201208 NAMIC Operations: Analytics: A Cross-Functional Solution to Information Overload

Analytics: A Cross-Functional
Solution to Information Overload

Presented to:
2012 NAMIC Operations Seminar
Charleston, S.C.
August 23, 2012

Presented by:
Steve Callahan, CMC
Practice Director
Robert E. Nolan Company

Today’s Discussion

 Why Analytics
 Recent Survey Results
 Case Studies
 Final Thoughts
 Questions

Analytics: A Cross-Functional Solution to Information Overload ©
Robert E. Nolan Company | Page 2

Analytics One of Top 5 Technology Topics


How Do the Top 5 Compare Today


Business Results Driven Priorities

 Admin / Legacy System Functional, Flexible, Supportable, Reliable

 Analytics / BI Optimal Information Driven Decisions

 Mobile Computing Service Delivered in Customer’s Hands

 Big Data Incorporate All Data into Decisions

 Cloud Computing Evaluate Universal Access / Variable Cost

Most Companies Have or Are Addressing Legacy Systems
Next Step is Analyzing Discrete Data and Focusing Decisions

2011 Survey: Most Decisions Rely on Experience and “Gut”

82% 60%

Robert E Nolan Company Executive Survey, 2011 6Robert E. Nolan Company
©
| Page 6
Analytics: A Cross-Functional Solution to Information Overload
June 2012

Retrospective Based on Experience versus Predictive

MY REAL-TIME
ANALYSIS TELLS ME
IT’S SMOOTH SAILING

7Robert E. Nolan Company
©
| Page 7
June 2012

More Informed Decisions Improves ROI

IDC Research
Leveraging the Foundations of Wisdom:
The Financial Impact of Business Analytics (© IDC) showed
30% tremendous
25% gains –

20% 10 Years Ago
15% (2002)

10% Median ROI:
5% Predictive: 145%
NonPredictive 89%
0%
1-50% 51-100% 101-500% 501-1000% >1,000%

2011 Survey: Leadership Decisions Moving To Data Driven


2011 Survey: Analytics Used Across Wider Variety of Areas


Analytics Capability Maturity Evolution
5
Tools and data rapidly evolving
Most 4 Continuous improvement loop
Companies
Here 3 Direct Link to Decision Making
Applied Across the Organization

Advanced Analytics Tools
2
Integrated Data
Limited Link to Decision Making
1 QA Standards Applied
Basic Analytical Tools
Limited Data Integration
Basic Data
Minimal QA

A Different View: From Reporting to Data Innovation


Relative Adoption by LOB
25.00%

20.00%

15.00%

10.00%
Predictive
5.00%
Retrospective
0.00%


Top Line Revenue is Improved As Well
Carriers effectively using predictive analytics achieved:
•1% improvement in profit margin
•6% improvement in year on year customer retention

Carriers not fully using predictive analytics:
•Dropped 2% in profit margins
•Decreased 1% in year on year customer retention

Higher on the Capability Maturity Curve = Better Results:
•Top 20% : 27% Year on Year Growth in Revenue
•Middle 50% : 12% Year on Year Growth in Revenue
•Bottom 30% : 1% Year on Year Growth in Revenue

Summary Key Benefits of Analytics

 Gain deeper, more relevant business insights to inform decisions

 Bring predictive analysis and regression modeling to entire organization

 Use analytics to identify and determine options for industry challenges

 Effectively and proactively manage risks

 Strengthen data governance at each level of the organization

 Reduce costs through more accurate, data-driven decision-making

 Use analytic capabilities and outcomes for change management

 Create a culture that thrives on fact-based decisions versus “gut”

Analytics: A Cross-Functional Solution to Information Overload © 15

Yet Companies Struggle to Implement

Most frequent reasons companies struggle with analytic initiatives:
•Too much management, not enough leadership
•Limited support and buy-in at multiple levels within the organization
•No guiding purpose or vision for people to rally around
•Overemphasis on technology implementation/success criteria
•Business benefits are too fuzzy to articulate and communicate clearly
•No consistent communication or messaging to stakeholders
•Poor identification of stakeholders and influencing factors
•Compensation structures and incentives not aligned

