why predictive analytics should be part of your 2015 strategy final

Why Predictive Analytics Should Be Part of Your 2015

BI Strategy

Joseph Brandenburg

Predictive Analytics Practice Leader · Dunn Solutions

Predictive Analytics / 2

A Little About Joe Brandenburg and Dunn Solutions Group

What is UDIB?

What is Predictive Analytics?

Why it is important for companies to use Predictive Analytics to

stay competitive?

How key decision makers can make better plans and improve

business decisions with Predictive Analytics

How and when departments and organization can incorporate

Predictive Analytics as part of the Business Intelligence (BI) strategy

to minimize effort and cost

How current technologies make it easier than ever for companies of

all sizes to implement Predictive Analytics

How Predictive Analytics is used in the real world for significant ROI

Agenda

• Been performing and managing Predictive Analytic projects for 17 years

• Have helped companies realize multiple $billions of dollars in new revenue and helped them avoid potential big loses

• Expertise across many industries Financial Services, Banking, Insurance, CPG, Retail, Higher Education, and more

• Predictive Analytics Practice Leader for Dunn Solutions Group

A little about Joe Brandenburg

Predictive Analytics/ 3



What is UDIB?



stay competitive?









Agenda

• UDIB is a simple concept for developing Predictive Analytics:

• It is the ability Understand, Determine, Implement, Benefit

• Understand what the problem areas are or what opportunities exist

• Determine which course of action you should take to address the questions

• Implement the techniques and models that best answer the issues

• Benefit from the implementation by measuring the results and setting baselines

UDIB of Predictive Analytics




What is UDIB?



stay competitive?









Agenda

• Predictive Analytics examines patterns found in internal and external data to identify future risks and opportunities.

• It is projecting where the Best and Sweetest Nuts will be in the forest hidden amongst all the trees and other nuts.

Based on:• Where you have been before

• What you know about the situation

• What others have told you



Predictive Analytics is the GPS of a Business Intelligence system because it provides the greatest business value by making forward-looking decisions possible

How has Business Intelligence evolved?


Source: The Data Warehouse Institute (TDWI)

• It helps to assess, predict, and solve business problems using lots of data

What does Predictive Analytics provide?


Strategic Modeling and Forecast Prediction

Market Blended

Econometric

Factors

(Home

Owners)

Broker and

Distribution

(Home Owners)

Claims

Experience

(Home

Owners)

Legal /

Regulatory

Environment

(Home

Owners)

Our

Competitive

Landscape

Compared to

the Industry

(Home

Owners)

Wealthy

Home

Owners

Buyer

Density

Wealthy

Home

Owners

Market Size

Market

Power

Ranking

Report

(Home

Owners)

Market

Penetration

(Home

Owners)

Category

Development

Index /

Sales

Strength

(Home

Owners)

Category

Development

Index p

(Home

Owners)

Market Outlook

(Home Owners)

Ft. LauderdaleStrong Strong Good Good Strong 6.5 125,451 7.40 10.6% 130.93 0.65

Spend More Effort

Expanding

Kansas CityStrong Strong Good Very Good Strong 6.0 45,804 7.10 8.4% 72.17 0.36

Spend More Effort

Expanding

Power Ranking Report

Detail – Policy and Claim

Data mining for Patterns

Transactional –

opportunity and

validation

Lifetime Value

Persistency

High Value

Customers - Products

Summary, Reporting,

Forecast, Optimization

Traditional Reporting• Works great for viewing current data

• You can use data warehouses to show historical trends

• An average person looks at 3 maybe 4 variables and get a “feel” of the behavior

• People are only good at identifying big trends and relationships

What Predictive Analytics brings to the table beyond Traditional Reporting?


Predictive Analytics• Forward Looking

• Measurable and Traceable

• Millions of Observations and Criteria

• Predictive Analytics is not just Forecasting

How is Predictive Analytics used?


Where is there opportunity for me to

expand?

What factors are causing my

customers to churn?

If I sell items in bundles will my

overall sales increase?

Which segments should I target?

Can I reduce advertising costs by

20% without impacting sales?

What channel is most effective to

reach certain customers?

When should I contact customers and

what message to get them to stay?

What products and combinations do

people want?



What is UDIB?


Why it is important for companies to use Predictive Analytics

to stay competitive?









