35
13 October 2016 Claims Management © 2016 Willis Towers Watson. All rights reserved. Leveraging analytics to improve performance Tom Helm

Claims management: leveraging analytics to improve performance, Tom Helm

Embed Size (px)

Citation preview

Page 1: Claims management: leveraging analytics to improve performance, Tom Helm

13 October 2016

Claims Management

© 2016 Willis Towers Watson. All rights reserved.

Leveraging analytics to improve performance

Tom Helm

Page 2: Claims management: leveraging analytics to improve performance, Tom Helm

Claims data – so powerful it can save lives

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 2

Page 3: Claims management: leveraging analytics to improve performance, Tom Helm

Agenda

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 3

1

Empowering people

2

Advanced Analytics

3

Data Treasure Trove

Page 4: Claims management: leveraging analytics to improve performance, Tom Helm

What is Claims analytics?

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 4

Claims analytics => “Leveraging data to provide insight”

Static

Descriptive

Diagnostic

Prescriptive

Predictive

Cognitive

Page 5: Claims management: leveraging analytics to improve performance, Tom Helm

Claims analytics to improve claims performance

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 5

Unlocking the potential

Claims Spend Optimisation

Operational Performance

Advanced Claims

Analytics

Claims

Analytics

Page 6: Claims management: leveraging analytics to improve performance, Tom Helm

Claims analytics – empowering the enterprise

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson cl ient use only. 6

Page 7: Claims management: leveraging analytics to improve performance, Tom Helm

Claims analytics – empowering the enterprise

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson cl ient use only. 7

Page 8: Claims management: leveraging analytics to improve performance, Tom Helm

Claims analytics – empowering the enterprise

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson cl ient use only. 8

Page 9: Claims management: leveraging analytics to improve performance, Tom Helm

Claims Inflation

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 9

Page 10: Claims management: leveraging analytics to improve performance, Tom Helm

Key challenge

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 10

Page 11: Claims management: leveraging analytics to improve performance, Tom Helm

Car rental battle

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 11

Page 12: Claims management: leveraging analytics to improve performance, Tom Helm

Initial approach

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 12

Management

Information

Practices and

processes

People

Engagement

Page 13: Claims management: leveraging analytics to improve performance, Tom Helm

New Claims System

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 13

Page 14: Claims management: leveraging analytics to improve performance, Tom Helm

The breakthrough – analytics empowering people

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 14

Generate a daily report Auto Emailed to Team

Leaders @ 8:30am daily Capture the data

Team Leader reviews all

hire cases & prioritises

Claims adjusters clear

line of sight pro-active

Analytics

empowering people

Results delivered

€200 saving per claim

€2.4m saving per year

Case Study: Reducing Third Party Car Rental Costs

Page 15: Claims management: leveraging analytics to improve performance, Tom Helm

Claims analytics - monitoring behaviours

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 15

0

100

200

300

400

500

600

I.14 II.14 III.14 IV.14 V.14 VI.14 VII.14 VIII.14 IX.14 X.14 XI.14 XII.14 I.15 II.15 III.15 IV.15 V.15 VI.15 VII.15 VIII.15 IX.15 X.15 XI.15 XII.15

Tier 1 Tier 2 / Approved Other 3 per. Mov. Avg. (Tier 1) 3 per. Mov. Avg. (Tier 2 / Approved Other)

Vol distribution of repairs within Network – Tier 1 vs Tier 2

Network operates a two

tier solution

Case Study: Customer Vehicle Repairs – Repair costs up 22%

Outsourced repair

Network Average costs £650 lower

in their tier 1 solution

£465k Saving

opportunity

Page 16: Claims management: leveraging analytics to improve performance, Tom Helm

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 16

Advanced Analytics

Page 17: Claims management: leveraging analytics to improve performance, Tom Helm

Advanced Analytics – a solution tool kit

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 17

Data Technology Modelling People / skill set

Page 18: Claims management: leveraging analytics to improve performance, Tom Helm

Liability

Investigation

Medical Experts

Employers

Insurer

Accident

Management

Solicitor

Third

Party

Customer

ClaimFNOLIncident

Rich data gathered throughout the claim lifecycle

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 18

Can I learn about

the accident

earlier?

Can I allocate the

claim to the right

people?

Can I identify

fraud or

exaggeration?

Can I identify

witnesses?

Images?

Can I establish

severity of

injuries?

Can I settle this

claim quicker?

Can I be the first

to contact them

and offer a

service?

Can I reserve

more accurately?

Can I gain a better

insight into the

claimants

wellbeing/conditio

n?

Can I identify

potential high

value claims

earlier?

Can I identify

hidden witnesses?

Can I identify

hidden images?