Robert E Nolan Company Executive Survey, 2011 © 16
June 2012

And the Barriers Are Diverse
Survey Comments on Barriers to Growth in Use of Analytics
“Resistance comes from most experienced, those requiring 100% accuracy”
“Access to critical data that is not captured in the system but is on paper”
“Getting away from tribalism, managing by anecdote and subjective decisions”
“Availability of resources and the money necessary to do it right”
“Data is spread all over and difficult to integrate or consolidate”
“Privacy will become a major issue as external data sources drive decisions”

Robert E Nolan Company Executive Survey, 2011 © 17
June 2012

With Opinions Varying Greatly
“The importance placed on analytics will grow, however there will be
a disproportionate reliance placed on results, until management
learns that garbage in/garbage out continues to cast its shadow.“

“It really doesn’t matter as most data currently produced comprises
the basis for most uses necessary. Advanced techniques do not
therefore produce ‘advanced’ data - the numbers are the numbers
no matter how produced. Indeed, give me a room full of ladies in
green eyeshades and Marchant calculators and maybe a punch
card reader or two and I could be perfectly happy with managing
the business, no matter how complex.“

“Those companies that do not embrace technology and analytics will
be left behind in the dust of those companies that do. “

©

Common Barriers to Using Analytics


Analytics is One Tool in the
Retroactive
Entire Product Life Cycle Production
Research Operations
Internal and
Feedback
External Data Predictive
Acquisition and Analysis
Cleansing
A clear but specific Exception
Conversion and
Formatting
vision enables a Handling and
Workflow
manageable project
Rule and Rate
Client and
Account Centric
structure with iterative Automation and
Integration deliveries. Enforcement

Population External Data
Analysis and Acquisition
Segmentation
Product Development & Assembly
Underwriting Rule and Rate Product / Forms Rule and Rate
Pattern and Data Rate Design and
Rules Design Model Prediction Creation Production
Analysis Modeling
and Modeling and Optimization and Assembly Roll Out

Iterative Process

Transforming Product Development
Improved use of analytics can be organized so that the key plan
areas can be developed along three parallel tracks.

Data Production
Product Processing
Evaluation
Development Integration
and Analysis

 Current Data  Modeling and  Rates and Rules
Evaluation Analytic Tools Integration
 Data Cleansing and  Rules Engines  Predictive Analysis
Alignment  Rating Engines Models
 Market Segmentation  Product  Workflow and
Assembly
Analysis Exception Process
 Process
 External Data  Legacy Integration
Integration and
Aggregation Management  “Dashboard”
 Trend Analysis Tools  Performance Development
and
and Modeling Scalability
©

Case Study A: Prospect Scoring
Scoring of prospects based on conversion and
Psycho- value, marketing strategy developed to match
graphic
Data Potential Value
Text High value, High value, High value,
Low Medium High
Data conversion, conversion, conversion,
2nd Priority Top Priority Top priority
Predictive Potential Good value, Good value, Good value,
Web Analysis Future Low Medium High
Log conversion, conversion, conversion,
and Value of Low Priority 2nd Priority Top Priority
Data Modeling Customer
Low value, Low value, Low value,
Low Medium High
Survey conversion, conversion, conversion,
Data Low Priority Low Priority 2nd Priority

Purchased Low Medium High
Data Propensity to Convert
©

Case Study B: Agency Management
60% of customers would switch carriers if so advised by their agent.
(Source: JD Power & Associates)

33%+ of agents are likely to change insurance carriers.
(Source: National Underwriter and Deloitte)

Insurers better manage their agents achieve competitive advantage.
New customers have high acquisition costs, retaining one more profitable.
New agents have high acquisition expenses and pose a greater risk of
inferior retention rates, resulting in lower profits.
Monitoring effectiveness of agents provide early warning that an agent may
be about to leave, triggering action and market differentiation.
Predictive scorecards tie traditional features like traffic lights and
speedometers to powerful analytics.
 Dashboard visuals provided at-a-glance access to the current status of new
KPIs, with automatic alerts for underperforming objectives and strategies.

