The Clear Advantage: Using Artificial Intelligence to Create a Learning Model for Training
Preston Faykus, Founder & CEO RankMiner Predictive Analytics [email protected] 469.387.3865
Introduction – Who is RankMiner
• RankMiner, Inc. a US corporation headquartered in St. Petersburg, FL www.rankminer.com
• Founders specialized in data and voice analytics with a focus on innovative applications built with machine learning
Overview 1. Introduction to Artificial Intelligence (“AI”)
2. Bird’s-eye overview
3. Current applications
4. Innovations made using AI relating to call centers
5. Leading developments
Introduction to AI What AT is like today… • Process enormous amounts of data
• Classify new observations according to past observations
• Forecast details of future events
• Solving complicated – but practical! – problems
What AI can’t yet do… • Making connections at an abstract level
• Higher order reasoning
• Self-actualization
The Power of Using Artificial Intelligence Tsunami of data in Call Centers
• Call Volume – thousands of calls made per day
• Metadata – tens to hundreds of
thousands bits of information generated per day
• Business Outcomes – defined and
discovered daily • How does it all Inter-relate? • How can humans analyze all of this
data fast and efficiently?
The Power of Using Artificial Intelligence
Artificial Intelligence or, Machine Learning is the process of designing, tuning and employing algorithms to identify patterns in large, seemingly unrelated datasets, and predict future patterns.
The Power of Using Artificial Intelligence
3 Main Types of Machine Learning
• Unsupervised: discovering hidden properties of data
• Supervised learning: classifying new data from known properties
• Reinforcement learning: making the best decisions now to maximize long-term reward
Innovations using Unsupervised Learning • Give cross-sell suggestions to consumers • Organize consumers into target markets / calling campaigns • Search engine optimization Basic Ideas: • Start with a population and determine distinguishing characteristics • Organize data with generic/suggested labels
Innovations using Supervised Learning • Proactively identify fraudulent transactions • IVR call routing • Speaker identification Basic Ideas: • Use historical labeled data as examples • Answer True/False or multiple choice questions
Innovations using Reinforcement Learning • Determine payment / cost adjustments • Adjust employee commissions / incentive programs • Guided sales negotiations Basic Ideas: • Many more choices in the decision-making process (typically) • No knowledge of ultimate result, only of immediate return
Well-known applications
• Unsupervised learning: Google search, Amazon ”if you like X…”, Netflix “Top picks for you”
• Supervised learning: BoA automatic fraud detection, Apple Siri, Microsoft WhisperID
• Reinforcement learning: Deep Blue, IBM Watson, Geico instant online quotes
AI Uses in Call Centers • Evaluate 100% of Your Call Volume
• Audio Analysis
– Extract and create audio Feature Vectors (FV)
– Identify Emotions and Behaviors – Know not only what was said, but also
how it was said
• Metadata Analysis – Algorithms used to identify key parameters of structured data – links vocal insights to data sources more
closely related to business outcomes – include pieces as part of feature vector – use other pieces to categorize business
outcomes into supervised learning choices (“classes”)
AI Uses in Call Centers
Predictive model building process
• Feature vectors paint a picture to each class
• Big data <==> huge number of
dimensions to base decisions on • End result: a decision-making
policy to take in new data (e.g. calls) and associate them with a class and ultimately the anticipated business outcome
AI Uses in Call Centers • Deploy Automated Analysis
Algorithms – Optimized Call Analysis
Algorithms – Optimized Metadata
Analysis Algorithms – All future calls and related
metadata are analyzed
• Deploy Optimal Prediction Model(s) – Can have one or more
prediction models – Predicts Business Outcomes
for all future calls
AI Uses in Call Centers • Call Center business
success is based primarily on interaction between customer & agent
• How to automatically improve opportunity for success? – Customer Predictive
Model – New Hire Predictive
Model – Agent Predictive Model
AI Uses in Call Centers • Analyze 100% of Floor
Agent calls • Use Agent Prediction
Model to identify Good / Average / Poor calls
• Quickly identify Floor Agents in need of additional training
• Implementations show 21.6% increase in Floor Agent revenue
AI Uses in Call Centers
• Soft Skills – Floor Agent Engagement – Empathy & Helpfulness – Active listening
• Correlate with: – Payment – Sells – Customer Satisfaction
• Identify Floor Agents in need of help…their success is your success
AI Uses in Call Centers • QA Staff can not evaluate all calls – Use predictions as a queueing
mechanism – Database driven scorecard
templates – Immediate insight for Supervisors
and Management
• Use existing QA process to improve predictive model – Verify model prediction – QA identified trends to improve
model performance
• Implementations show productivity improvements in Quality Assurance of over 50%
AI Uses in Call Centers
• Discover new factors leading to SUCCESSFUL business outcomes – Increase Floor Agent revenue
streams – Accelerate conversion cycle – Increase Floor Agent ROI
• Upgrade Business Practices – Hire Floor Agents with successful
attributes – Train Floor Agents for success – Quickly identify and react to
“bad” trends to prevent loss of revenue
– Quickly identify and react to “good” trends to increase revenue and build a successful culture
Leading Developments
• Automated Reinforcement Learning – Business Environments
constantly change – Prediction Model(s)
dynamically updated – Analysis Algorithms dynamically
updated
• Allows continuous uninterrupted evaluation of 100% of call volume 24/7
SUMMARY • Machine Learning based on AI is changing the way call centers operate – Analyze all data, including
audio for emotions and behaviors
– Discover new trends and factors for success
• Increase revenues – Improve Floor Agent success – Improve QA efficiency – Improve Management insight – Create a culture of success
Questions
Preston Faykus, Founder & CEO RankMiner Predictive Analytics [email protected] 469.387.3865