Can I determine

nature of impact

through images?

Images

Emergency

services

Highway agency

Vehicle recovery

agency

Witnesses

Hospital

Location

Weather

conditions

Vehicle damage

severity

Injured parties

Witnesses

Circumstances

Injury details

Hospital details

Location

Weather

conditions

Third party vehicle

details

Third party details

Vehicle damage

repair estimate

Damage images

Circumstances

Vehicle

damage/repair

estimate

Acting solicitor

Injury details

Employment

status/occupation

TP Mobility status

Injury severity

Age

Vehicle damage

costs

Vehicle damage

severity

TP vehicle

mobility status

Hire vehicle need

Injury details

Salary

Prognosis

Pre-existing

condition

Treatment

Specialists

involved

Job Nature

Commute

distance

Seat belt

Images

Witness

statementsTotal Loss

Vehicle

registration

document

Purchase details

MOT certificate

Page 19: Claims management: leveraging analytics to improve performance, Tom Helm

Variety of data sources and techniques

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 19

Data digitalisation

Page 20: Claims management: leveraging analytics to improve performance, Tom Helm

WTW modeling process

20

Data Design Data ExplorationFeature Generation

& SelectionData Transformation

& Visualization

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only.

sources

timeframe

sampling

fuzzy matching

what is a

record

correlation

linearity

grouping

clustering

principle

components

independent

components

singular value

decomposition

factor analysis

supervised:

GLM, LDA,

SVM, SGD, NN,

trees, gaussian

processes,

naïve bayes,

ensemble

unsupervised:

GMM,

clustering,

matrix

factorization,

NN, manifold

learning

genetic equation

search

topic modeling

one-ways

stats/common

sense checking

match rate

standardization

scaling

normalization

binarization

encoding

imputation

high-order

segmentation

hierarchy

separator

sparse coding

filter based

permutation

based

formula based

bag of words

holdout

cross validation

grid search

quantifying

the quality of

predictions

persistency

validation

curves

Model

Development Model Validation

Topic modeling

Page 21: Claims management: leveraging analytics to improve performance, Tom Helm

Topic modelling summary

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 21

A topic is a set of co-occurring of words which

can describe specific events or ideas

E.g., topic describing recurring sci-fi ideas

future, technology, space, aliens, science

In an insurance context, topics represent

common events related to the insurance

process

For loss adjuster notes, topics reflect how:

An adjuster handles a claim

A claimant recovers from the loss/injury

For UW notes, topics reflect how:

An insured relates to, manages, or cares for

the insured item

An insured item was reviewed and documented

Page 22: Claims management: leveraging analytics to improve performance, Tom Helm

Text Mining

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only.

22

Traditionally, if an insurer wants to systematically summarize information within

text documents, then word indicators are used:

Word indicators ignore relationships among words, and so part of a document’s

meaning is lost

Employee injured

lower back by lifting a

heavy box

Unused Words in Claim

Employee injured

lower back by lifting a

heavy box

Adjuster’s Note for Claim

Claim # Surgery Ind Lift Ind

123 0 1

Restated Adjuster Note

via Word Indicators

Page 23: Claims management: leveraging analytics to improve performance, Tom Helm

Topic modelling – advanced text mining

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 23

Advanced text mining techniques like Topic Modeling can capture the

content and meaning encoded in your text documents by restating

them as a blend of common topics or themes inferred from your

collection of documents

Topic modeling can create structured data from text documents

without significant loss of meaning

Employee injured

lower back by lifting a

heavy box

Adjuster’s Note for Claim

Claim # % of Topic 1 % of Topic 2

123 0.32 0.15

Restated Adjuster Note

via Topic Modeling

Page 24: Claims management: leveraging analytics to improve performance, Tom Helm

Claims Triage - Sleeping Giants

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 24

Case Study : Using topic modelling to identify claims that may “blow up”

Financial Benefits

€ Reduced claims settlements

€ Reduction in claims leakage

€ Improved reserving

€ Improved pricing

Opportunity to investigate

Opportunity to intervene

Opportunity to settle early

Increased control of legal costs

Investigate the claimant

background

Investigate fraud / exaggeration

Time for surveillance

The Problem Claims Handling Benefits of

Earlier Identification

High value claims

hidden amongst

the lower value

claims

=> late

identification and

late reserving

Page 25: Claims management: leveraging analytics to improve performance, Tom Helm

Topic Modeling Results

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 25

Topic 1 dentist tooth dental teeth lip rebar patent crown jaw

Topic 2 lifting felt muscle heavier pulled weighs lbs strain pop

Topic 3 herniated esi disc stenosis epidural spineneuro

surgeonbulge fusion

Summary:

