1. Data Driven Decision Making
Hatim Maskawala
October 19, 2017

2. How many of you believe that investing in Social Media is the way forward And how many companies have kept this as part of their strategy and have actually done something?

3. Facebook Analysis - Insurance

4. Facebook Analysis - Insurance

5. Facebook Analysis - Banks

6. Insurance vs Banks

7. Underlying Forces An explosion of data and contents
Reliance on external data
An explosion in computing power
The rise of information activism
Self service BI
There is now more and more data available with us
The processing power of systems have gone up tremendously
The new generation has been brought up using data and computers

8. Challenges No single view of customer
Ineffective customer segmentation
Multiple distribution channels
High volumes of fraud, wastage and abuse
Most insurers in MENA rely on policy admin systems for analytics
Underwriting vs Financial/Accident years
Data is segregated across systems
Manual data / excel files
Many calculations are in excel
Need for Customization Requests

9. Analytics Maturity

10. Descriptive Analytics
The rearview mirror is descriptive analytics. Its based on past data and shows where you have been.

11. Predictive
Analytics

12. Diagnostic
Analytics
The rearview mirror is descriptive analytics. Its based on past data and shows where you have been.

13. Maximizing analytic value
HIGH
Predictive
analytics
Analytic value
Diagnostic
analytics
Descriptive analytics
(What is likely to happen?)
LOW
(Why did it happen?)
Reporting and viz tools
BUSINESS USERS
DATA SCIENTISTS
Need to combine all in one tool

14. Difference Between Flat Reports and BI
Flat reports give too much information about what happened.
Used for transaction
BI tells WHY it is happening
Allows you to interact with the data
Answers your next question

15. Difference Between Flat Reports and BI
Criteria
Operational Reports
Business Intelligence
Business Function
Tactical
Analytical
Users
Staff
Executive Management, Analysts, LOB Heads
Data Sources
Core system Only
Any including core e.g. TPA Data
Mode of Operation
Query then Analyze
Analyze then query (on the go)
Type
Static
Dynamic
KPI Measurement No
Yes
Real Time
Near Real Time
Analysis Across Subject Areas
Yes e.g. Combined Ratio (as it has expenses, etc)

16. DAR Concept Dashboard High level KPIs
Something which gives you an overall picture in two minutes
Analysis
Has graphs and tables and tells the users “how” or “why”
Report
Transactional level information
Has more details like policy details, claim details

17. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

18. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

19. Product Development
Health/Motor insurance has become extremely data oriented
Pricing / decision making should be analytics driven. Always validate perceptions with data
Start at the top and drill down
Pricing factors and pricing tool
Use of pricing tool will provide standardised rates
Medical pricing is highly data dependent and here due to eclaims we have lots of data. Some general principles that we all should remember and this is not only for actuaries but for all:
Data driven. Don’t base your conclusions on perceptions. We cant give this cover what if customer goes to US and Canada and gets it treated there? We have given this cover in the past and how many have gone? Always look for evidence from data to validate or invalidate perceptions / hypothesis.
Don’t spend time on trivial things. Like in medical IP / OP and Mat generally make up 90% of the claims. So don’t spend hours evaluating alternative medications or optical etc.
Start at macro level. Like start with OP and then start breaking it down. Rather than worrying about 8000 CPT codes and pricing for each.
Data will give you facts. You need to have a plan to benefit from those opportunities. Like data might show that pharmacy costs fall by 40% when we apply 20% coinsurance. So the plan should be that we will give 30% discount for 20% coinsurance and push this as an option for all quotes.
The feedback loop is required in all cycles. Routinely do back testing and reassess your hypothesis in light of actual data after you had made the hypothesis.
In pricing do not forget the pareto 80/20 principle. Don’t spend majority of your time on pricing riders.

20. Deeper Pricing Analysis

21. Deeper Pricing Analysis

22. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

23. Sales - Client Client segmentation
By size – analyse your existing portfolio to find out which segments are profit making
By industry – generally difficult to evaluate. Can be done as part of data cleaning or one time exercise
Campaign optimization / lead generation
Where to cross sell
What to upsell

24. Sample Production

25. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

26. Sample Health Performance

27. Sample Motor Performance

28. Loss Ratio by Body Type and Sum Insured Bands

29. Performance Monitoring

30. Analysing Target Segments

31. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

32. Customer Service Customers are less loyal and more price conscious
Insurance managers are looking for more insights / advices
Clients wants transparency / more analytical reports
Need proper analytical tools which can handle big data
Previously it used to be just print cards. Now they want detailed reports.
Health premium is growing year on year and impacting their budgets. A company with 5,000 was paying AED 1,100 per member and now you quote them AED 1,400 per member. Their costs have increased by AED 1.5 mn. When you say loss ratio was bad their reaction is that its your job to maintain the loss ratio.
The larger customers have stopped trusting insurance companies and want constant report. Many time staff of insurance companies spend time in assembling the report and not assessing them.

33. Use of Systems for Fraud
Complex Probab-ilistic Rules
Calculate scores for claim
Calculate scores for patient based on past history
Determ-inistic Rules
Rule edits
Diagnosis / treatment rules
Statistical Reports
Distribution of claims by provider
By diagnosis / treatment
Standard Reports
Utilization per provider
Provider KPIs
Averages per diagnosis
Historical / Retrospective
Predictive / Real Time
Historical data / experience will help create new rules and setup complex probabilities.

34. Gap within Consultations

35. Normal vs C-Section

36. Rank Analysis on Drugs

37. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

38. Risk Accumulation Visually
We can show accumulation by geography also and the spots would be updated based on selections. The size and colour can represent different dimensions.

39. Risk Accumulation Reports

40. Using Analytics Product Development Production Performance Servicing
Reinsurance
Finance
Using Analytics

41. Automated Financial Reports

42. Financial Loss Ratio

43. Are you data driven
Data Denial
You distrust data and avoid using it
Data Indifferent
You don’t care about data and have no need for it
Data Informed
You use it only when it supports your opinions or decisions
Data Driven
You use it to shape and evaluate all your decisions
This is one time cost that reaps ongoing benefits. No use reinventing the wheel
To become a Data Driven Company it has to come from top

44. CONTACT 2107 SIT Towers, PO Box 341486, Dubai Silicon Oasis,
Dubai, UAE
Phone:
Fax: