Big Data

Big Data An insurance business imperative

David Helmuth and Suresh Selvarangan Deloitte Consulting LLP Tuesday, April 8, 2014

Copyright © 2014 Deloitte Development LLC. All rights reserved. 2

Agenda

What is Big Data? 1

Where can Big Data bring value in Insurance? 2

3 The Journey to Big Data – steps to get there

What is Big Data?


Big Data is more than just growth in data volume. Big Data includes data that is unstructured, generated from non-traditional sources, and/or real-time – in addition to being large in volume.

Clarifying the definition

Type Size Examples

Admin Kilobytes Policy Administration, Claims Administration, Billing

CRM Megabytes Segmentation, Offer Details, Customer Touch Points, Support Contacts, Campaigns

Web Gigabytes Web Logs, Offer History, Dynamic Pricing, Affiliate Networks, Search Marketing, Behavioral Targeting, Dynamic Funnels

Big Data Terabytes Call Notes, Social Network, External Demographics, Business Data Feeds, Imagines, Audio, Video, Speech to Text, SMS

Size of Data

Big Data

Web

CRM

Admin C

ompl

exity

of D

ata

Illustrative


Creating value with the three V’s of big data

Velocity

Volume

Variety

Value

+

+

=


Identifying the types of big data in insurance

Big Data is highly prevalent within insurance, but remains underutilized.

Type Which V Why is it “Big”

Structured Claims Data

• Volume • On average, 30 years of historical claims data is stored

Claims Notes and Emails

• Variety • Notes and emails are considered unstructured data

Telematics • Volume • Velocity • Variety

• Streaming data is captured frequently (minutes); the sheer volume and velocity of the data poses challenges for traditional relationship systems

Weather Patterns and Seismic Data

• Volume, • Variety

• Data can be provided in relational format or using geo-spatial parameters

• Volume is a long-standing issue with analyzing weather patterns

Social Media • Volume • Velocity • Variety

• A large amount of social media data is generated • Data is transmitted in varying formats, all unstructured • Data is created at a rapid pace


Enterprises face the challenge and opportunity of storing and analyzing Big Data, respectively. Insurers, in particular, may expect to be challenged with: • Handling more than 10 TB of data

• Data with a changing structure or no structure at all

• Very high throughput systems: for example, in globally popular

websites with millions of concurrent users and thousands of queries per second

• Business requirements that differ from the relational database model: for example, swapping ACID (Atomicity, Consistency, Isolation, Durability) for BASE (Basically Available, Soft State, Eventually Consistent)

• Processing of machine learning queries that are inefficient or impossible to express using SQL

Implications for the enterprise

“Shift thinking from the old world where data was scarce to a world where business leaders demonstrate data fluency” - Forrester

“Information governance focus needs to shift away from more concrete, black and white issues centered on ‘truth’, toward more fluid shades of gray centered on ‘trust.’ ” - Gartner

“Enterprises can leverage the data influx to glean new insights – Big Data represents a largely untapped source of customer, product, and market intelligence” – IBM CIO Study


Big Data is supported and moved forward by a number of leading vendors throughout the ecosystem. In many cases, vendors play multiple roles and are continuing to evolve their technologies to meet changing market demands.

Taking a look at the big data ecosystem

Big Data File and Database Management

Big Data Integration

Big Data Analytics

Stream Processing

and Analysis

Appliances

BI/Data Visualization

Big Data Ecosystem

Where can Big Data Bring Value in Insurance?


Making big data and analytics top of mind

Insurance is a tough market. Big Data driven analytics can provide an edge in both day-to-day management decisions and in finding top line growth Industry-wide investment is turning analytics from an emerging issue into a core competency: • 82% of insurance executives

surveyed cite data and analytics as a key strategic priority

• 81% of insurance companies surveyed intend to increase spending on data initiatives in the coming years

• By 2016, it is estimated that 25% of large global companies will have adopted big data analytics for at least one security or fraud use case

Manage the Business • Gain visibility into operational performance • Improve statutory and market conduct reporting • Streamline core processes • Identify fraudulent claims

Doing Nothing is Not an Option

• Competitors and emerging startups are changing the industry, pushing analytics from an advanced capability to a core competency

Grow the Business • Create personalized pricing for customers • Build stronger distribution channels • Proactively cross- and up-sell current customers • Target opportunities in new geographies


Big Data

Telematics

Visualizations

Advanced Analytics

Claims Analytics

Applying big data in insurance


Streaming telematics data

Stream

• Latitude and longitude captured at predefined intervals during a trip, typically within 1–3 minutes intervals

• Average number of drivers, taking an average number of trips per day — volume grows large very quickly

Event

• Excess speed, acceleration, breaking, turns, and other values derived from sensors

• Volume of events more variable depending driving conditions and driver behavior

Trip Score • Relative score based on various factors captured during a trip

Event Stream

Trip Score


Identifying the risk


Integrating telematics data to gain insights

Traditional Data

When combined with policy / demographic factors / claims experience / driving history, you can really start to answer the important questions.

