The consumer now has more information to make decisions with multiple points of contact which demands a more refined targeting strategy.Shifting $ from mass to individually targeted & engaged medias is the new frontier of competitive advantage…..this is creatinga new CRMrevolution.
The reason to buy (DB) is changing (1.0 DB/DP to inform campaign targeting => 2.0 DbM measurement and insight to inform strategy and $ allocation)IMO (DB layer, Tech layer and services layer)
Results over the last 12 months deteriorating, and Life Line working with Merkle to review its targeting strategies.
Detailed analysis of the promotion history data showed significant performance difference between prospects that received prior contacts vs. those who did notPromotion history provided 50% of the explanation in the response variable. As a result, the decision was to develop separate prior and no prior models
Driving Results through Strategic Data Sourcing and Optimization: Life Line Global Case Study
October 5th, 2011<br />Driving Results through Strategic Data Sourcing and Optimization: <br />Life Line Global Case Study<br />Trish Mathe – Vice President of Database Marketing, Life Line Screening<br />Ozgur Dogan – General Manager, Data Solutions Group, Merkle<br />
Presenter Backgrounds<br />Trish Mathe<br /><ul><li>Vice President of Database Marketing at Life Line Screening
Over 10 years of database marketing experience both in financial services and healthcare industries
Areas of expertise include: building and maintaining marketing infrastructure and automation, prospect and customer database management, campaign management and measurement
Experienced in marketing to the fifty plus crowd, healthcare professionals, and several other specialty market segments</li></ul>Ozgur Dogan<br /><ul><li>General Manager of Data Solutions Group at Merkle
Oversees the delivery of analytical data sourcing and optimization solutions for Merkle’s clients across all industry verticals
Spent 7 years at Merkle and has 15 years of industry experience in building, implementing and integrating database marketing solutions
Technical MBA Degree from the University of Georgia</li></ul>2<br />
Session Overview<br />Evolution in the CRM Data Landscape<br />Developing a quantitative framework to assess value of data<br />Future Trends and Innovation Opportunities<br />Life Line Data Sourcing & Optimization Case Study<br />3<br />
Global Market Trends<br />Fundamental changes in the consumer decision making and buying process<br />Advancing and evolving technology use<br />Expanding fragmentation – media and channels<br />Data explosion driven by emergence of digital media<br />Clutter and confusion in the data landscape<br />Increased Accountability and Measurement<br />Ultimately, these influencers are changing the way marketers will create competitive advantage in the future.<br />5<br />
Consumers are More Connected Today than Ever<br />Blog<br />Search<br />Email<br />27% actively read blogs<br />86%<br />87%<br />87% use email 1+ times per day<br />86% use search frequently<br />27%<br />Social<br />Display<br />63% use Facebook weekly<br />20% click on banner ads<br />63%<br />Mobile<br />IM<br />51%<br />20%<br />33%<br />33% use IM regularly<br />51% are active texters<br />6<br />
Database Marketing Landscape is Evolving<br />DbM 1.0<br />DbM 2.0<br />Direct/Identified Model<br />New Entrants <br />Domestic<br />US and International Solutions<br />Single Campaign/ Media Targeting<br />Integrated Media Optimization<br />Cost Pressure<br />Increased Cost Pressure<br />Offline focus<br />Digitalization<br />Key Trends<br />7<br />
Data Explosion!<br />Today, the codified information base of the world <br />is believed to double every 11 hours<br />15 out of 17 sectors in the United States have more data stored per company than the US library of Congress<br />“We create as much information in two days now as we did from the dawn of man through 2003.” <br />Eric Schmidt, Google CEO <br />“Organizations are overwhelmed with the amount of data they have and struggle to understand how to use it to drive business results.” (2010 MIT Sloan/IBM Study)<br />8<br />
Business Impact of Analytical Data Sourcing<br />Leading direct marketer saved $2 MM in list sourcing cost in it first four 4 months through analytical data sourcing optimization without negatively impacting response<br />2010 Costs 2011 Costs Savings<br />10<br />
Common Data Types and Constraints<br />Type of Data<br />Examples<br />Common Constraints<br />Compiled & Aggregated Data<br />Experian INSOURCE, Epsilon TotalSource, Data Source<br /><ul><li>Can only afford one source
It is difficult to determine unique value so only purchase single source</li></ul>Syndicated Research<br />MRI, Scarborough<br /><ul><li>Unable to implement beyond basic messaging and product design</li></ul>Vertical Lists<br />New parents, magazine subscribers<br /><ul><li>Too many choices on the market, hard to evaluate
Selection limited to a small number of data card attributes</li></ul>13<br />
Analytical Data Sourcing and Optimization<br />14<br />
How to Assess the Value of Data<br />Framework<br />Predictive Power<br />Descriptive Power <br />Composite Score<br />Source Quality<br />Universe Coverage<br />Key Dimensions for Evaluation:<br /><ul><li>Predictive Power: Does the source add incremental lift to my predictions?
