Optimizing Customer Relationship Management: Adding Optimization to Segmentation and Automation at MarketSwitch 1 What is Marketing Optimization? The “Marketer’s Dilemma” Products • each with specific goals and requirements Prospects -Opportunity set is huge -Resources are limited -How do I satisfy everyone? Classic Optimization Problem... Channels • each with specific costs and capabilities 2 Optimizing The Demand Chain Customer & Product Data Managment & Analysis Data Accumulation Response Modeling, Profit Modeling, Personalization, Segmentation Likelihood's and Propensities per Product MultiDimensional , ConstraintBased Optimization Optimal Offer/ Optimal Channel Intelligent Marketing Resource Allocation Multi-Channel Management & Delivery Marketing Activity 3 The Role Of Optimization Oracle Hyperion Cognos MicroStrategy WebTrends Claritas E Data Experian First Data Trans Union Others SAS RightPoint Net Perceptions Black Pearl SPSS Quadstone Info Advantage Others Siebel E.piphany DoubleClick BroadVision Xchange Prime Response Clarify Others 4 The Real World Problem Customers (millions) Offers (hundreds) Constraints... Minimum sales Maximum cost per customer Maximum total budget ? Minimum NPV per customer Minimum solicited offers Maximum solicited offers Maximum number of promotions per customer “What if” Scenarios... Limit budget Which offers should I send to which customers? Maximum promotions per customer Maximize profit Acquire the most customers 5 Offer Optimization Offers -Ads~Promotion -Products Offer Eligibility Conditions Optimization Engine Offer Economics -CPM rate (ads) -Delivery cost Response & Profit Models -Propensity Score -Profitability per prospect Real Time Campaign Deployment INPUT Optimization Goal -max. NPV -max. click-through/imp. -min. Budget Optimization Schema Optimization Strategy Development (Maximize Business Goals and simultaneously satisfy all Constraints) Application of unique mathematical algorithm to generate optimization schema Business Constraints -min. NPV -max. Budget -min/max contacts 6 Real Time Offer Optimization How it Works Online Promotions Optimization Workflow 7 Real Time Offer Optimization Offer~Ad Request Optimization Web Sites Scoring S e r v e r Optimal Offer Selection Real Time Adjustment Activity Log Ads or Offers 8 Real Time Offer Optimization Case Study 9 Snowball.com Business Model • Leading network for Generation-I – aggregates sites – brings content to audience – brings audience to multiple products and services • 25th largest web property – 14+ million unique monthly visitors – registering 20,000 new members per day 10 Opportunity • Business Challenge – Maximize the value of the 20,000 daily network registrants • businesses can offer special promotions to registrants • revenue generated by serving offers and generating conversions Cool - I’ll Register Visitor Arrives General Network Registration Flow Personalize Value Exchange E-mails & Offers 11 12 13 14 15 Snowball.com Marketer’s Dilemma • Most relevant offer • Achieve revenue and profit goals • Satisfy our specific business commitments – such as number of impressions per offer and conversions • Execute in a dynamic environment Customers (millions) Offers (hundreds) ? – new offers rolling in and old offers rolling out Which offers should be promoted to which customers? 16 Today’s Process How is it accomplished today? – Combinations of business rules • if question 2 is answered A or C • and • question 1 is answered D • and • gender is Male • then show offers X and Z 17 Today’s Limitations Limitations of this approach – Significant profit being left on the table • optimal promotions at any point in time not being served – Extremely resource intensive • team of marketers trying to manage “If~Than” business rules – Not flexible/adaptable • not scalable…as offer pool grows so does the resource requirements 18 Business Objective Clear Objective: – Implement solution that maximizes the revenue derived from registration promotional offers • do it fast • do it with minimal resources • do it in a scalable fashion 19 Requirements • Target limited offers (max 6) from a large pool • Manage inventory of multiple offers • Access entered data and serve offers based on – – – – – User profile Performance completion models Inventory levels Revenue per user Dictated constraints 20 Typical Constraints • Constraints applied – – – – – By particular network By age (i.e. users age 20 or older) By gender (i.e. females only) By geographic location (city or state) Offer can only be selected once per user. If selected, it should not be served again to particular user • Data used – – – – – – – Birthday Gender Zip Code Network Historic offers selected Offer details, profitability and goals Other... Offer Pool Response Modeling Customer Data Multi-Dimensional Optimization 21 Key Functionality • Marketing inputs – required conversions – ability to model – – – – – deal term included networks targeted demographics financials required click-throughs and impressions • Calculate impressions to achieve benchmarks • Ability to update in real time • Ability to use transactional data sources 22 Expected Benefits “Third” page real estate put to optimal revenue generation use – Snowball presents promotional offers mathematically proven to maximize revenue • cross-network registrants get promotion offers that are relevant to them – real world business constraints are met – marketing team can focus on other efforts • versus “If X and D, then A, B and C” 23 Conclusion • Constraint-based optimization is required to maximize a function – personalization is a tool…optimization puts it to work to achieve financial goals – ability to deal with constraints is critical • Optimization unlocks value from business agreements $ $ $ – many promotional partners $ $ – many sites – many registrants $ $ $ $$ 24 Outbound Marketing Optimization Cross Selling Case Study Cross Selling High Speed Data and Complementary Telephone Services Cross Selling credit cards, affiliate programs, balance transfer programs, insurance... 25 Impact pact • Revenue increases ranging from 8-20% ImpB2B Challenges • Significant reduction of management time in developing offer assignments • Scalable and adaptable as results develop. 26