This document discusses building a talent analytics function and the speaker's journey in doing so at LinkedIn. Some key points:
- The speaker discusses guiding principles for crafting a talent analytics function, such as identifying its core purpose, staffing it with a blend of skills, focusing on questions that create business value, investing in maturity, and prioritizing efforts.
- Examples are provided of how the speaker focused on quick wins to build credibility, developed a framework to help stakeholders ask strategic questions, and is leveraging analytics across the talent lifecycle.
- Case examples show how the speaker used analytics to help with engineering hiring needs, improve operational planning and retention, and identify critical skills and attractive hiring regions.
4. More than 400M members in
over 200 countries around the
world
Professionals are signing up to
join LinkedIn at a rate of more
than 2 new members per
second.
There are over 39 million
students and recent college
graduates on LinkedIn. They are
LinkedIn's fastest-growing
demographic.
LinkedIn operates the
largest professional network on the internet
Regional membership
127M+ EMEA
96M+ Europe
78M+ Asia and the Pacific
15M+ Southeast Asia
7M+ DACH
17M+ MENA
55M+ LATAM
5. Total revenue advanced 37% year-on-year to $780M
5
2015 Q3 distribution by revenue stream
Premium
Subscriptions
18% - $138M
Marketing
Solutions
18% - $140M
Talent
Solutions
64% - $502M
Source: Linked.com 2015 Q3 Revenue
8. 8
We are a team of HR experts, data scientists,
and consultants who help HR solve problems
9. A center of excellence for Strategy & Analytics
9
Business
Operations
& Analytics
Talent
Analytics
BizOps
Analytics
Field
Sales
Online
Sales
Product
Strategy
Growth
Marketing
Strategy
Talent Analytics sits in Business Operations
10. 10
Talent analytics is getting bigger
Source: LinkedIn self-reported member data based on job title, start date, and company
13. 13
4,500companies have
employees focused
on talent analytics
43%
of the Fortune 1000
have talent analytics
teams
55%
of all talent analytics
functions have been
started within the last
5 years
70%
of all talent analytics
functions only have
1-2 team members
Source: LinkedIn, estimate based on member reported job titles and skills
15. Guiding principles for
crafting a talent analytics function
Identify the core purpose of your team
Staff your analytics team with a blend of skills
Focus on questions that create business value
Invest capacity to develop maturity
Prioritize your efforts to maximize impact
16. Problem-solving focus drives action and impact at LinkedIn
What is the core purpose of your team?
Research Problem Solving
Focused on methods
Long maturity curve
Business may not be
ready for results
May not help HR
become more strategic
Focused on impact
Short maturity curve
Helps the business
where they need it
Problem solving with HR
build business acumen
20. 20
Invest capacity to impact your organization
across all three maturity curves
Technological
Maturity
Stakeholder
Maturity
Analytical
Maturity
Integrated data and
automated reporting
Stakeholders ask strategic
questions and act on results
Predictive analytics
Integrated systems but
manual reporting
Stakeholders ask strategic
questions but don’t act
Advanced analytics
Cleaner data-entry
but no integrations
Stakeholders ask for
insights
Manual reporting with
a few deep dives
Messy data in
disparate systems
Stakeholders ask for reports
Manual reporting and
data cleaning1.
2.
3.
4.
21. 21
Prioritize your efforts and focus
where you can have an impactImpact
Org Readiness
Focus NowFocus Later
Avoid Automate
23. 23
We focused our efforts on quick wins
What we had Business
demand
Our solution
Analytics Infrastructure Reporting
Team resource allocation
Building the IT
infrastructure is a long
journey…
Reporting will consume
100% of capacity and
never be 100% accurate
Prioritize quick wins that
solve business problems to
build credibility
24. 24
We developed a framework to help our
stakeholders ask strategic business questions
Data - Oriented
Research questions need evidence in order to be answered
Questions that are not data-oriented usually need to be more specific
Objective
Testable
Specific
Great questions do not have the desired answer built-in
Make sure the answer to your question has the possibility of being positive
and negative
Great questions allow you to test your instinct
Sometimes the greatest learnings happen when the answer is unexpected
Specific questions focus the insights you are looking for and make them
easier to find
Broad questions can usually be broken down into several specific
questions
25. 25
We are leveraging analytics to be
the connective tissue across our lifecycle
Hiring
Data-driven recruiting using LinkedIn data
Monitoring org shapes to inform hiring strategy
Onboarding
Monitoring onboarding satisfaction
and time to productivity
Inclusion
Using analytics and natural
language processing to
inform inclusion strategies
Learning
Measuring the impact of learning
and identifying learning needs to
inform content creation strategy
Engagement
Measuring and monitoring
key drivers of engagement
Retention
Predicting attrition with
advanced analytics and
developing strategies to
retain top talent
Succession
Measuring succession risk and
creating strategies for reducing
readiness time
Performance
Evaluating programs based
on impact to performance
28. How many engineering recruiters do we need?
Forecasted hiring needs
# of Hires
Headcount forecasts
# of FTE
2015 2016 2017 2015 2016 2017
29. Are we hiring the right mix of people?
Org. shape has shifted over time
% of Engineering FTE
2013 20142013 2014
Senior+
Mid-Level
Entry-Level
Hiring has focused on entry level…
% of new hires
30. Partnered with HRBP and talent acquisition leads
to double mid-level and senior hires
# of new hires
1H 2014 1H 2015
Senior+
Mid-Level
Entry-Level
31. How do we improve operational
planning & better retain top talent?
32. 32
We built an algorithm to forecast
sales attrition using machine learning
Future
Attrition
Rate
Comp
Predictors
Background
Predictors
Role
Predictors
33. Need to focus on retention strategy before predictive analytics
33
The attrition forecast matched the number that
our HRBPs provided for headcount planning
Attrition forecast (Algorithm)
Q2 Q3 Q4
Attrition forecast (HRBPs)
Q2 Q3 Q4
34. Enables predictive analytics to be actionable and scalable
34
Developing retention playbook
and then scaling forecast process
X
XXXXX X
X
O O OO OO
O
O O
O
35. What are the most attractive
regions to hire software engineers?
36. Supply of software engineers in region
DemandforsoftwareengineersWhat is the supply and demand for software engineers?
Seattle
Chicago
Boston
Washington D.C.
New York
SF Bay
Phoenix
Houston
Denver
Philadelphia
Dallas
Toronto
Raleigh-Durham
Detroit
Montreal
Austin
San Diego
LA
Atlanta
Minneapolis
High
Low
Low High
37. LI Profile features
LI Profile
Features
Candidates from ATS
Machine learning
algorithm
Classification model Classified profiles
TrainPredict
Used profile data to classify
software engineers into tracks
38. Where do we find critical skills?
Engineering track concentration by region
Below average Above average
Systems
& Infra Apps Data Mobile
Eng
Manager
Eng
Services OpsIT
39. Conclusion
Talent Analytics is
growing but still maturing
Focus on questions that
create business value
Build a team that helps
HR solve problems
Leap frog the maturity
curve by finding quick wins