The Promise and Peril of Big DataOr Trust but VerifyMid-Atlantic Marketing SummitApril 24, 2014

Who Is Enroll America?

• Landmark Patient Protection and Affordable Care Act passed into law in 2010.

• In 2011 Enroll America was founded to ensure that all Americans enroll in, and retain, health coverage.

• To achieve their mission Enroll first had to find the uninsured.

• Enter big data…

Building a model

Started by thinking about the individual:

• Used public and commercial data

• GeographyState >> City >> ZIP >> Household

• Demographics Race, Age, Marital status Purchasing habits, Medical history

Surveyed a sample and built the model:

• Surveyed 10,000+ people • Asked insurance status • Individual likelihood of

uninsured score• Strong predictors

Age, gender, race, income, voting history

30

44

52 59

37

32

35

47

41

44

Micro-targeting down to the household possible

How big data informed decisions

What we learned:

• Two-thirds of the uninsured live in 12 states

•  Half live in just 114 counties(<4% of all counties)

• A lot of education needed• The more informed, the more

likely to sign up•  Required a longer, multi-

touch approach

What we did:

• 10 target states, incl ZIPs• Field teams in each state• Started knocking on doors of

likely uninsured• The model held up early on• Where else can we apply the

model?

Uninsured microtargeting media strategy

• $5M campaign with a mix of display, mobile, search, social

• Test multiple tactics with a heavy emphasis on BIG DATA

• Targeted select ZIPs in 10 target states, same as Field

• Maximize the number of email addresses acquired at lowest possible CPA

• Demo-geo-targeting and re-targeting/behavioral attribute targeting (i.e. uninsured, intent to enroll)

AUGUST SEPTEMBER OCTOBER 1

Research revealed core motivations:

• Personal stories• Financial security• Just the facts• Affordability• Cuz mom says so

Responsive website and digital ad campaign

Countdown to October 1

Research-driven online experience

The site:• Wordpress platform• Responsive design• Custom subsidy calculator

How we built it:• Agile style approach to

design and development• Functional prototypes

instead of static page comps

• Iterative reviews• Designers and developers

often collaborating side by side

Research-driven creative

Last minute rush to produce TV

• Last week of September

• Seven interviews

• Two TV spots highlighting the personal impact of affordable health insurance

• The spots launched in target markets in time for the beginning of open enrollment

Not the results we expected

• Personal stories didn’t resonate the way we thought they would. People cared more about how insurance would impact them personally as well as its affordability

• After a few weeks, we realized our budget was horribly misaligned— forgot to prioritize basic Direct Response tactics

• Let the promise of Big Data lead our strategy instead of informing it

• Confused voter and consumer:ROI justifies the political campaign investment but that’s rarely true in consumer marketing

But why didn’t it work?

Big Data is expensive and (for us) too targeted

• We had a relatively small media budget given the size of the opportunity

• Big Data media had to work a lot harder relative to lower cost alternatives

• Didn’t reach influencers

Recalibration and the results

• Higher volume at lower costs

• Find the low hanging fruit

• Display & Lead Gen: Simple geo (ZIP) and demo (age, gender, income) targeting

• Paid search: ZIP targeting • Paid social: Targeting

lookalike audiences

• Retargeting of site visitors

• 2 million unique site visitors

• 1 million consumer email addresses in 10 states

• Connected 380,000 people to their marketplace or application assistance via digital only

Optimization worked

1-Oct

8-Oct

15-O

ct

22-O

ct

29-O

ct

5-Nov

12-N

ov

19-N

ov

26-N

ov

3-Dec

10-D

ec

17-D

ec

24-D

ec

31-D

ec7-

Jan

14-Ja

n

21-Ja

n

28-Ja

n

4-Fe

b

11-F

eb

18-F

eb

25-F

eb

4-Mar

11-M

ar

18-M

ar

25-M

ar -

100,000

200,000

300,000

400,000

500,000

600,000

700,000

800,000

900,000

1,000,000

$-

$20

$40

$60

$80

$100

$120

Consum

er

Em

ails A

cquir

ed

Eff

ecti

ve C

PA

Campaign CPA

Consumers Emails

Recalibrated the website tooFrom stories to information and assistance

Takeaways

Trust but verify

Don’t let strategy be dictated by the newest, flashiest thing

Is a Big Data driven media strategy right for you?

• Goals & Budget

• Accessibility of your target audience

• Does the marginal return justify the cost

Adam StalkerNational Digital Director | Enroll America

Email: [email protected]: @adamtstalkerLinkedIn: linkedin.com/in/adamstalker

Leigh George, PhDVice President | Social@Ogilvy

Email: [email protected]: @leighgeorgeLinkedIn: linkedin.com/in/leighgeorge

Connect with us