Intelligent Interactions: Improve Response Rates by Getting to Know Your Customers Through Data Analytics
0 likes378 views
The document discusses the use of data analytics and predictive modeling to enhance telemarketing strategies and improve customer acquisition. It highlights trends in the contact center industry, such as the shift to multi-channel communication and the need for customized solutions based on detailed customer profiling. Key findings show significant improvements in response rates and revenue through tailored offers and intelligent routing of customer inquiries.
Intelligent Interactions: Improve Response Rates by Getting to Know Your Customers Through Data Analytics
- 2. Intelligent InteractionsIntelligent Interactions Improve Response Rates by Getting to KnowImprove Response Rates by Getting to Know Your Customers Through Data AnalyticsYour Customers Through Data Analytics Steve BrubakerSteve Brubaker Chief of StaffChief of Staff InfoCision Management Corp.InfoCision Management Corp. www.infocision.comwww.infocision.com
- 3. Agenda •The impact of modeling on acquisition •Using business intelligence to drive results •Online lead generation •Multi channel marketing using business intelligence
- 4. Top trends in the contact centerTop trends in the contact center industryindustry 10. Cell phones – erosion of landlines10. Cell phones – erosion of landlines 9. Trend back to the phone call – technology is driving9. Trend back to the phone call – technology is driving down call center costs while paper and postal costs aredown call center costs while paper and postal costs are increasing direct mail costs.increasing direct mail costs. 8. VOIP – Voice Over Internet Protocol8. VOIP – Voice Over Internet Protocol
- 5. Top trends in the contact centerTop trends in the contact center industryindustry 7. Salaried contact center agents7. Salaried contact center agents 6. Highly/Specially trained agents with ability to free flow6. Highly/Specially trained agents with ability to free flow conversations and not always work off a scriptconversations and not always work off a script 5. Skill based inbound customer service – impacts up-5. Skill based inbound customer service – impacts up- selling and cross-selling. Inbound doesn’t make $. By up-selling and cross-selling. Inbound doesn’t make $. By up- selling and cross-selling you can make $.selling and cross-selling you can make $.
- 6. Top trends in the contact centerTop trends in the contact center industryindustry 4. Social Media monitoring in the call center4. Social Media monitoring in the call center 3. Work at Home Agents/Virtual Contact Centers3. Work at Home Agents/Virtual Contact Centers 2. Multimedia communication channels – blending email,2. Multimedia communication channels – blending email, chat, phone. Agents are expected to interact at differentchat, phone. Agents are expected to interact at different levels.levels.
- 7. Top trends in the contact centerTop trends in the contact center industryindustry 1. The use of data analytics to develop a1. The use of data analytics to develop a multichannel approach to reach out to a widemultichannel approach to reach out to a wide variety of consumers in the most personalizedvariety of consumers in the most personalized and effective way.and effective way. Tweet questions or comments with hashtag #ATAdata
- 8. Traditional direct marketing often was like trying to force a square peg into a round hole. Today, a customized solution is the only cost effective approach.
- 9. The Implementation and Impact of PredictiveThe Implementation and Impact of Predictive Modeling on Telemarketing AcquisitionModeling on Telemarketing Acquisition Case StudyCase Study
- 10. •Many clients traditionally use rental or exchange lists for acquisition efforts •A 20% success rate is typical •The goal is to develop and use a predictive model to improve results utilizing rental lists
- 11. •First step: •Apply the model to rental lists to develop segmentation strategies •Improve performance and drive down costs •Second Step: •Improve performance and drive down costs through dynamic request strategies
- 12. Define the current customer base with profiling Apply the model to rental list and segment prospects Model the current customer base to target for acquisition •First step: Tweet questions or comments with hashtag #ATAdata
- 14. •First step: • Analyze current customer base and define key demographic and psychographic attributes: • Age • Education Level • Home Value • Income • Family Position • Gender • Create “Model” donor • Overlay model onto response list and score prospects
- 15. The Implementation and Impact of BusinessThe Implementation and Impact of Business Intelligence on Telemarketing AcquisitionIntelligence on Telemarketing Acquisition
- 16. •Second step: •Now that the audience is scored and segmented • How do we now impact the offer? •Analyze various affluence indicators and their relationship to offers •Apply this information to develop a dynamic offer utilizing variable scripting technology Tweet questions or comments with hashtag #ATAdata
- 17. •Findings: • Household income displayed the highest correlation to gift amounts • Household incomes were then broken into five income bands ranging from low to high • Each income band was given a specific gift ask • The key metrics we were looking to influence were: • Response rate • Average gift • Dollars per call • Efficiency
- 18. •Dynamic GRC results against control: •Revenue per call increased by 27% •Response rate increased by 16% •Average gift increased by 11% •Also showing an increase were credit card rates at 12% •Not only were gross conversions impacted but stick rate and ROI dramatically improved
- 19. Superior Lead Generation throughSuperior Lead Generation through Real Time Scoring and Targeted RoutingReal Time Scoring and Targeted Routing Online Application StudyOnline Application Study
- 20. Tweet questions or comments with hashtag #ATAdata
- 21. Here’s how R3 works: Fast Response A request comes in from your website Quick Routing An InfoCision communicator promptly contacts the lead Intelligent Transfer Calls are transferred to agents or counselors if needed
- 22. •Step 1: Potential customer clicks on online ad or webpage and is directed to online application •Step 2: Customer fills out form and presses “contact me” option •Step 3: Self reported data is “pinged” against the consumer database to append additional demographic information
- 23. • Step 4: Customer data is then scored against pre-built model • Step 5: Offer is customized and/or altered based on score • Step 6: Call is directed to appropriately skilled communicator and an outbound call is generated and routed • Step 7: Calls are transferred to agents or counselors if needed Tweet questions or comments with hashtag #ATAdata
- 24. Market Applications: Education Student requests information about specific campus or educational program Financial Prospect requests more information about a specific type of loan or offer Commercial Customer expresses interest in a specific product line or service Calls are routed to Agents or Counselors who are trained and knowledgeable on those specific products and markets
Editor's Notes
- #11: Of the lists that we rent, we only see that 20% perform to the degree that we can roll them out. Using a model, we can score this universe and determine which records are the best to call thus improving the number of lists that we can roll out while at the same time improving their subsequent performance. This has the double impact of increasing the callable universe and increasing results.
- #13: We first overlay the data we wish to model and do a profile. This allows us to better understand the audience we are modeling and allows us to better understand the model results. We then will model the records using regression analysis to determine which attributes contribute most heavily to performance. These attributes are then scored so that when the model is applied the records we are applying it too can be scored. When a new list is brought in, we score it using the model and rank order the scores into deciles.
- #14: These are the different types of data that we can use in the model. Transactional is the one that is built based on prior results. We will look at transactional with a focus on our behavior of interest, i.e.. Recency, frequency, monetary. This will depend on the type of model we are building.
- #15: These are examples of attributes in the database that may be used in the model. There are no certain ones that will be used in each model as it will depend on their relationship to the behavior of interest that we are modeling. Below are the main attributes that we look at for a model: OVERLAY GROUPS  Group A Individual Information  1. Age Range202 2. Gender205 3. Married220 4. Estimated HH Income213 5. Census Education Level170 6. Race206 7. Family Position Code207 8. Image Children Present236 9. Number of Children230 10. Voter Party512 11. Net Worth Indicator514 12. Homeowner211 13. Religion Code311 14. Donor515 15. Donor Index516 16. Occupation Code (Group)237 17. Voter Indicator513 18. CBSA Code135 19. DMA Code131 20. Household Composition523 21. ZIP Level Household Income Decile V1.9286 22. Census Income Percentile182   Group B Housing Information  1. Length of Residence209 2. Dwelling Type212 3. Census Median Home Value186 4. Own/Rent317 5. Nielsen County Size140 6. Number of Persons in HH224 7. Online HH Access241   Group C Mail Response Information  1. Mail Order Responder215 2. Mail Order Buyer216 3. Mail Order Books370 4. Mail Order Books/Magazines371 5. Mail Order Children’s372 6. Mail Order Gifts376   Group D Credit Information  1. Credit Active217 2. Bank Card218 3. Retail Card219 4. Credit Cards: Premium AMEX331 5. Credit Cards: Premium DISC332 6. Credit Cards: Premium OTHE333 7. Credit Cards: Premium STR334 8. Credit Cards: Premium V/MC335 9. Credit Cards: Regular AMEX336 10. Credit Cards: Regular DISC337 11. Credit Cards: Regular OTHE338 12. Credit Cards: Regular STR339 13. Credit Cards: Regular V/MC340   Group E Donor Information  1. Donor: Animal424 2. Donor: Arts/Cultural425 3. Donor: Children’s426 4. Donor: Environment427 5. Donor: Health428 6. Donor: Political Conservative430 7. Donor: Political Liberal431 8. Donor: Religious432 9. Donor: Veterans433   Group F Transaction Information  1. Internet Shopper591 2. Continuity Shopper592 3. Internet: Purchase Online369  Group G Interest Information  1. Veteran in HH342 2. Hobby: Self Improvement358 3. Music Pref: Christian/Gospel389 4. Music: Country391 5. Reading: Bible404 6. Reading: Children’s407 7. Reading: Computer409 8. Reading: Country 410 9. Reading: Medical414 10. Reading: Military415 11. Reading: Natural Health417 12. Reading: Sports422 13. Reading: World News423 14. Sporting: NASCAR445 15. Sporting: Hunting444   Group H Cluster Information  1. Health/Insurance Responder779 2. Mindbase Groups634 3. Mindbase Segments636 4. Mature Data Profiles595 Â
- #17: As a second step, we studied various economic indicators to determine if they had an impact on giving. We found that Household Income showed the strongest correlation to giving. We used this information to derive variable gift asks. Other indicators we looked at were net worth, home values, education, and zip level income percent. It is my belief that income is the best indicator due to the nature of our business. We are asking for a monetary response in a matter of 5 seconds. People will quickly think about how much money they have readily available to give. It is also my belief that net worth is a better indicator for direct mail as people do not have to make a split second decision and as such think about what they can afford in a broader scope.