Agenda

• A recent study from the Economist Intelligence Unit (EIU) revealed that:

Business leaders rate creating a data strategy for marketing and predictive analytics as one Marketing's most important priorities.

• A recent study by the Nucleus Research says:

• Analytics pays back $10.66 for every dollar spent

• Dunn Solutions projects have been more like $100 for every dollar spent

• This is a $100 your competitors are making through changes and adjustments that you are not realizing

Why is Predictive Analytics key for everyone?


• Predictive Analytics has proven over and over again that it leads to increased sales and profit.

• Marketing, sales and customer relationship management are some of the areas where the returns from Analytics are the highest.

• Example: One online retailer was able to increase its sales by thirty percent for a single campaign and by 3 percent for the whole company. This added an additional $80 million in revenue from one area alone

Why are more and more companies are using it?




What is UDIB?



stay competitive?









Agenda

• Successful predictive analytics projects are a driving force, the movement to customer‐oriented and actual‐needs focused organizations is accelerating.

• One Forrester study concluded, "Predictive Analytics Brings Fresh Momentum“

Why Predictive Analytics changes the game?


• Analytically Driven companies use Predictive Analytics to help their companies just like yours make the most of their existing data to uncover future trends, risks and opportunities that result in bigger returns, or avoid big losses.

How Predictive Analytics can lead to better decision making


Source: Harvard Business Review

• Seldom trusts analysis

• Makes decisions unilaterally

• Applies judgment to analysis

• Listens to others but is willing to dissent

• Trusts analysis over judgment

• Values consensus

Reliance on institution

Reliance on analysis

38% of employees are informed skeptics

19% of employees are visceral

decision makers

43% of employees are unquestioning

empiricists

• Visceral Decision Makers often make decisions based on their gut which could be contrary to what their customers want and operations needs

• Ideally, analytically-driven companies must have predictive analytics embedded in all their business processes, thereby moving away from decisions based on gut feeling or intuition.

Taking advantage of ‘Real’ Information to make decisions


• Management and the company become better users of information and make the most out of it just like they do in other areas such as managing talent, capital, and brands.

• Too many executives treat data as something the IT department handles and often considers themselves too novice to get deeply involved in how data is shared across the organization.

• Managers and Senior Management need to wake up to the fact that their data investments are providing limited returns because their organization is underinvested in understanding the information.

By Embracing Data, BI and Predictive Analytics


• There are four phases in most industries customer life cycle • Acquisition

• Relationship

• Retention

• Win-back

• Each phase should have several embedded analytical models, which can enhance operations considerably and provide a strategic advantage.

Tailoring your management and experience to the customer




What is UDIB?



stay competitive?




Predictive Analytics as part of the Business Intelligence (BI)

strategy to minimize effort and cost




Agenda

• There is no better time than the present but it is best to incorporate predictive analysis in with other implementations of Smart Tools and BI

• However, there are some key things to think about!!!

When to implement Predictive Analysis?


• Predictive analytics is made up of three major components:

• Data

• Statistics

• Assumptions

• It is important to be prepared to make some changes to data and the arrangement in a data warehouse or in extracts

• Many analytics softwares today have the ability to do this automatically

Predictive Analytics Requirements and Components


• Data• Predictive Analytics needs the right kind of data and information to help

answer the main questions you are trying to predict or forecast.

• It is best practice to collect as much relevant data as possible in relation to what you are trying to predict. This means tracking past data, customers, demographics, and more.

• Statistics• Most Predictive Analysis starts with curiosity about a business problem or

customers

• Assumptions• Because predictive analytics focuses on the future, assumptions that the past

will mirror the future are imperative.

• Need to understand what you are trying to solve

• However, with modern techniques, there are options to do exploratory analysis and identify problems or opportunities

Data, Statistics, Assumptions


• Having and implementing “Big Data” does not help you make better decisions!!!

• The process of taking big data and developing analytics and especially predictive analytics drives better decision making

• Big Data provides a ton of great data but predictive analytics synthesizes that data into actionable information – if done correctly

Big Data and Predictive Analytics


• Collection Analytics

• Marketing Activities• Cross-sell• Customer retention• Direct marketing• Portfolio, product or economy-level prediction• Analytical customer relationship management (CRM)

• Risk Management• Fraud detection• Underwriting• Pricing

• Disease Prevention and Management• Clinical decision support systems

Most Common Uses of Predictive Analytics


• Earlier, up-selling through campaigns used to be carried out through trial and error.

