Invited talk at Stanford CSEE392I (Seminar on Trends in Computing and Communications) April 24, 2014. Covered three topics: (1) Data science at LinkedIn. (2) Mobile data science — how is it different, challenges and opportunities. Examples of how data science impacts business and product decisions. (3) Mobile today, and LinkedIn's mobile story.

1. A Perspective from the
Intersection of Data Science,
Mobility, and Mobile Devices
Stanford, EE392I April 24, 2014
Yael Garten
Manager, Data Science for Mobile Products @ LinkedIn

2. Let’s talk today about:
2
1. Data science at LinkedIn: driving product and business insights
2. Mobile data science:What’s different, what’s important?
3. Mobile today. LinkedIn’s mobile story.

3. Our mission: Connect the world’s professionals
to make them more productive and successful
3
www.linkedin.com

4. Whatever you do, wherever you are
4
Connecting

5. LinkedIn at a Glance
5
• Founded in 2003
• 300M+ members
• 2 new signups per second
• Executives from all Fortune 500 companies
• 4B annual people searches
• Over 200 countries territories
• 20 different languages

6. Your Professional Identity
Amazing dataset that we can slice and dice.
By seniority, by job function – we can ask
many interesting questions.

7. So what data do we have?
Content
Social Graph
Behavior,
Engagement
Interest and Intent:
• Queries
• PageViews
(different types)
• Actions
(invites sent,
articles clicked)
Mechanism
• Device
• Channel

8. What does the Data Science team at LinkedIn
do with the data?
8
• Product and Business Insights
Measure, understand and improve products
Drive strategy
• Build Data Products
• Extract Insights we Share Externally

9. Knowledge: What’s going on?

10. If you can’t measure it, you can’t fix it.
Measure everything. wisely.
10
Know thyself:What’s going on?
In the form of reporting, knowing the numbers, understanding usage of products,
data patterns, segments of users, tracking growth and health of the ecosystem.
What
are
the
right
success
metrics
that
we
can
define?
What to measure?

11. Insights: what should we build or improve?

12. What to build? Rethinking our Mobile App
12
What do people on this page?
Where does the opportunity lie?
What drives them in?
Where do they go next?
Where can we reduce friction?
How many drop off?
What is the stickiest product?
What works, what doesn’t?

13. Data to inform product design of the iPad app:
iPad
12am
1am
2am
3am
4am
5am
6am
7am
8am
9am
10am
11am
12pm
1pm
2pm
3pm
5pm
6pm
7pm
9pm
10pm
11pm
Desktop
At what times in the day are people using different devices?

14. What to ship? Controlled online experiments
(aka A/B testing)
14
1. What are the right success metrics?
2. How do we track (instrument) this? Data quality
3. Experimental Design
4. Analysis, Interpretation, Recommendation

15. Wisdom: What’s the next needle mover?
dataà knowledge à insights à wisdom

16. Strategic Analyses:
Using data to drive the business via rigorous analytical framework
• What is the value of an action that a user takes on the site?
• What early behavior on the site is predictive of future
engagement?
• How does mobile usage compare to and impact desktop
site engagement?
• What is the value of a (mobile) user?
• How do people use LinkedIn organically? Drivers and flows

17. What does the Data Science team at LinkedIn
do with the data?
17
• Product and Business Insights
• Build Data Products
• Extract Insights we Share Externally

18. Data Products on your LinkedIn homepage
18

19. Alumni Data Product
19

20. The Skills Data Product
20

21. How do we do it?
21
Extract

22. Skills:Assigning Skills to People
22

27. What does the Data Science team at LinkedIn
do with the data?
27
• Product and Business Insights
• Build Data Products
• Extract Insights we Share Externally

28. Data Stories: Growing and Shrinking Industries

29. Where did all the people go from the collapsed
financial institutions in 2008?

30. The 10 Most attractive start ups to Bay Area Engineers

31. Big Data that scales.
What Technologies do we use?
31

32. Let’s talk today about:
32
1. Data science at LinkedIn: driving product and business insights
2. Mobile data science: What’s different, what’s important?
3. Mobile today. LinkedIn’s mobile story.

33. Connecting: Whatever you do, wherever you are

34. Ubiquitous Computing
Constant access to information and products

39. Price
Point
Screen
Size
Connectivity
App
or
Web
Operating
System
Location
App
Version

40. 1. Recognize the complexity
2. Invest in tracking
3. Leverage the data
Control the chaos… Delight users

41. Price
Point
Screen
Size
Connectivity
App
or
Web
Operating
System
Location
App
Version
Screen
Size
PhoneTablet
Phablet

42. Price
Point
Screen
Size
Connectivity
App
or
Web
Operating
System
Location
App
Version
App
Version
1.0 2.0

43. App
Version
Price
Point
Screen
Size
Connectivity
Operating
System
Location
App
or
Web
App
or
Web
• Can A/B test (+) • Limited A/B capabilities (-)
• Transactional • Organic, richer experience

44. App
or
Web
App
Version
Price
Point
Screen
Size
Connectivity
Location
Operating
System
Operating
System
3X more reach

45. Operating
System
App
or
Web
App
Version
Price
Point
Screen
Size
Connectivity
LocationLocation

46. 1. Recognize the complexity
2. Invest in tracking
3. Leverage the data
Control the chaos… Delight users.

47. Data to inform product design of our iPad app:
iPad
12am
1am
2am
3am
4am
5am
6am
7am
8am
9am
10am
11am
12pm
1pm
2pm
3pm
5pm
6pm
7pm
9pm
10pm
11pm
Desktop
From coffee to couch and everything in between

48. Follow the trends

49. Landscape or Portrait ?
45%
65%
55%
35%iPad
iPad mini

50. 50% and growing. Optimize your email
2X higher CTR

51. The data says: Simplify
1.0 2.0
6%
8%
10%
12%
14%
16%
18%
20%
1/11
2/11
3/11
4/11
5/11
6/11
7/11
8/11
9/11
10/11
11/11
12/11
1/12
2/12
Mobile % of Unique Visiting Members
1.0 2.0

52. Leverage learnings across platforms

53. Let’s talk today about:
53
1. Data science at LinkedIn: driving product and business insights
2. Mobile data science:What’s different, what’s important?
3. Mobile today. LinkedIn’s mobile story.

54. time
% of traffic /
revenue
mobile moment

56. Q1 2011

57. Q3-2013

58. Q1-2014

59. This year

60. 25%
44%
Singapore
Netherlands
Turkey50%
50%
50%
45%
45%
United Arab Emirates
45%
Sweden
Australia 45%
Denmark
38%
32%
UK
US
Brazil
India
End of 2013

61. Always on with mobile
3am
6am
9am
12pm
3pm
6pm
9pm

62. Sources: IDC March 2012, Gartner April 2013,
2012 2016
Smart phone
PC
Tablet
Per day…
= 100M
Every day there are 4.5x new mobile devices sold than babies born!
378K
1.3M
371K
Mobile is changing the world

63. Mobilize systems.
Different architecture. New considerations.
Feature parity? Same designs?
Resourcing.
Mobilize. Mobile First.
Promotions
Tracking
Internationalization
Deep Linking
Analytics
Automation
Personalization
A/B
Experimentation
Page Load

64. LinkedIn App
So we have one great app. Segment use cases.

65. Contacts
Recruiter
Pulse
LinkedIn App
Multi-app. Cross promote.
Slideshare

66. Mobile first. Platform first

67. Organic vs. Transactional

68. Opportunity.
Essential.
Transformational.
Connecting.

69. Thank you.
http://www.linkedin.com/in/yaelgarten
@yaelgarten We’re
hiring!