Transform Your Business Intelligence with Predictive Analytics
Welcome / Introductions Analytics Market Trends Demo: Customer Churn Q& A with BluePay Predictive Analytics Best Practices Questions Conclusion/Key Takeaways Slides will be distributed to all attendees
SOFTWARE SOLUTIONS MANAGED SERVICES DIGITAL MARKETING INFRASTRUCTURE SOLUTIONS
Business Intelligence and Data Management Roadmaps Operational, Prescriptive and Predictive Analytics Enterprise Data Management Business Intelligence Managed Support
THE INC. 500 5000 The Channel Company CENTRAL REGION PARTNER OF THE YEAR AWARD PROUD MEMBER OF THE GLOBAL EDITION ONE OF AMERICA S FASTEST GROWING PRIVATE COMPANIES. 2012 2013 2014 2015 2016 CRN Solution Provider 500 Managed Services Provider 500 CUSTOMER LOYALTY & SATISFACTION Top 100 Managed Service Providers 2012 2013 2014 2015 2016 2012 2013 2014 2015 2016 The Computerworld Honors program, founded in 1988, recognizes organizations and individuals who have used information technology to change the world for the better The Interactive Media Awards honor individuals and organizations for their outstanding achievement in website design and development.
Predictive Analytics Discussion
Business Intelligence Trends Businesses have access to more data today than ever before. The datafication of nearly every workload and result has organizations turning to advanced analytical processes to make sense of these massive data sets. With recent advancements in analysis tools, the ability to perform advanced data analytics has quickly become highly PRACTICAL, EFFICIENT and USER-FRIENDLY Organizations of all sizes are embracing ADVANCED ANALYTICS
However Market Trends: Advanced Analytics Organizations that utilize advanced analytics on top of Big Data 20 will grow % (Gartner) More than their peers 5x Companies using advanced analytics are more likely to make faster decisions 85 % of business leaders believe big data will dramatically change the way they do business Most organizations analytics projects are focused on the past, Analyzing their historical data Only13 % are using advanced analytics to Look to the future
Market Trends: Predictive Analytics 34% Actively using predictive analytics 52% Investigating predictive analytics (TDWI Research Study sponsored by Tableau) Predictive analytics is quickly becoming a decisive advantage to inform strategy and achieve business outcomes Key Drivers for Predictive Analytics Top rated reasons organizations cited they will use Predictive Analytics Understanding Customers Predicting trends Understanding customers Predicting behavior Improving Business Processes Drive better business performance Make more strategic decisions Improve operational efficiency
Predictive Analytics vs. Business Intelligence Business Intelligence Helps answer questions such as what happened or what is happening, and perhaps even why it happened. Generally provides static reports or dashboards and can be inflexible. Predictive Analytics Utilizes methods and technologies for organizations to spot patterns and trends in data Test large numbers of variables, develop and score models, and mine data for unexpected insights Users can estimate outcomes (often called targets) of interest Facilitates fluid decision making, transitioning from batch/historical analysis to real-time and even streaming based decision making
Predictive Analytics Example: Customer Churn CUSTOMER CHURN (Attrition) is one of the most universally applicable applications for predictive analytics. Not only does churn affect nearly every type of organization, it is highly relevant for nearly every area of an organization. Most organizations have the right data points to do churn analysis, but need help discovering those data points. Data Preparation is the single biggest challenge to getting a predictive analytics project launched.
