Actionable intelligence via_speech_analytics
- 2. Actionable Intelligence via Speech Analytics Dr. Ofer Shochet SVP Verint Systems July 2008 IBM Speech Technologies Seminar
- 3. Speech analytics transforms recorded customer interactions from idle data to actionable intelligence
- 4. CONTENT CATEGORIZATION Transcribe and index entire call and extract concepts Analyze impact on known issues (lower false alarms) Three Levels of Speech Analytics KEYWORD SPOTTING Spot 20-200 defined words Find isolated calls of interest (high false alarms) BUSINESS VALUE INTELLIGENCE ROOT CAUSE ANALYTICS Mine categorized calls and suggest root cause Find out what you do not know to look for
- 5. Word Spotting Categorization Root Cause Analytics Technician did not show Received wrong information Did not receive credit Offer not clear to customers Product does not work well Product is too expensive Perceived as better value Product quality driving churn Price attracting customers Large sample of customer interactions Customer complaints Interactions about new product offering Interactions involving competition Another way of looking at it: Word Spotting, Categorization, Root Cause
- 6. The Value of Speech Analytics • Delivers value from the “voice of the customer” – “Focus groups on demand” with a more complete view of the customer experience • Enhances Quality Monitoring – Evaluate calls that represent “what matters most” to you • Connects the contact center and the enterprise SalesSales Back OfficeBack Office FraudFraud CollectionsCollections Risk Management Risk Management ComplianceCompliance R&DR&D MarketingMarketing Contact Center Contact Center Intelligence from Customer Interactions Intelligence from Customer Interactions
- 7. Verint Analytics Addresses Key Business Issues
- 8. Customer Details • Fortune 500 Insurance provider with over 4 million customers • First call resolution at 60% • Abandonment rate of 28% • Customer service rating of “Poor” • No clear insight into why customer issues not resolved Customer Case Study • Improve First Contact Resolution First Contact Resolution
- 9. Classifies calls via automated speech recognition and categorization technologyIdentifies key reasons why customer issues were not resolved Customer Case Study How it works First Contact Resolution Success (65%)
- 10. Terms automatically surfaced indicating root causeSurfaces root cause of first call resolution issues Customer Case Study How it works “calling back about my claim” “calling back about my claim” “Check with my supervisor” “Check with my supervisor” “I don’t know”“I don’t know” First Contact Resolution “waiting for a claim form” “waiting for a claim form”
- 11. Solutions • Outdated policies reviewed and changed and agents were trained to fully understand them • Agents empowered to solve customer issue on first call • Integration of frontline transaction processing • Clarification of timelines on claim forms Customer Case Study First Contact Resolution
- 12. Results Customer Case Study 0% 20% 40% 40% 30% Before SpeechAnalytics AfterSpeechAnalytics Unresolved Calls 25% increase in First Call Resolution! First Contact Resolution
- 13. Additional Results • 83% improvement in average speed of answer • 68% improvement in their service level (% of calls answered in 30 seconds) • 25% improvement in abandonment • 20% reduction in average handle time • 15% reduction in seasonal call volumes • eliminated the need to hire 22 additional agents • greatly improved staff morale Customer Case Study First Contact Resolution
- 14. Speech Analytics Delivers the Power of Why How can I improve performance? • Review outdated policies • Empower agents • Revise claim forms • Improve frontline processing What am I analyzing? • First contact resolution Root cause of why my results are poor/excellent? • Agent knowledge • Agent empowerment • Outdated policies • Confusing claim forms Execute a Plan • Increase first call resolution by 25%
- 15. Customer Details • Credit card provider • Historical record of converting 65% of inbound customer inquires • Sales conversion rate stagnating in previous three years • Marketing currently testing new offers Customer Case Study • Pinpoint best (and worst) selling circumstances and behaviors • Improve up-selling/cross-selling capabilities • Increase closing rates Sales Effectiveness
- 16. Identifies the most effective approaches for agents when selling to customersClassifies calls via automated speech recognition and categorization technology Customer Case Study How it works Sales Effectiveness Success (65%)