Descriptive Power: Does the new source provide the ability to better segment my target audience or lend new insights?
Universe Coverage: Does the source provide access to new and unique prospects (or overlay to existing customers)?
Source Quality: Does the source provide accurate and high quality data? </li></ul>15<br />
Evaluating Value of Data Sources - Example<br />Key Dimensions for Evaluation<br />Predictive Power<br />Descriptive Power <br />Example<br />Composite Score<br />Source Quality<br />Universe Coverage<br />17<br />
Analytical Data Sourcing & Optimization<br />Analytical Data Sourcing<br />Traditional Data Sourcing<br />Incentive<br />Incented to increase list <br />performance and <br />reduce list costs<br />Incented to increase list<br />volume<br />Fully aligned with Client’s <br />cost efficiency and growth goals<br />Not fully aligned with Client’s <br />business goals<br />Alignment<br />Analytically Driven Optimization<br />Approach<br />Recommendations driven by<br />Experience and Relationship<br />Recommendations<br />Team<br />Dedicated Team focused on <br />Driving performance<br />Driven to increase commissions<br />Analytics<br />World Class Analytics Team with <br />dataoptimization experience<br />No real analytics or science <br />18<br />
List Optimization Dynamics<br />The purpose of the list optimization process is to balance cost and value<br />19<br />
Analytic Approach to List Universe Optimization<br />Existing Universe Lists<br />Future Universe Lists<br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />List <br />“N” lists<br />Merkle’s approach is to inform the source /list pool and universe optimization process with analytics to define the right mix and number of lists that maximize ROI<br />ROI<br />N lists<br /># of Lists<br />20<br />
Optimized Source Mix Illustration<br />The ratio of the Base File names increases in the optimized source mix scenario<br />21<br />
Optimization Performed At Multiple Levels<br />LEVEL 1 <br />LEVEL 2<br />LEVEL 3<br />Today’s Focus<br />22<br />
Data Sourcing and Optimization As Enabler of Customer Centricity<br /><ul><li>Effective ICM™ demands a broad set of core competencies in order to be effective. Data plays a central role in delivering on the vision of ICM.
Understanding the optimal mix of data, both third party and customer enables optimal analytics.
Analytics informed effectively through data enables segmentation, customer optimization, marketing mix, media targeting, and predictive modeling in support of the four functional areas within ICM.</li></ul>25<br />
Data Sourcing As Strategic Engagement<br />Phase I - Evaluation<br />(Months 0 – 3)<br />Phase 2 - Implementation<br />(Months 3+)<br />Establish KPI’s<br />List Optimization<br />Illustrative<br />Simulation/Optimization on Historical Campaigns<br />Refine Optimization Models<br />Evaluation of New Compiled & Vertical Sources<br />Early Harvest<br />Execute Test Campaign<br />Eliminate list sources with high duplication rates<br />Develop list optimization tool<br />Rollout<br />Optimized list sourcing for Highlights (incl. brokerage services) <br />Strategic data research and analysis <br />26<br />26<br />26<br />
List Optimization Engine Automates the Process<br />27<br />
Economic and Environmental Data Integration<br />Economic and Environmental Data<br />Examples<br /><ul><li>New house starts and vacancy rates
Unemployment rate and per capita personal income
Disaster areas</li></ul>Business Impact<br /><ul><li>Better targeting of products and services that are sensitive to environmental factors
More predictive media mix optimization and allocation models
Ability to explain performance changes due to environmental factors</li></ul>28<br />
Digital Data Innovation and Integration<br />29<br />
Key Take Aways<br />CRM data landscape is changing rapidly due to digital media emergency and data explosion<br />Innovative optimization approach delivers ROI by reducing data costs and increasing marketing performance<br />It’s important to cut through the clutter and identify the most valuable data assets in the market place including newly emerging sources like digital<br />Integrating analytics expertise with data market knowledge is necessary to gain access to best and most comprehensive marketable universe<br />30<br />
Advised to share with physician for appropriate follow-up.