Developing a Predictive Analytics Project Strategy A sound process is the key component to making your predictive analytics project successful. Predictive Analytics Process Define Hypothesize Model Test Evaluate Operationalize Determine problem to solve with the analysis Via analysis of the data and feature engineering Organize data & build the predictive model Training, testing and cross-validation Does the model perform? Evaluate others models? Integrate into applications and analytics assets
Demo: Predicting Customer Churn
Demo Details What Did We Learn? Telco customer churn business issue The variables relevant to the problem: Strongest: Number of customer service calls made high number of calls Total daytime minutes used Poor indicators of churn included: Length of account Evening, Night, & International Charges Region where the account is located Others with positive influence on model: Whether or not the customer had an international plan Charges incurred for daytime usage Total evening minutes used
Q&A: An Analytics Journey From Data Discovery to Predicting Merchant Attrition
Introduction Peter Schmidt, Lead Data Scientist with BluePay BluePay is the leading provider of technology enabled payment processing solutions Received my B.S. in Computer Science and M.S. in Analytics from Northwestern University Worked across a variety of industries for such companies as U.S. Cellular, MyPoints, United Airlines, Officemax, and Centro
Project Mission To understand the influential factors contributing to merchant attrition In a way that establishes a set of predictor variables, uses advanced analytic techniques and will establish new metrics So that meaningful questions can be asked/explored, effective strategies/actions can be taken and subsequent predictive modeling efforts can be supported Measured by decrease in merchant attrition rate, increase in merchant time on file and increases in profitability
Analytic Process Summarize Sell Success Model Comparison Business Problem Hypothesis Brainstorm Dataset Selection Exploratory Data Analysis Profile Visualize Descriptive Statistics Distributions Data Discovery Correlations Patterns Missing Data Evaluate Visualize Finalize Test Predict Assess Visualize Model Assessment Conclusions & Action Model Selection Data Preparation Build & Develop Train & Tune Repeat & Cross Validate Feature Selection Variable Importance Check Assumptions Visualize Pre-processing Stage Clean High Correlations Near Zero Variance Feature Transformations Impute Partition Data Center & Scale Traditional Software Development Lifecycle Req Analysis Design Dev Test QA Impl
Attrition Risk Score Score the Population of Merchants Merchant Account
Which Merchants Should be Targeted? High Brought in profitability metrics Value Focus Examine Score vs Value Better quantify the Risk Low Low High Focus on the most valuable merchants Attrition Risk
Best Practices & Learnings Align your project to corporate objectives Understand the customer journey, Know it, Document it, Understand the touchpoints, Understand how that is tracked (entry/exit) from your IT systems. Review and re-review the data Confirm the business definitions Find that in-house knowledge expert Have the right metric with the right definition in place Status, time on file, last purchase date, last interaction Use the capabilities of your toolsets Tableau Data Discovery & sharing analytic insights. Realize this is just the 1 st step Clustering customers Forming a targeted response Strategy
Case Studies
How has SWC helped others with churn? Professional Membership-based Organization Problem: Volatile membership churn High levels of members dropping off Solution: Brainstorm and identify member data sources Prepare and cleanse data Built predictive analytic experiments and models Identified key reasons causing members to leave the organization Next steps: Client developed strategy & efforts to retain at-risk members Implement approach and evaluate results CORRELATION TO INVESTIGATE: Website visits = Retention
How has SWC helped others with churn? Banking Holding Company Problem: Acquire 1-2 banks in rural areas each year Many customers from acquired bank stay on for about 30 days, then close their accounts Solution: Brainstorm and identify internal and external data sources Prepare and cleanse data Build predictive analytic experiments and models Identify key reasons causing customers to close accounts Develop strategy & efforts to retain at-risk customers Implement approach and evaluate results CORRELATION TO INVESTIGATE: Additional data sets were required to predict customer churn
Getting Started With Predictive Analytics
What s the ROI? 40 % By 2018, there will be a projected of organizations struggle to find analytics talent (MIT Sloan Study) Demand Professionals 1.5 MILLION SHORTFALL of data professionals in the U.S. 60 % of Big Data Projects Will Fail to go Beyond Piloting and Experimentation A successful advanced analytics strategy is about more than simply acquiring the right tools. Many organizations incorrectly view these initiatives as simply another IT project. Need multiple business areas to unite by around a shared, achievable goal. Specialized skillsets and training are required to ensure project results and data quality.
Ways to Get Started BI Discovery and Roadmap BI Pilot and Prototype Solution Development
Key Takeaways More organizations than ever before are using Business Intelligence to gain insights from large sets of data to make more strategic, informed decisions. Predictive Analytics can make a tremendous impact on solving your business problems (like customer churn) and give you the intelligence needed to succeed. Advancements in analytics tools have put BI in reach for organizations of all sizes. However, you won t succeed without a strategy, proven process and achievable goals.
Questions? Please fill out your feedback forms
Thank You Rob Wellen Director of Business Intelligence Rob.Wellen@swc.com 630-230-8678 DIRECT 708-790-9821 MOBILE Chad Dotzenrod Director / Practice Lead, Business Intelligence Chad.Dotzenrod@swc.com 630-286-8108