Premium leakage? Are my drivers driving more than the estimate provided during underwriting?

What is the relationship between the driving behavior and driving history?

Do my drivers with lower scores have higher claims?

Traditional data

Stream Event

Insight


Moving from basic to advanced analytics

Technologies around Big Data have emerged to handle exponentially growing volumes, improve velocity to support real-time analytics, and integrate a greater variety of internal and external data.

Big Data and Advanced Analytics Attributes

Reactive

Gigabytes

Weekly/monthly reporting

Predefined, structured data

Strategic

Terabytes

Weekly/monthly modeling

Expanded, still structured

Real-time

Petabytes

Real-time modeling

Dynamic, includes unstructured data

Decisions

Volume

Velocity

Variety

Yesterday Today Tomorrow

Foresight Hindsight Insight

Reporting Predictive modeling Big Data and advanced analytics

Hypothesis Testing


Evolving the actuarial process

More sophisticated customer digital interactions require and enable increasing insight into customer behavior. Organizations that leverage big data and advanced analytics can have accelerated growth through greater insight and understanding of their expanded customer interactions.

New Signals Predictive models to push and alert business of opportunities and insights

Profitable Growth Investments in analytics infrastructure and tools to improve insight into financial and market information

Hidden Insight Social media has given

rise to new ways to connect with customers

and uncover patterns

Computing Capacity Real-time

processing and data mining are now

possible

Volume and Variety

Global data volumes

continue to grow


Empowering actuaries with analytics

A well-constructed and maintained Enterprise Data Warehouse frees up actuaries, analytics modelers , and data scientists to focus on the data itself and their loss/predictive models.

Trying to figure out how to draw and integrate data from a number of different sources takes valuable time away from actuaries and IT.

Product Analysis Design

Consistent Simplified

Basis

Rationalized Model Inputs

Methodology Analysis

Policyholder Data

Lapse/PUP/Surrender Rates

Expenses

Mortality

Bond Rates

Unit Allocation Rates

Commission


Automate unstructured claims data with analytics

Analytics on unstructured data is a process to automate the interpretation of language to find the useful information hidden in documents and text within the enterprise and from external sources.

Usage

Unstructured analytics enables the following capabilities: Capture early signals of customer

discontent Quickly target product deficiencies Find fraud Route documents to those who can

best leverage them Comply with regulations such as

XBRL coding or redaction of PII

Data Retrieval

The Data retrieval engine searches across all relevant content to provide a summarized output

Text Mining

Text mining tools extract and identify relationships between entities of interest

Other Capabilities

Linguistic and statistical techniques to extract concepts and patterns Transformation of language into data Unlocking of meaning and

relationships


Use analytics to improve loss outcomes

Claim adjuster notes and call center notes, often stored as free-form texts, contain valuable information that can be leveraged for better claim outcomes and improve efficiency within claims organization.

Big Data platforms allow insurers to perform advanced analytics on the unstructured claim adjuster notes and to

provide near real-time updates, which opens up the following

possibilities


Realize claims efficiencies

First Notice of Loss Call center notes used to

predict severity Social media data along

with notes used to predict potential fraudulent activities

1 Triage/Assign Claim With improved

severity prediction, claims are classified and assigned in timely manner, reducing costs

Improved claim segmentation leads to “best-fit” adjuster being assigned

2

Initial Claim Setup Improved severity

prediction allows more accurate reserves to be allocated

3 Perform Investigation Adjuster can

search for similar claims and replicate best practice

4

Negotiate / Settle Claim Improved

predictions lead to improved loss outcome

5

Performing advanced analytics on unstructured claims data can improve claim loss outcome and related costs by improving the efficiency and effectiveness of the claim adjuster’s claims handling activities and improving reserving practices.


Leverage visualization techniques

Most insurance companies have access to similar data sets; leading players use visualizations to combine these sets in complex ways to extract unique, actionable insights.

Example scenario

• The Chief Risk Officer or Chief Information Officer for a large Property & Casualty Carrier needs to prepare for an impending natural disaster, in this case a hurricane heading up the eastern seaboard

• Information about the path of the storm, insured risk, loss prediction models all need to be evaluated in combination with team location data

Examine the hurricane’s projected path using a real-time, publicly available information from NOAA

Overlay the path with the book of business. Projected losses correlated to in force policies and loss projections from Catastrophic Loss Models

Gain insights about where agents and adjusters are location, if they are likely to be impacted by the event, and what other field service personnel can be brought in for support

1. Examine 2. Overlay 3. Assess

The Journey to Big Data


Move from basic to advanced information management. Big Data is the next step in the evolution of analytics to answer critical and often highly complex business questions. However, that journey seldom starts with technology and requires a broad approach to realize the desired value.

