1 Speech Analytics ROI Uncovering Key Business Intelligence

1

Speech Analytics ROI: Uncovering Key Business Intelligence Can Save Revenue From Dropping off the Bottom Line

Agenda September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California • Speech Analytics background – – – Why speech analytics? State of the art in speech analytics Analytics – more than words Typical application stack Application scaling • Case study – Reducing churn for broad band provider • Variables • Cost of churn • Methodology for saving revenue with speech analytics • Revenue saved • Return on investment • Sample applications • Summary 3 www. ITEXPO. com

Why Speech Analytics? September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California • 90% of customer contact outside of the storefront comes through the call center – Cable companies – Wireless & wire line carriers – Consumer goods • Most overlooked source of customer information – Over 225 million calls recorded daily – Less then 1/10 of 1% is ever analyzed – More than 223 million opportunities to analyze customer needs are being irretrievably lost. 4 www. ITEXPO. com

State of the Art in Speech Analytics September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California • Underpinnings: – Advanced speech to text engines – converting audio to minable words and data – Natural language processing – 2 nd order recognition using patterns – Auto identification of key words and phrases and other meta data • Applications: – – Dashboards for KPI reporting Trend analysis Visualization tools to uncover complex data relationships Advanced analytics to drill-down and across data sets 5 www. ITEXPO. com

Analytics - More than words September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California Conversation Analytics – Defining context in content – Words/Phrases together with: • • • Who is talking When is it said in reference to other words - temporal awareness Silence – Boundaries have meaning Tempo – rate of speech, emphasis Stress – Speaker identified important words – Meta-Data: • E. g. - customer number, purchase history, products purchased, preferences, how often the customer called, length of call, called number, call reason, inquiry, support, complaint, etc… 6 www. ITEXPO. com

Application Stack September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California Analytics: search, discovery, reporting, alerts Core. Engine Virtual Server Room Call Center Recording Systems 7 www. ITEXPO. com

Application Scaling September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California Site #2…n Site #1 Core. Engine Call Recordings Audio Database • Retrieve calls • Convert audio to Minable text and metadata Analysis Reporting Visualization Core. Engine Call Recordings Audio Database • Retrieve calls • Convert audio to Minable text and metadata Call. Miner Database Analysis Reporting Visualization 8 www. ITEXPO. com

Case study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – Reducing churn for broad band provider: • Variables • Cost of churn • Methodology for saving customers with speech analytics • Revenue saved • Return on investment 9 www. ITEXPO. com

Case study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – Variables: – $4 billion annual revenue – $200 per customer acquisition cost – $516 annual revenue/customer – 18% per year churn 10 www. ITEXPO. com

Case study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – The cost of churn: • $720 million/year • Only a 1% reduction in churn saves $20 million dollars from dropping off the bottom line • The $20 million could be used to recruit 200, 000 new subscribers • Takes ~ 5 months to recover the customer acquisition cost 11 www. ITEXPO. com

Case study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – Methodology for saving customers with speech analytics: Profile historical churn factors Produce churn risk report Retention group calls out to customers Customer Saved Customer lost 12 www. ITEXPO. com

Case study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California Profile historical churn factors: – Some speech applications can tell you what you don’t know – e. g. what are the top 50 reasons why customers are calling? • • Billing complaints Competitor mentions Service issues Hardware problems Cost Reliability Performance 13 www. ITEXPO. com

Case study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California Profile historical churn factors – tag cloud: yes, I'm calling to pay my bill, no it's not overdue, why did I have to wait so long. Yes, I'll order the basic service, what is the cost? this is the third time I've called to get help I want to speak to your supervisor yes thank you hours, I have a business to run, can't you help? the system has been down for two Yes I'm a business first customer the agent told me it would be fixed by 5 pm, and the system is still down I appreciate your help no, I don't want that service, it's too slow, I should have switched to satellite are you kidding me! The sales person said the service would cost only 40 dollars 14 www. ITEXPO. com

Case Study September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – Product churn risk report: 15 www. ITEXPO. com

Revenue Saved September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – Churn Reduction: • Current churn costs $720 million/year • Only a 1% reduction in churn saves $20 million dollars from dropping off the bottom line • Using a conservative outcome of. 25% churn reduction results in $5 million in saved revenue 16 www. ITEXPO. com

Case Study - ROI September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California – A conservative. 25% reduction from speech analytics yields $5 million in saved revenue – Implementation cost: • Number of call centers – 15 • Total seats – 5, 250 • Total implementation cost $2. 625 million – Time to recover costs 6. 3 months 17 www. ITEXPO. com

Sample Uses September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California • Identify Call Trends – Identify often used word & word combinations over time – ID calls and view trends – Search for “repeat” language • Classification – Categorize calls by type • Scoring – Validate that words or phrases were spoken • Reduce and/or Eliminate the Cost of Call Monitoring – Automate call scoring • Use Tireless Supervisor to automate the process of call evaluation 18 www. ITEXPO. com

Sample Uses September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California • Listen to the right calls – Call. Miner can immediately reduce the required time to review calls by an average of 50% – Auto selects calls that match (or don’t match) specific criteria • Reduce Calls into the Call Center – ID reasons for calls into the contact center – Use Trend. Miner and Tireless Supervisor to identify why customers are calling - specific categories of calls may be directed to other channels like the web or IVR. 19 www. ITEXPO. com

Summary September 10 -12, 2007 • Los Angeles Convention Center • Los Angeles, California • Make the most the enterprises most important assets – what customers are saying • Conversation analytics uncovers critical insight into what customers are saying and how agents are responding • Analyzing and reporting on actionable information can result in significant cost savings and opportunities to increase revenue • Vendor choice critical – not all speech analytics are created equal 20 www. ITEXPO. com

Cliff La. Coursiere, SVP Business Development cliff@callminer. com