Implemented an agency dashboard based on new KPI’s that were
modeled with a predictive analytics tool.
Analytics: A Cross-Functional Solution to Information Overload 23
©
June 2012

Case Study B: Agency Management

© 24

Case Study C: Loss based Pricing
Territory average loss ratios
generate prices that are too high
for some and too low for others.
$812.50

Detailed risk
$438.00 analytics generate
more accurate loss
cost estimates by
$1187.00
discrete segments
of business.

Result: More equitable and competitive risk adjusted pricing.
ISO Price Analyzer Tool used for graphics
© 25

Case Study D: Retention Strategies
Step 1: Determine Life time Value
Post Purchase
Activity –
Increases in Future
predictive value Value
over time as
behavioral
patterns
develop
Predictive
Analysis
Customer behavior
shifts focus from
Time of current to future
Purchase value
Demographics Current
-Loses predictive Value
value over time
as relevance is
superseded by
inforce behaviors 26
©

Step 2: Predict Potential Lapse
Source of Business
influences lapse tendencies
based on channel behaviors

Predictive
Analysis –
Model
Channel
Transaction behavior and
influences lapse tendencies
per consumer behaviors Consumer
Behaviors

© 27

Step 3: Develop Strategy Matrix
Match effort to
risk and value –
•High value low
risk gets medium
effort, save money
on retaining low
risk customers
•Low value
customers get low
cost efforts
across the board

•Targeted high
efforts on high
value / high risk

© 28

Case Study E: Claims Fraud
• About 10% of all insurance claims are fraudulent.
• Annual fraud losses for P&C industry total $30B in US alone.
• Need to detect unknown patterns of financial fraud.
• Keep track of new fraud schemes.
• Unsure exactly what to look for.

• Rules: Captures fraud on known patterns previously used
Ex: Two claims in different time zones within short window
• Anomaly Detection: Detect unknown patterns (ind & aggr)
Ex: Statistics (mean, std dev, uni/multivariates, regression)
• Advanced Analytics: Detect complex patterns
Ex: Knowledge discovery, data mining, predictive assessment
• Social Network Analytics: Determine associative links
Ex: Knowledge discovery via associative link analysis (entity
map)

©

Automated Fraud Detection Points
Re-estimate duration
Prioritized investigation SIU Reassess loss reserving
Focus on organized Prioritize resources
fraud Fraudulent rescoring
Minimize claim padding Review litigation
Reduce false positives propensity
Fraud Referrals Fraud Referrals

FNOL Assign Evaluate Update Close
Claim Claim Claim Claim

Fast Track
Predict duration Claim Cross-sell options for
Forecast loss reserves Negotiate / satisfied customer
Customer retention
Optimize fast track Initiate
claims
Prioritize resources Services Identify salvage and
Fraudulent scoring subrogation opportunities
Initiate Indicate deviations
Litigation propensity Reports on overrides
Settleme
© 30
nt

Claims Analytics:
Fraud Red Flag Dashboard

Analytics: A Cross-Functional Solution to Information OverloadCourtesy of Attensity © 31
June 2012

Other Brief Claims Examples

Optimized Claims Adjudication process.
Using data mining to cluster and group claims by loss characteristics
(such as loss type, location and time of loss, etc.).
Claims scored, prioritized and assigned by experience and loss type.
Higher quality, more consistent, and faster claims handling.

Adjuster Effectiveness Measurement.
Adjusters typically evaluated based on an open/closed claims ratio.
Analytics create key performance indicator (KPI) reports based on
customer satisfaction, overridden settlements and other metrics.

Claims using attorneys often 2X settlement and expenses.
Analytics help determine which claims are likely to result in litigation.
Assign to senior adjusters to settle sooner and for lower amounts.