The included topic factors were statistically significant, time

consistent, semantically coherent & reasonable for this

application, they were also more important than many factors already

in the model (including text mining flags)

Case Study : Using topic modelling to identify claims that may “blow up”

Page 26: Claims management: leveraging analytics to improve performance, Tom Helm

Topic Modeling Results

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 26

Double lift chart example from GLM including results from 100-topic STM with (supervised) content covariate

Case Study : Using topic modelling to identify claims that may “blow up”

Page 27: Claims management: leveraging analytics to improve performance, Tom Helm

Advanced Analytics – Opportunities across the claims process

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 27

Fault vs Non Fault

Fast settlement for high

value customers Supplier optimisation

Quickly identify the likely

claim outcome

Auto (re)allocate the

claim to the right people

Resolve fault claims

quick – save money

Improve subrogation

Link with underwriting

information to identify

high value and

trustworthy customers

Auto feed into the claim

notification process

Enable day one

settlements

Evaluate performance

between suppliers

Assess image of

damage – route claim to

most effective solution

Source repair or

replacement via an

online tender or

selection process

Determine the top

performers

What are the handling

characteristics of a top

performer?

Share the insights and

monitor to drive

performance

improvement

Claims Adjuster Performance

Page 28: Claims management: leveraging analytics to improve performance, Tom Helm

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 28

Claims => Data Treasure Trove

Page 29: Claims management: leveraging analytics to improve performance, Tom Helm

Claims Data – Providing deeper customer insight

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 29

Customer Data

Collected

At policy

inception

During

Claims

Process

New Correlations New Customer Insights

Lifestyle

Family

Travel / Holiday

Attitude to claiming

Attitude to risk

Assets

Page 30: Claims management: leveraging analytics to improve performance, Tom Helm

Claims Data – Providing deeper customer insight

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 30

Customer Data

Collected

At policy

inception

During

Claims

Process

New Correlations New Customer Insights

Lifestyle

Family

Travel / Holiday

Attitude to claiming

Attitude to risk

Assets

Business Analytics & Intelligence – Job Advert Walmart

• Our work is about more than just knowing what our customers want, it’s

about understanding who they are.

• It all comes down to working better so customers can live better.

• Our team is a customer-understanding powerhouse.

• Retail may be what we’re known for, but science is what’s building our

next-generation of business

• No matter what role you play, you’ll be at the– at the heart of customers’

needs.

Page 31: Claims management: leveraging analytics to improve performance, Tom Helm

Claims analytics – valuable asset across the business

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 31

Sales &

DistributionBig Data

Pricing &

Underwriting

Financial Reporting

Reinsurance &

Capital Management

Reserving

Claims

Analytics

Page 32: Claims management: leveraging analytics to improve performance, Tom Helm

Summary

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 32

Claims Data => A treasure trove to be unlocked

Smart Analytics => Business Experts + Analytic Experts

Recognise increased value in data

Claims community engaged

Empower people through data

Claims analytics can help:

solve business challenges

create new claims handling solutions

Powerful tool kit

Huge potential

Blend skills

Culture Change

Page 33: Claims management: leveraging analytics to improve performance, Tom Helm

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 33

Claims data – so powerful it can save lives

Page 34: Claims management: leveraging analytics to improve performance, Tom Helm

Discipline Everyone knows the value of data

Transforming claims analytics

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only. 34

Mapping the journey

Current State

Data Availability

Data Quality

Data Frequency

Analytical

Tools/Techniques

Analytical

Resources/

Skills

Analysis

Reports/

Outputs

SME

Key Questions

Proficient

Advanced

Stakeholders

ROI

DeliveryRich new insights

Data Identify the source

Determine What key outputs will provide value ?

Page 35: Claims management: leveraging analytics to improve performance, Tom Helm

For further information please contact:

35

[email protected]

Tom HelmHead of Claims Consulting

Risk and Financial Services

71 High Holborn

WC1V6TP London

United KingdomT: +44 20 7170 2262

M: +44 7773 040703Confidentiality Statement

This document has been prepared for the

sole and exclusive use for participants of the

Czech Insurance Conference 2016,

Willis Towers Watson Czech Republic.

Distribution or disclosure of, or quotation

from, or reference to this document to any

other party, is prohibited without the prior

written consent by Willis Towers Watson

(“WTW”, “we” or “us”.)

© 2016 Willis Towers Watson. All rights reserved. Proprietary and Confidential. For Willis Towers Watson and Willis Towers Watson client use only.

[email protected]

Roger GascoigneDirector

Risk Consulting and Software

Klimentská 1216/46

110 00 Praha 1

Czech RepublicT: +420 222 191 239

M:+420 602 313 408