Expand on your capabilities

Reporting

Data Analysis

Modeling and Predicting

“Fast Data”

“Big Data”

Data Management

Standardize business processes

Focus less on what happened and more on why it happened

Establish initial processes and standards

Leverage information for predictive purposes

Analyze streams of real-time data, identify significant events, and alert other systems

Leverage large volumes of multi-structured data for advanced data mining and predictive purposes


Journey to big data

Develop a Strategic Plan Identify strategic priorities

Identify Opportunities Brainstorm and ask “crunchy” questions

Determine Data Sources Assess the landscape, current capabilities, and priorities

Adopt in Production Prioritize and implement successful, high- value initiatives in production

Identify and Define Use Cases Based on the assessments and business priorities, identify and prioritize big data use cases

Pilot and Prototype Identify tools, technologies, and processes for use cases and implement pilots and prototypes

1 2

3

6

4 5


Step 1: Develop a strategic plan

Every Big Data project starts with a short planning and scoping phase.

Conduct analysis

Evaluate current situation

Mission vision values

Situation assess-

ment

Key issues

Analysis of external sources

Analysis of internal sources

Synthesis

Future

Industry

Scenarios

Future industry

scenarios

Formulate strategy

Create transformation

plan

BI & analytics roadmap

Strategic Big Data

plan

Action-plans

Strategic options

Reward

Strategic direction

Workshops Interviews Brainstorm sessions, Workshops, and analyses

Implementation plan writing

Creativity and ideas

Think outside of the box


Step 2: Identify opportunities

Identifying strategic opportunities starts with asking “crunchy” questions for “sticky” business issues. This process is independent of the underlying data (volume, variety, and velocity) and therefore applicable to both traditional and big data analytics.

Sales • How many of our leads have

converted into sales? • What is the profile of those

leads? • What campaigns are

generating the higher response rate and have the best ROI?

Risk How can we eliminate offers to those adversely effected by underwriting decisions?

Customers • Are our customers frequently changing products? • What are the key customer metrics across LOB’s for acquisition, retention

rates, and customer satisfaction? • Who are the next 1,000 customers we’ll lose — and why? • How do factors such as politics and demographics affect the price our

customers are willing to pay?

Product How can we improve product pricing by analyzing data from different sources?


Asking the right question can go a long way. Big Data introduces new technologies and tools for coping with the volume, velocity, and variety that characterize data sources in current business ecosystem. The opportunities are exciting, but a multitude of difficult questions first need to be answered.

Step 3: Determine data sources

Selection Criteria

Data Structure What structure can be derived from nontraditional data sources to make storage, analysis, and ultimately decision-making easier?

Governance What data governance is appropriate when analysis is distributed, needs change, and data definitions and schemas evolve over time?

How is data quality managed across so many sources of data, many of which come from outside the organization, such as public social networks?

Architecture What levels of availability and reliability are possible in mission-critical applications when data volumes are so large?

What intellectual property, licensing, and data protection considerations apply when Big Data environments are distributed across boundaries?

Infrastructure Is specialized hardware required for a particular need, or can low-cost commodity hardware be leveraged to scale processing?

How can current IT skill sets best be leveraged in evolving the infrastructure to include Big Data?


Step 4: Identify and define use cases

Identify and define use cases to unlock the value of Big Data.

Identify

Identify key information needed and the data sources required.

Access

Access internal and external data sources to provide an integrated view of the organizational data.

Analyze

Analyze the data using statistical tools and techniques to discover patterns and generate insights.

Act

Act on the insight from the analytical models and visualizations to produce business results.

Visualize

Visualize the data to engage non-technical business users and focus attention on the right problems.


Steps 5 and 6: Pilot and adopt

Valuable time and money can be saved by adopting a business user driven prototyping approach that targets value providing initiatives.

Governance and stew

ardship

End User Environment

Collection

Ingestion

Discovery and

cleansing

Integration

Analysis

Delivery

Production

Extract & Load

LOB applications Files Data marts

Marketplace — external data

Data quality

Analysis cubes

Data warehouse

Transform

Analysis Reports Dashboards & scorecards

Analyze

Business user

Hypotheses / questions ? Pilot

Spreadsheets, Specialized Tools, Sandboxes

Value? Yes 1 2

POC

prototype

3 Implement

AND

4

Repeat the POC / prototyping process with more value-providing initiatives

Repeat process Adopt Implement successful, high- value initiatives in production

Big Data environment

Visualize

Analytical environment

Hadoop | MPP | Appliance | In-memory

This publication contains general information only, and none of the member firms of Deloitte Touche Tohmatsu Limited, its member firms, or their related entities (collective, the “Deloitte Network”) is, by means of this publication, rendering professional advice or services. Before making any decision or taking any action that may affect your business, you should consult a qualified professional adviser. No entity in the Deloitte Network shall be responsible for any loss whatsoever sustained by any person who relies on this publication. As used in this document, "Deloitte" means Deloitte Consulting LLP, a subsidiary of Deloitte LLP. Please see www.deloitte.com/us/about for a detailed description of the legal structure of Deloitte LLP and its subsidiaries. Certain services may not be available to attest clients under the rules and regulations of public accounting. Copyright © 2014 Deloitte Development LLC. All rights reserved. Member of Deloitte Touche Tohmatsu

Big Data - An insurance business imperative

Business

big data bring

big data ecosystem

making big

big data analytics

external sources

advanced analytics

structured