©

Case Study G:
Life Underwriting via App + Social Data
Second child born last year Actively
High investment risk tolerance pursue for
Lived in home 2 years issuance of a
Owns home preferred
Commuting distance 1 mile policy without
Reads Design and Travel Magazines requiring
Urban single cluster fluids or
Premium bank card medical
Good financial indicators records.
Active lifestyle: Run, Bike, Tennis, Use strong
Aerobics
Health food choices retention
Little to no television consumption tactics.

Life UW using a GLM predictive model to assess risk:
Use info on app plus social data, No fluids or files
Integrate 3rd party publicly available information.
© 33

Case Study:
Life Underwriting via App + Social Data

Current residence four years Do not send
Lived in same hometown 15 years offers. Do not
Currently renting pursue
Commuting distance 45 miles
Works as administrative assistant aggressive
Divorced with no children retention
Foreclosure/bankruptcy indicators strategies. If
Avid book reader applies,
Fast food purchaser pursue
Purchases diet, weight loss equipment additional
Walks for health medical
High television consumption records and
Low regional economic growth
tests.
Light wine drinker
In a test over 30,000 applicants, behavioral and lifestyle
factors provided 37% of the risk assessment influence and
performed as well as additional, more intrusive medical
tests and fluids.
©

Social Analytics:
Customer Engagement Dashboard
 Automatically monitor
social conversations
 Filter out irrelevant
posts
 Analyze posts to extract
key insights
 Engage customers with
proactive outreach
 Improve experience
customers are having
on the site
 Improve brand image
and emphasize
business legitimacy


Social Analytics:
Conversation Sentiment Tracking

Courtesy of Attensity

Social Analytics:
Website Sentiment by LOB

Courtesy of Attensity

Available Third Party Data is Extensive
Third party marketing datasets are often used to develop the predictive
models, they include over 3,000 fields of data, contain no PHI, are not
subject to FCRA requirements, and do not require signature authority.
The match rate with insured’s is typically around 95% based only on
name and address. Third party marketing data includes:
Survey Data:
•Self-reported information
Rewards programs •Contains many lifestyle elements
Magazine subscriptions Basic demographics
Email lists •Age, sex, number & ages of kids, marital
Websites status
Grocery store cards •Occupation categories, education level
Book store cards Financial information
Public records •Income level, net worth, savings, investments
•Home value, mortgage value, credit card info
Lifestyle data
•Activity: running, golf, tennis, biking, hiking,
etc. Robert E. Nolan Company | Page 41
©

•Inactivity: TV, computers, video games,

Social Analytics:
Overall Sentiment Ratings Dashboard


Closing Notes: Bloomberg Qualitative Research Findings

 Analytics rapidly advancing past “emerging stage”
 Organizations proceeding cautiously in adoption
 Business experience driving factor in decision making
 Analytics for big issues, focus on improving bottom line

 Key adoption challenges:
– Data quality, acquisition, integration
– Many carriers lack proper analytical talent
– Culture critical
– Executive sponsorship key
– Start small, work big


3 Guidelines to Implementing Analytics

1. Have executive sponsored roadmap clearly outlining:
 What resources will be needed for how long,
 Where and when predictive analytics will be used,
 Which tools will be used, and
 How will success be measured.

1. Use data that is comprehensive, accurate, and current.
 Not necessarily 100%, some have used only 70%.
 Must be representative.

§ Staff with talented and engaged people.
1. Completely understand business problem, proficient with analytics.
2. Every person does not have to meet both qualification.
3. A team can be used with some business and some analytics experts.

©
June 2012

Questions?

Thank You...
Steve Callahan, CMC
Practice Director

steve_callahan@renolan.com
www.linkedin.com/in/stevenmcallahan
@stevenmcallahan
(206) 619-7740

201208 NAMIC Operations: Analytics: A Cross-Functional Solution to Information Overload

More Related Content

What's hot (18)

Similar to 201208 NAMIC Operations: Analytics: A Cross-Functional Solution to Information Overload (20)

More from Steven Callahan (20)

Recently uploaded (20)

201208 NAMIC Operations: Analytics: A Cross-Functional Solution to Information Overload