Post on 25-Jun-2020
, MACHINE LEARNING: KNOWLEDGE FOR THE MASSES Matt Seaman
Knowledge Centered Service
Dynamic Content in the Customer’s Workflow
Predictive Customer Service
OF CUSTOMER INTERACTIONS ARE INITIATED BY SUPPORT 80
2020 Vision
ENABLERS
IoT Connected
products / software
KCS Knowledge
captured / relevant
Machine Learning
Augmenting Humans
connecting people to contextual information at massive scale
THE FOUNDATION: KCS
600+ knowledge workers
18% new issue rate
42 knowledge domain experts
129,000 public articles
211,000 total articles
Questions
Online
Knowledge Centered Service
Dynamic Content in the Customer’s Workflow
2011 2016
155,000
108,000
107,000
615,000
10 % 469 %
0.9 % deflection in Case Logger
“Search” technology does not work well
Dylan A neural network that learns patterns of words and assigns vectors
6 Years of Case Descriptions
Natural language
Engineers’ Associations with Articles
Customer description PTC Knowledge Article
Dylan
“Please help poor performance” Bing Translator into Klingon
33% deflection w/Dylan
0.9 % Deflection pre/Dylan
45,000 Cases resolved
30% Cases reduction
November 2016 October 2018
35% Dylan accuracy rate
Dylan
$Millionsinsavings
Predictive Customer Service
OF CUSTOMER INTERACTIONS ARE INITIATED BY SUPPORT 80
Paid service
always-onmonitoringsmart&connectedsystems
Proactive Monitoring
manualuploadonsupportwebsiteOn – Demand System Scan
Part of standard support
PREDICTIVE CUSTOMER SERVICE
Listen at IoT scale
Rules Data Processing
KCS Methodology (modified) • High Risk • Follows KCS process • Heavy review / inspection • Limited publishers
Creation of Rules • Hardcoded rules • Machine run but not self learning
Customer Value • Tailored, specific, impactful recommendations
PREDICTIVE CUSTOMER SERVICE
SOLUTION SENT
THE DATA-DRIVEN JOURNEY
KCS Robust knowledge capture is the required baseline
IoT Listening to People, Software, & Machines provides a new input
Processing data at massive scale provides complete new value streams
Machine Learning