hese bit s
and pieces back t oget her again t o come t o
your conclusion.
10. mat t er how powerf ul, can only help you
t aking a problem apart and
underst anding it s pieces. There' s
anot her t ool t hat can do t hat , and we all
have it , and t hat t ool is t he brain. I f
t here' s one t hing a brain is good at , it ' s
t aking bit s and pieces back t oget her
again, even when you have incomplet e
inf ormat ion, and coming t o a good
conclusion, especially if it ' s t he brain of
an expert .
11. List the two most
(important/interesting/
informative)
insights from this talk ?
12. The first insight is that - Data is of course a
massively useful tool to make better
decisions, but things go wrong when data
starts to drive those decisions.
13. NO data or data analysis is perfect. Even
established companies can go wrong.
One has to learn to take decisions and
take risks.
14. Why and How are these
insights relevant to a
manager in India?
15. ●
It gives managers an idea of how
wrong improper techniques of
data-driven decisions can go. It is
especially important for medium
and small enterprises with limited
budget.
●
Sometimes, risks pay off. Managers
must also learn to trust their
instincts. A manager must listen to
his brain along with data analysis.