• Nowadays privacy norms forbid operators from reaching out to customers repeatedly.

• Hence, all instances of customer contact become significant. In such a scenario, analytics can help develop a ‘sharp-shooting' method to run campaigns.

• Based on historical data and customer profiles, it is possible to classify customers according to their likelihood of buying a product or a service through a campaign.

• Thus, every campaign can target the set of customers with better purchasing potential for that service/product.

• While these statistics-driven campaigns yield higher ROI, they also reduce the irritation caused by non-relevant communication, thereby indirectly reducing customer dissonance.

• Campaign analytics is most beneficial to product management teams new methods exclude unprofitable customers.

Campaign Analytics


• Churn is expensive, costing companies millions of dollars every year. Probably the most widely used analytics models are churn modeling to help reduce the impact of voluntary churn to the bottomline.

• Typically, the model measures the ‘Churn Propensity' of a customer, and profitable customers are wooed through a set of personalized actions, like offering freebees.

• The value of the offers can be tailored to suit customer profitability. Churn modeling is an easy analytics option, with relatively low complexity, wide acceptance among functions and immediate results.

• The marketing department usually initiates these activities, and most companies use their customer service and sales teams to execute these plans.

Churn Modeling


• A very real challenge in most industries is how to increase yield from the current subscribers/customers, and how to improve Average Revenue Per User (ARPU).

• Cross-selling and up-selling activities can now be supported by predictive analytics based on transaction histories and like customer behavior.

• Analytically driven cross-selling and up-selling campaigns provide remarkably higher returns. Most of our projects see a 40% lift in Cross-sell and Upsell and by moving beyond financials, they also increase retention and reduce the number of contacts required for cross-selling and up-selling.

• These models may be integrated with real-time decision engines and developed as real time cross-selling and up-selling systems.

• The campaigns are usually planned by the marketing department and executed through in-bound customer facing teams, mainly call centers and web transactions

Cross-Selling and Up-Selling


• Not all customers are the same. Although most organizations follow this credo at one level, it is important to assign a dollar value to each customer, in order to prioritize various sets of customers.

• The Customer Lifetime Value (CLV) model provides the predicted yield from each customer over the customer life cycle.

• High priority customers can be given loyalty bonuses, preferential treatment through personalized service, better credit norms for contract subscribers etc.

• These analytic models may be utilized across all the functions like marketing, credit risk, customer service and so on.

Customer Lifetime Value Analytics


• This has been one of the main analytical models in the retail, financial services, CPG and Service industries.

• Customers are segmented both at the prospect and purchase phases.

• At the first stage, segmentation helps reach out to prospects with higher predicted conversion rates, thereby increasing the campaign success rate as well as the ROI.

• During campaigns, prospects are divided into segments to which specific campaigns are targeted.

• This approach helps eliminate meaningless segments that unnecessarily clutter the thinking and execution of the campaign.

Customer Segmentation


• Fraud is a key root cause of lost revenue in the many industries.

• Efficient fraud detection systems can help companies save a significant amount of money.

• Fraud detection systems depend on data mining algorithms to identify and alert the company to fraudulent customers and suspicious behavior.

• Remember, there are several methods of fraud, requiring other analytic models to aid in detection.

• Risk management teams are the largest users of fraud management systems.

Fraud Analytics


• Companies spend heavily on mass media.

• Spends cover television, newspapers, radio, magazines, email and the internet.

• The marketing spends are often apportioned on the basis of instinct rather than hard facts.

• A Marketing Spend Optimization or better know as marketing mix model helps marketing managers and product managers take decisions based on what works and what does not.

• This analytics model has been of considerable benefit to the marketing function, and is hence widely used to improve marketing Return on Investment (ROI).

Marketing Spend Optimization – Marketing Mix


• Price optimization contributes significantly to revenue development and profitability and is especially important in the corporate sales segment, where awareness of the impact of the various pricing options offered is critical.

• Simulated scenarios can help evaluate the revenues at various price points.

• These models are widely used by product managers and finance teams.

Price Optimization and Elasticity


• Traditionally, customers have been measured in terms of the revenue they bring to the company.

