CASES SA
CAPITAL ONE (CAPGEMINI)
- Business challenge
o Lack of differentiation and weak customer targeting in the credit market
o Goal was to transform a data-driven digital bank that could innovate quickly, personalize services
and outperform competitors
- Data used
o customer behavior, IT systems, internal analysis, location, large-scale experimentation
- Methods
o Advanced analytics & statistical modeling
o Mass A/B testing
- Results
o Strong predictive capabilities and CA through data-driven decisions
o Profit before tax CAGR of 10.78% outpacing US banks
o Market share increase
o Higher customer retention and lower acquisition costs
- Critical view
o Risk about transparency and ethics and privacy
o Long term scalability
CASE VIDEO GAME INDUSTRY (BRYDON & GEMINO, 2008)
- Business challenge
o Predicting market success of videogames
- Data used
o Sales data, product characteristics (limited and incomplete)
- Methods
o Classification trees (inductive), combined with decision trees (deductive)
- Results
o Classification models inform better managerial, decision trees improve controllability of decision
- Critical view
o Data quality is weak (missing values), limiting predictive accuracy
o Path dependency
o Real option
CAPITAL ONE (CAPGEMINI)
- Business challenge
o Lack of differentiation and weak customer targeting in the credit market
o Goal was to transform a data-driven digital bank that could innovate quickly, personalize services
and outperform competitors
- Data used
o customer behavior, IT systems, internal analysis, location, large-scale experimentation
- Methods
o Advanced analytics & statistical modeling
o Mass A/B testing
- Results
o Strong predictive capabilities and CA through data-driven decisions
o Profit before tax CAGR of 10.78% outpacing US banks
o Market share increase
o Higher customer retention and lower acquisition costs
- Critical view
o Risk about transparency and ethics and privacy
o Long term scalability
CASE VIDEO GAME INDUSTRY (BRYDON & GEMINO, 2008)
- Business challenge
o Predicting market success of videogames
- Data used
o Sales data, product characteristics (limited and incomplete)
- Methods
o Classification trees (inductive), combined with decision trees (deductive)
- Results
o Classification models inform better managerial, decision trees improve controllability of decision
- Critical view
o Data quality is weak (missing values), limiting predictive accuracy
o Path dependency
o Real option
