2. About Me [@aj2z]
Data Science & Analytics Team
- Data ops (ETL, Munging etc)
- Feature Test & Learn
- Measure/Predict Customer Quality
- Acquisition Portfolio Mgmt
About us
Previously
- Postdoc Scientist at UCLA
(Bioinformatics, Graph theory, Viral Genomes,
Molecular Dynamics Simulations)
- PhD, Univ. Chicago
(Statistics of 3d defect formation in 2d films)
Age Strength Location
2.5 yr ~40 Santa Monica, CA
3M Series A (Dec 2013) & Growing Fast!
3. Next 30 mins
What we do at CARD.COM
Where’s our data?
Any new science? (some examples)
Some startup tips (if time allows)
Anyone heard of us before?
4. Our mission
A Core Financial Service for ALL
● Alternative to B&M Banking
● Mobile
● Affordable
● Fair
● Direct to consumer
5. … and also fun and fashionable
+
General Purpose Reloadable Prepaid Card
Customized with whatever inspires you!
Mobile App
Our Rewards: Inspiration in your wallet!
6. How does it work?
Let’s take a look:
Bank
& Service
Providers
Artwork
& Brands
Member
Transactions
DM
7. .. a closer look
Artwork
& Brands
Bank
Partner
Transactions
CARD.COM
Site / App
CARD.COM
AdTech Platform
APIs
RTB Ad Xchgs
CARD.COM
Analytics Platform
Members
Visitors
1
2
3
● 3 key types of data
● Platforms in PHP, R, Python
… let’s make some data
predict
deploy
learn
11. And now?
If I showed you this first?
We’ve just generated new IP!
12. Our (Data) Science Questions
- How likely is a person to cross over from a
design they like to a Financial Service?
- Is Star Trek a stronger Financial Brand
than Care Bears? …or a Cute Baby Elephant?
- Is a Spock CARD pulled out more at
a Starbucks than Care Bears?
Some examples next...
Brand
& Artwork
Marketing &
Human
Behavior
Finance
14. Design = Performance?Demographic + Other Variables
Design
De-Average, Learn, Grow
(aka the Capital One way)
Different strokes inspire
different folks!
There’s a CARD for every
demo! 2K+ Designs
15. But Design is not Everything
Human Behavior!
e.g.: When (Hour of Day)
you sign up indicates how
soon you’ll use your CARD
Time
Recommended reading: “Predictably Irrational” - Daniel Ariely
17. Our (Data) Science Vision
Things we CAN do (with bigger data):
- Financial Brand Equity Ratings
- Affinity Specific Behavioral Science
- FICO-like CARD.COM score for underbanked
- Provide pathway to creditworthiness
- Personalized promotions to members
All this while keeping our privacy commitments
Brand
& Artwork
Marketing &
Human
Behavior
Finance
18. Learning to learn
Transactions
CARD.COM
Site / App
CARD.COM
AdTech Platform
APIs
RTB Ad Xchgs
CARD.COM
Analytics Platform
Members
Visitors
1
2
3
predict
deploy
learn
Sample wide (better to know for
sure that something won’t work)
Know your feedback loop!
marketing > website > behavior
Focus resources where you see
evidence of large impact with small
investments
Understand business impact to
learning cadence… i.e. know your
efficient frontier early!
19. Infrastructure
Transactions
CARD.COM
Site / App
CARD.COM
AdTech Platform
APIs
RTB Ad Xchgs
CARD.COM
Analytics Platform
Members
Visitors
1
2
3
predict
deploy
learn
Will your data be:
-- Structured or Unstructured?
-- Growing Fast
-- Need to learn fast (RT?)
Structured data & R:
When analyses need more than
10M lines of data. It’s time to move
to distributed data and
computation
20. We’re hiring!
- Sr Statistical (R) Programmer
- Ad Tech Eng
- Full Stack Eng
- Drupal Eng
- Front-End Eng
and more...
card.com/careers
Our Mostly FOSS Toolbox