• In this age of social media, how can a company measure the value of a customer who is socially influential' among his/her peer groups?

• It is well-known that early adopters in families and offices influence a large number of followers.

• Such ‘influencers’ require a special, differential treatment, even though their billing is often low and customer lifetime value is not very high.

• This kind of differentiation is possible through Social Network Analytics.

• This model is of great help in the areas of churn prevention of profitable customers, cross-selling of new products and so on.

• There have also been instances where social network analytics have been used to test advertisements.

Social Network Analysis


• Listening to the customers and understanding their needs is crucial, especially in this era of viral communication.

• While it has been an established practice to undertake sentiment analysis based on content from call center logs and social media through reporting and dashboards, predictive models can be utilized in social media analysis to listen to the written word.

• Data mining techniques add tremendous value in this area.

• Similar feedback around key discussion areas pertaining to the company's activities can be brought together using various methods.

Social Media Analytics


• Web Analytics has traditionally been an area where only reporting or dashboarding was implemented.

• Now organizations are utilizing predictive analytical models to take web analytics to a new level.

• Using data mining techniques, customer profiles that help determine which factors differentiate the behaviors of one group from another, may be generated.

• Models can be used to identify new visitors and predict their future behavior, for example, if they will subscribe to a particular service or not.

Web Analytics




What is UDIB?



stay competitive?






How current technologies make it easier than ever for

companies of all sizes to implement Predictive Analytics


Agenda

• Predictive Analysis (SAP)

• Infinite Insight (KXEN) now SAP

• SAS

• SPSS

• R

*Future webinars will discuss technologies and give demos of the top tools.

Most common tools used in Predictive Analytics




What is UDIB?



stay competitive?








How Predictive Analytics is used in the real world for

significant ROI

Agenda

CASE STUDY

Top U.S. Bank

SolutionsPredictive Modeling Engine & Dashboards

Challenge

• Top 10 U.S. Bank faced with mounting delinquencies in all banking areas

• A huge backlog of foreclosures

Solution

• Designed a predictive analytics modeling engine and dashboard reporting environment

that tracked all consumer lending and mortgages

• Tracked customers through models to help determine when and what action best fit

the situation

Result

• Generated models on defaulting loans which justified an additional $2.2 billion in TARP money

• Staffing Model and Capacity Planning for Collections and Defaults

• Use automated dialers to call customers

• Using targeted messages to better handle the collections and default, saving $9 million each month in extra phone capacity

• Increased collections and curing by 25%

• Better Customer Risk Targeting $40 Million

• Which collections and foreclose first ?

• Which properties and customers were best to restructure and make other offers?

• Prioritize properties, loans and targeted outcomes


CASE STUDY

Online Mass Retailer

SolutionsMarket Basket Analysis

Predictive Recommendations

Challenge

• Online Mass Retailer was missing out on the opportunity to add additional revenue by

providing differentiated customer experiences and by offering the right items in the right

order to increase cross-selling

Solution

• Identified optimal product mix and cross-sell offering

• Helped to customize the consumer experience and optimize revenue through more

revenue per customer

Result

• Recurring annual benefits of more than $80 M in additional cross-sell revenue and an

average increase of $.94 per customer purchase


CASE STUDY

Fortune 100 Insurance Company

SolutionsPredictive Forecasting Simulation Tool

Marketing Mix Modeling

Challenge

• A subsidiary of a Fortune 100 insurance company provides health insurance services

in 29 states and was facing specific new competition

• Products were underpriced for claims experience

• Larger competitors were quickly gaining market share

Solution

• Performed Segmentation of Customers

• Developed Forecasting Simulation Tool to predict product performance

• Developed Marketing Mix and Market Potential

• Market based assessments to better target customers

Result

• Transformed the subsidiary into the fastest growing individual health insurance company in the nation at a time when

most in the industry were shrinking

• Doubled the portfolio with targeted high-value customers in just 18 months to over 370,000 customers and $500 million

in new annual revenue


Thank you

Joseph Brandenburg

Predictive Analytics Practice Leader · Dunn Solutions

why predictive analytics should be part of your 2015 strategy final

Documents

predictive analyticshow

predictive analytic

business decisions

dunn solutions groupwhat

dunn solutions groupa

significant roiagenda

better plans

key decision makers