- 17. Automatically detects sales success and failures based on key phrases and metadata Customer Case Study How it works Sales Effectiveness
- 18. Surfaces root cause of negative sales performanceTerms automatically surfaced indicating root cause Customer Case Study How it works “I am not sure that we offer that…” “I am not sure that we offer that…” “I’m confused” “I’m confused” “Are you interested in the choices I presented?” “Are you interested in the choices I presented?” Sales Effectiveness
- 19. Terms automatically surfaced indicating root causePositive behaviors are reinforced, negative behaviors are correctedSurfaces root cause of positive sales performance Customer Case Study How it works “May I ask you a few questions?” “May I ask you a few questions?” “the best deal for you is…” “the best deal for you is…” “this is a better offer because…” “this is a better offer because…” Sales Effectiveness
- 20. Solutions • Agents trained to engage in conversation to uncover what customer values • Agents trained in presenting offers appropriately • Marketing began providing competitive data to agents prior to campaign launch • Marketing revised offers based on findings Customer Case Study Sales Effectiveness
- 21. Results Customer Case Study 50% 60% 70% 80% 65% 77% Before SpeechAnalytics AfterSpeechAnalytics Sales Conversion Rates 19% increase in conversions! Sales Effectiveness
- 22. Speech Analytics Delivers the Power of Why How can I improve performance? •Train agents to qualify •Create simple marketing offers What am I analyzing? •The factors that drive success or failure in sales calls Root cause of why my results are poor/excellent? • Agent knowledge • Probing questions • Simplicity of offers Execute a Plan • Increase closing rates by 19%
- 23. KPI Past Performance Performance three months after Verint deployment Impact Quality Scores 70% 81.20% +16% Improvement Revenue Per Call $0.33 $0.67 +103% Improvement First Call Resolution 76.8% 79.1% +3% Improvement Customer Satisfaction /Executive Complaints 8700 @ $82 Countless occasions to be proactive Potential saving of $713K Manager Productivity 1-2 call evaluations/month 5-6 call evaluations/month +300% Improvement Customer Churn Analysis underway Analysis underway 25% reduction Speech Analytics Delivers Quantifiable ROI Communications Provider
- 24. Analyst Praise for Verint Analytics “Saddletree Research views the Verint approach to speech analytics managed services as the most comprehensive and efficient offering on the market today…Verint has set the competitive bar” Paul Stockford - Saddletree Research
- 25. Why Verint Speech Analytics? • Automated root-cause – Delivers the Power of WHY • Integrated recording and QM platforms – Lower TCO and future proof • #1 Market Leader in Speech Analytics – Market proven ROI – Expert turnkey service offering
Editor's Notes
- #10: The first step in using speech analytics to improve sales effectiveness was to first separate calls with sales opportunities from calls that were about other issues. We were able to do this with a combination of speech analytics and metadata. In this case it was about a 50/50 split. Once we were able to classify calls with sales opps from other calls we then used speech analytics to determine which calls closed with a successful sale. Keywords from agent scripts were used and were automatically detected via speech analytics. For example, there was a disclosure statement that was required to be read at every sale. This made it easy to find sales calls. We also found calls that were not successful (once again in the sales opps bucket). Agents also had scripts they were required to read in the event of an offer rejection. It worked out to be about a 65/35 split, along the lines with their historical data.
- #11: For the unsuccessful calls a couple of key phrases were identified via automated root cause analysis. Terms like “I am not sure we offer that..”, I’m confused”, and “are you interested in the choices I presented” were automatically found by the system. The business users did not know to search for those terms but the system found that they were more frequently said in failure calls than all other calls in the contact center. Similarly, on successful calls there automated root causes. The terms ““May I ask you a few questions?”, “the best deal for you is..”, “this is a better offer because” all were found by the system to be more frequently said in success calls than all other calls.