If anything critical participant is provided a “Doctor’s Review Kit” immediately and advised to go to a physician or emergency room within 24 hours.</li></ul>Participant Screened At Local Venue: Church, Club, Community Center<br />Screening Scheduled<br />Results are reviewed by a board certified physician <br />33<br />
Life Line’s Global Expansion Strategy<br />34<br />
Life Line Projected Global Presence<br />35<br />
Life Line Business Challenge<br /><ul><li>Interested in rapidly growing the customer base in US and across the globe
Using multiple compiled lists provides support to the large-scale Direct Mail acquisition program
Limited universe and heavy mailing volume causing contact fatigue
Applying the learnings generated in US to support the global expansion strategy with UK as the first pilot market</li></ul>36<br />
CRM Solution Roadmap<br />High<br />Targeting<br />Insight<br />Program Development<br />Measurement<br />Source Incremental P&L and Hierarchy<br />Integration of Promotion History <br />Prospect Segmentation<br />“Silo” Sources<br />Marcom Contact Strategy per Segment<br />Prospect and Customer level Insights<br />Brief knowledge on the 50-75 years old target population <br />Impact<br />LTV & Profitability Tracking @ The Customer Level<br />Integration of Sources<br />Multi-Source Interaction Campaign Approach<br />Creative & Source Testing<br />Single level source campaign level measurement<br />Phase I<br />Phase II<br />Phase III<br />Low<br />High<br />Program Sophistication<br />37<br />
Analytics and Targeting Solution for US<br /><ul><li>Started with an in-depth analysis of Life Line’s historical campaign data and quantified the impact of contact history on campaign performance
Learnings from the analysis were used to develop a segmented modeling strategy based on prior contact history that drove the selection of best prospect names
A new targeting methodology was developed and tested against the current compiled data vendors in a head to head test
Segmented modeling solution increased response rate by 38% and generated 62K incremental customers given the same mailing quantity</li></ul>38<br />
Analytics Solution Framework<br /> STEP 1 – PERFORM CONTACT HISTORY ANALYSIS<br /> STEP 2 – DEVELOP A PREDICTIVE MODELING SYSTEM<br /> STEP 3 – DEVELOP OPTIMIZATION ALGORITHM TO MAXIMIZE DIRECT MAIL CAMPAIGN PERFORMANCE<br />39<br />
Targeting Evolution – Gen3.0<br /><ul><li>LLS models continue to be redeveloped to keep current and the approach refined to gain incremental lift.
Gen3.0 segments out prior contacts from non-prior and also urbanicity. Promotion history as a predictor is removed and used outside of the model to remove bias that comes from having it in the model.
In head to head testing Gen3.0 is winning over Gen2.0 in 5 out of 7 campaigns and driving an incremental 6% improvement on average over an already strong Gen2.0 model.</li></ul>Modeling Approach<br />Gen1.0 – Gen3.0<br />40<br />
UK Predictive Modeling Solution<br /><ul><li>We developed a Modeling System consisting of multiple Customer Clone and Response Models to support Life Line’s UK business
Detailed analysis of the promotion history revealed that two separate response models were needed (Prior and No Prior) given the large performance differences between the two contact strategy segments
All of the models performed well and will provide a steady stream of high performing target prospects going forward</li></ul>41<br />
UK Modeling and Selection<br />Leveraging the learning's from the US:<br />A customer clone model is used to eliminate 50-75 year olds who do not look like current Life Line customer customers<br />Prospects are then separated between those who received an offer from Life Line in the past 12 months vs. those who did not<br />Segment-specific response models are used to improve identification of prospects with prior and no prior contacts<br />UK Models<br />National Canvas<br />50-75 yr olds<br />Customer Clone Model<br />Priors Response Model<br />No-Priors Response Model<br />Optimization Algorithm To Combine The Predictive Models<br />42<br />
UK Segmented Model – Summary<br /><ul><li>Modeling process identified the characteristics among each segment that best defined the responders
Predictors of response for households without prior contact: