Conclusion
- Big Data jeopardizes century-old models (e.g. risk mutualization in insurance sector) : for the Greater Good or for profit seeking purposes ?
- Algorithms are tools like any other in the Human species' evolution : we need to harness it.
- Currently cowboys' behaviors are common : not sustainable
- "Black Box" algorithms are highly dangerous : more transparency (and Ethics) needed
- The challenge of the coming years is not Big Data anymore ... it's privacy
A business perspective on Big Data trends
Happy to answer questions and welcome inquiries to speak about Big Data
- Big Data gives the impression to make any marketer's dream come true: one-to-one personalized relationship
- Data collection has become easier and cheaper (IoT)
- More data from the same sources is not always meaningful
- Real-time processing is an important evolution made possible by mutualization (AWS, Azure, ...)
- Sociological approaches to behaviors worryingly disappearing : Big Data doesn't explain causation ! Qualitative approach still needed.
Big Data : business Trends and Challenges
- New data sources are key for better predictions but may be more intrusive
- Firms will increasingly look to enrich their data with other businesses (e.g. banks + retailers, banks and social networks, ... )
Big Data in a nutshell
2. The future of insurance
- Some industries more advanced than others
- Retailers, banks, insurances, telecom are heavy data producers and users
- Data often used for profits
- 2 types of (big) data usages : keep costs down vs. increase revenues
- Adaptive pricing ("Pay for your own risk")
- insurance companies need more "user-centric" data
- Example 1 (real) : car insurance
- Example 2 (fictitious ?) : health insurance
Increase revenues = sell more
Keep costs down = control risks
Profile risk evaluation (insurances)
Mortgage risk evaluation
Selfscan control
forecasting
Churn prediction
Cross- and upselling
Behavior-based pricing
Couponning
- Insurance companies will investigate new (intrusive) forms of data collection
- In Belgium, smart boxes will not be adopted
- Data currently collected does NOT improve significantly predictions
- market structure impedes adoption (4m units in Italy vs. 30k in Belgium)
- New variables need to be included in model (e.g. health-related with wearables)
1. The future of retail
- Parallels between retail evolutions & Big Data
- 3 major evolutions in retail landscape
Retail evolutions
data capture of in-house consumption: replenishment and forecasting
Getting customers away from competitors
understanding total purchases to make consumer switch (coupons)
order automation to get rid of temptations (Nespresso)
Dr. Pierre-Nicolas Schwab
Big Data manager RTBF
Founder IntoTheMinds
Researcher ULB
- Data inflation seems to be the only way to go
- New variables increasingly captured (time of consumption for instance) giving new possibilities (forecasting) but also increasing firms' bargaining power over consumers
- New data collection methods threaten established business models
- Real-time processing will enable new customer experiences (personalized coupons when you enter the store ... not after you have shopped)