- #12: These findings were eye-opening to the business. They had assumed that successful agents just followed script. Instead what they found was that deviating from the script and engaging in conversation was what led to sales. So, they redesigned their sales process. Agents were retrained to engage in conversation and present offers appropriately Marketing began providing competitive data to agents prior to campaign launch and revised their offers based on these findings to make them less complicated.
- #13: The results were astounding. After implementing these changes sales conversions jumped from 65% to 77%, which was an all-time high and an improvement of 19%. Without speech analytics there would have been no way to uncover the root cause of sales effectiveness. But with it, the reasons were quite clear. And the solutions were actionable.
- #14: These findings were eye-opening to the business. They had assumed that successful agents just followed script. Instead what they found was that deviating from the script and engaging in conversation was what led to sales. So, they redesigned their sales process. Agents were retrained to engage in conversation and present offers appropriately Marketing began providing competitive data to agents prior to campaign launch and revised their offers based on these findings to make them less complicated.
- #15: To summarize, this customer used speech analytics to analyze the factors behind why sales are successful Automated root cause analysis surfaced the reasons why The findings pinpointed specifically how to improve And the customer executed a plan to enact changes
- #17: The first step in using speech analytics to improve sales effectiveness was to first separate calls with sales opportunities from calls that were about other issues. We were able to do this with a combination of speech analytics and metadata. In this case it was about a 50/50 split. Once we were able to classify calls with sales opps from other calls we then used speech analytics to determine which calls closed with a successful sale. Keywords from agent scripts were used and were automatically detected via speech analytics. For example, there was a disclosure statement that was required to be read at every sale. This made it easy to find sales calls. We also found calls that were not successful (once again in the sales opps bucket). Agents also had scripts they were required to read in the event of an offer rejection. It worked out to be about a 65/35 split, along the lines with their historical data.
- #18: Now that we know what, the important question is Why? Why do sales calls sometimes close and sometimes don’t? What do successful agents do differently than unsuccessful agents? Is there something in our process that could be tweaked? Do customers respond to the offers? Do we have the right offer? Too many variations? To few? This is what Speech Analytics can help us uncover.
- #19: For the unsuccessful calls a couple of key phrases were identified via automated root cause analysis. Terms like “I am not sure we offer that..”, I’m confused”, and “are you interested in the choices I presented” were automatically found by the system. The business users did not know to search for those terms but the system found that they were more frequently said in failure calls than all other calls in the contact center. Similarly, on successful calls there automated root causes. The terms ““May I ask you a few questions?”, “the best deal for you is..”, “this is a better offer because” all were found by the system to be more frequently said in success calls than all other calls.
- #20: The business examined the reasons why these terms were being said and found that in the failure calls terms like “I am not sure we offer that” signified that agents were just taking orders instead of being consultative. On calls where customers said “I’m confused” and “are you interested in the choices I presented” , they were being overwhelmed by offers. On successful calls there automated root causes. The terms ““May I ask you a few questions?” signified that the agents were engaging in conversation instead of strictly reading a script The term “the best deal for you is..”, surfaced agents that actually didn’t read all offers but picked the one most relevant to the customer. And finally “this is a better offer because” signified that successful agents were actually conducting their own research
- #21: These findings were eye-opening to the business. They had assumed that successful agents just followed script. Instead what they found was that deviating from the script and engaging in conversation was what led to sales. So, they redesigned their sales process. Agents were retrained to engage in conversation and present offers appropriately Marketing began providing competitive data to agents prior to campaign launch and revised their offers based on these findings to make them less complicated.
- #22: The results were astounding. After implementing these changes sales conversions jumped from 65% to 77%, which was an all-time high and an improvement of 19%. Without speech analytics there would have been no way to uncover the root cause of sales effectiveness. But with it, the reasons were quite clear. And the solutions were actionable.
- #23: To summarize, this customer used speech analytics to analyze the factors behind why sales are successful Automated root cause analysis surfaced the reasons why The findings pinpointed specifically how to improve And the customer executed a plan to enact changes
- #24: Speech analytics has been vital to many organizations and has a quantifiable ROI. The communications provider in this slide has achieved the following results: