2. Why We Need to Make a Case for Analytics?
• What’s the net present value (NPV) of business analytics?
• No academic large-scale study exists that links analytics to firm
performance
– Brynjolffson et al. (2011) make an attempt
• Most of current evidence is based on case studies
– Most cases are published by analytics consultant/vendors/solution
providers
• IBM, SAS, SAP, Oracle, Teradata, etc.
3. The Circle of Mistrust
Marketing
Top Mgmt IT
Finance HR
4. Obstacles to Analytics Adoption
• Culture
• Lack of business sponsor
• Personal vs. organizational goals (short-/long-term)
• Few employees who question the data and make judgments
• Analytics skills are with too few employees
• Poor information management
• Lack of behavioral and anthropological training to IT
6. Analytics Usage and Organizational Type
Analytics Usage
Descriptive and All three types
High
Predictive
Descriptive Descriptive and
Low Predictive
Low High
Data Driven
7. Analytics Usage and Organizational Type
Analytics Usage
Descriptive and All three types
High
Predictive
Identify
Descriptive Descriptive and the
Low Predictive hindrance
Low High
Data Driven
8. Convincing Marketing Department
• What are the benefits you are looking for?
– Tracking customer satisfaction
– Assessing and increasing ad effectiveness
– Media planning
– Social media metrics
– Detecting trends
– Segmentation and positioning
– Something else…
• E.g., Wal-Mart and 9/11
9. Convincing Marketing Department
• Descriptive analytics
– Use external vendors on a small scale for demonstrations
• Predictive analytics
– Work with academic institutions to build models
• Run targeted experiments
– Exploit insights from predictive analytics
– Generate measurements for sales, market share, revenue
growth, customer satisfaction, churn rate, repeat
purchase, awareness, etc.
• Evaluate the effectiveness of analytics insights
10. Managing Human Resources
• Should you have an in-house analytics division?
– Corporate or SBU division?
• There are pitfalls to doing analytics in-house
– Demand for skilled analytics labor is extremely high
– Supply of skilled labor, unfortunately, is limited
• Other options
– Outsourcing
– Hiring young graduates and training them
– Training your existing employees
11. Outsourcing Analytics
• Outsourcing poses problems
– Data are sensitive
• Privacy issues
• Proprietary trade information
• Legal barriers
– Control on the analytics
• Quality
• Alignment of the objectives
• Coordination
12. Hiring and Training
• Hire young graduates from
– Engineering
– Economics
– Statistics
– Business management
• Train them on data analysis and/or business management
– Several online courses are available (e.g., Coursera)
– Tie up with local business schools (e.g., ESSEC, SMU)
13. Training Existing Employees
• Locate talent inside the organization
– Organization-wide search
– May have to overcome the departmental politics
– There may be a large variance in the skill levels
• Training alternatives
– Using in-house facilities for training
• Getting consultants and business schools to offer structured workshops
– Part-time business analytics programs
14. Getting to the ROI
• Analytics ROI at a staggering 10.66x (Nucleus Research 2011)
– Does it make sense?
• Survivorship bias (dolphins and 1,000 sailors), selection bias
– If that’s true, what’s stopping everyone from using analytics?
• ROI calculations are not straightforward
– Attributing cost savings, incremental profits, etc.
– What about the risk?
– More difficult with intangible benefits
17. Working with the IT
• Main challenges influenced by the culture
– Data capture/collection (e.g., MeritTrac)
– Data accessibility/sharing
– Organization-wide data integration
– Using real-time data dissemination
• In the initial stages
– Stick to available data formats
– Avoid merging multiple databases
– Avoid using too much unstructured data
18. Summary
• Making a case for analytics needs systematic approach
• In a non data-driven organization, there are many hurdles to
overcome
– ROI of analytics is one of the toughest one
• Each function (HR, marketing, etc.) may have their own concerns
for taking analytics route
19. Thank You
Prof. Ashwin Malshe
ESSEC Business School
malshe@essec.edu
Twitter: @ashwinmalshe