Segmento is developing an algorithm to predict psychographics from online data in order to deliver better targeted display ads. They conducted a study of over 300 participants in the automobile industry and found that metrics like risk preference, social comparison bias, and narcissism correlate better to eventual purchase decisions than demographic data alone. Their goal is to build a self-service platform allowing companies to send personalized emails and landing pages based on the algorithm's predictions. They plan to initially generate revenue by selling this service to ad agencies and companies with customer relationship management databases.
Targeted Ads with Segmento's Psychographic Algorithm
1. Segmento
Samuel Smith gianna whitver
m.a. advertising b.a. advertising
b.a. psychology b.a. english
B.A. Philosophy
a.a.s. comp. sci.
with special thanks to our advisors:
brian sheehan, former ceo saatchi & saatchi paul morarescu, associate dean, data science
bob sollish, lead engineer, synacor sean branagan, entrepreneur
2. DISPLAY ADSsimply put, a
display ad is
any picture ad,
on desktop or
mobile
they're often
seen next to
content
7. we did research
we sampled over 300
participants from
across the u.s.
our hypothesis:
psychographic
targeting is more
effective than
demographic
targeting in the
automobile industry.
we measured:
risk preference
social comparison bias
narcissism
we learned:
these metrics correlate
better to eventual
purchase decisions than
demographic data.
8. future researchq. How can we possibly
determine a random
internet user's risk
preference?
a. perhaps via natural
language processing,
checkins, and shares
q. how do we expand
this outside the auto
industry?
a. every industry will
have its own set of
psychographic variables
to target, and brand
managers know what they
are.
q. how do we find new
psychographics that are
even more effective?
a. haven't gotten this far.
9. competitionlots of interest, little competition.
1. Google adWords
threats: the elephant in
the room. can squash me.
weaknesses: targets
behaviorally, which is
highly unreliable
too massive to make
monstrous changes in
how it targets ads
2. Google ad arts
threats: currently
working on personalized
advertising
weaknesses: only
customizes select parts
of the message, can't
select the right message
to send to a particular
customer
10. competition
descriptive personalization
Varsity ad: "Hey EVAN, HOW'S YOUR TUESDAY GOING?
I DON'T KNOW ABOUT YOU, BUT I'M THINKING ABOUT
HEADING TO VARSITY AFTER THIS FOR A SLICE OF PIZZA!"
BEHAVIORAL TARGETING
THE NATURE OF THE INTERNET TAKES CURIOUS PEOPLE
TO FARFLUNG CORNERS OF THE WEB. NOT EVERYONE
WHO LANDS ON A WEBPAGE WANTS TO BE SOLD THAT
PARTICULAR GENRE OF PRODUCT
AND IT DOESNT WORK FOR LANDING PAGES OR EMAILS AT ALL. THE DATA'S NOT THERE.
11. market size
$17 billion and growing
potential to disrupt all
of display and mobile
advertising
2010
$9b
2011
$12B
2012
$15B
2013
$17B
2014
$20b
12. deliverable
we are prototyping
a self service
platform for using
our personalization
technology to
send emails and
personalized
landing pages
in case you haven't noticed yet,
the magic is in the algorithm.
this is a proof of concept.
13. biz model
Key partners
direct mail
heroku
website
key activities
market scientist
coders
key resources
code
algorithm
value props
optimized
messaging
better, more
efficient
advertising
customer
relationships
self service
channels
emails, twitter,
word of mouth,
pr, sxsw, growth
hacking, sales
team
customer
segments
ad agencies
contractors
companies with
crm databases
costs
employees
heroku
revenues
selling the service
templates
14. exit
Email and landing page
serving could achieve
significant profit in its
own right
Anticipating a pivot
from ad serving to
using our massive data
on sales
Opportunities for
acquisition at Google,
Yahoo, Bing, Salesforce