Utilizing Text Analytics in Your VOC Program Analyzing
- Slides: 28
Utilizing Text Analytics in Your VOC Program: Analyzing Verbatims with Poly. Analyst™ Sergei Ananyan Megaputer Intelligence (812) 330 -0110 sananyan@megaputer. com © 2007 Megaputer Intelligence
Outline • Project Highlights • Value of Verbatim Analysis • Historical Process and Need for Text Mining • Capabilities of Poly. Analyst™ • Text Analysis and Report Generation • Benefits of Text Analysis © 2007 Megaputer Intelligence
Customer: Hospitality Company XYZ • Global leader in development, operations and sales of Vacation Ownership resorts • Over $1. 5 Billion in sales • More than 300, 000 Timeshare Club Owners • Distinctive resorts with more than 8, 000 villas © 2007 Megaputer Intelligence
XYZ Company Surveys • 3 main areas of surveys: – Operations experience – Sales and marketing experience – Service experience • 10 surveys run on a constant basis • Mixed structured and open-ended questions • Guest Satisfaction Survey (GSS) – 100, 000 responses per year • Sales and Marketing survey – 150, 000 responses per year © 2007 Megaputer Intelligence
Guest Satisfaction Survey • Offered to – – – Owners Guests Rental guests Owner exchangers (non-XYZ) Preview package guests • 10 open-ended questions accompanying structured questions – e. g. What was your overall impression of the resort property? – (If rated 7 or below): What causes you to feel that way? • Questions follow customer touch-point map – Pre-arrival, online, gate house, check-in, wake-up calls, service, landscaping, restaurant, owner’s seminar • Goal: provide actionable feedback for onsite managers © 2007 Megaputer Intelligence
Need for Text Analysis © 2007 Megaputer Intelligence
Value of Verbatim Analysis • Go beyond structured questions – very limited information • Listen to what XYZ customers have to say – in their own words • Tie quantitative scores to verbatim comments • Provide proactive and actionable means for improvement at the Division, Site, and Regional Level • Define what topics are reported at varying levels of satisfaction • Assess whether XYZ is asking the right questions © 2007 Megaputer Intelligence
Historical Text Analysis Process • XYZ was categorizing verbatims from all surveys manually Read each response Manually select categories © 2007 Megaputer Intelligence
Challenges of Historical Process • Helped create initial Category Map: 4 levels and 217 categories BUT • Required a person to read each comment and assign categories • Different processes were used; No consistency • Slow: it was taking one hour to read and assign 100 comments • Reports were manually created • Addition of new categories was based on human interpretation • To handle the analysis of verbatims, XYZ needed a Text Mining tool © 2007 Megaputer Intelligence
Requirements for Text Mining Tool • Import survey results data and run word extractions on text • Create categories (or buckets) to group similar comments • Define patterns for automated text categorization • Perform automated categorization of text responses • Delineate positive/negative comments • Save reusable analysis scenarios for future categorization projects • Run extractions against the uncategorized comments © 2007 Megaputer Intelligence
Requirements for Text Mining Tool • Export categorization results and link back to specific comments • The output must be compatible with standard reporting tools • Added bonus: a scheduling component – At scheduled date/time it would retrieve/categorize data • Provide insight into ratings and comments reported – For example, which words are most frequently reported when the customer provides a structured score 3 or below? • Ability to create a custom thesaurus that would group frequently reported words that relate to the business – e. g. room, villa, suite, condo, etc. © 2007 Megaputer Intelligence
Poly. Analyst™ text mining tool • Knowledge discovery tool for business users • Easy-to-understand actionable results Data Overload Useful Knowledge Poly. Analyst © 2007 Megaputer Intelligence TM
Capabilities of Poly. Analyst • Unlocks value hidden in massive volumes of data and text • Solves all typical text analysis tasks: – – – – Categorization Clustering Taxonomy building Entity extraction Natural language search Multi-dimensional reporting Visual link analysis • Enterprise level scalability • Visual creation of analysis scenarios • Interactive visualization and drill-down • Executive reports © 2007 Megaputer Intelligence
Poly. Analyst extra features • In addition to meeting all requirements of XYZ, Poly. Analyst offered the following extra features: – – – – Automated spelling correction Words and patterns search Ability to discover unexpected issues Ability to automatically build taxonomies Dictionary editor for synonyms, abbreviations and stop-words Interactive reports for sharing results with business users Substantial ROI © 2007 Megaputer Intelligence
Survey Analysis with Poly. Analyst Data Analyst Collecting & Storing Data Automated Text Analysis Generating Reports Decision Maker © 2007 Megaputer Intelligence
Step 1. Data Analysis © 2007 Megaputer Intelligence
Poly. Analyst Analysis Scenario © 2007 Megaputer Intelligence
Text Categorization © 2007 Megaputer Intelligence
Text OLAP © 2007 Megaputer Intelligence
Step 2. Reporting Poly. Analyst for Business Users © 2007 Megaputer Intelligence
Site Manager’s Report: Food & Beverage © 2007 Megaputer Intelligence
Site Manager’s Report: Villa Cleanliness © 2007 Megaputer Intelligence
Benefits of Text Analysis with Poly. Analyst © 2007 Megaputer Intelligence
Benefits • Extracting value from massive volumes of text • Dramatic reduction in the cost of data analysis • Increase in quality and speed of the analysis – Poly. Analyst successfully categorized 95% of text verbatims – The analysis time dropped from 1, 000 hours to 10 minutes per survey • Automated monitoring of data for known problems • Timely discovery of emerging issues and trends • Joint analysis of text and structured data • Objective and uniform data-driven analysis • Delivering interactive report to decision makers © 2007 Megaputer Intelligence
Return on Investment • Guest Satisfaction Survey – 100, 000 responses per year (for all questions) • XYZ runs 10 surveys annually • Manual analysis – It takes over an hour to manually process 100 verbatims – Manual analysis of all verbatims would take over 10, 000 man-hours – Projected annual cost of manual analysis of text responses - $500, 000 • Based on $50 per hour gross cost – Projected 5 year cost of manual analysis - $2, 500, 000 • Poly. Analyst analysis – Categorization of all verbatims takes an hour of machine time • upon the initial taxonomy setup – 5 year cost of Poly. Analyst survey analysis process – less than $400, 000 • 5 year Poly. Analyst savings - $2, 100, 000 © 2007 Megaputer Intelligence
Handled Business Tasks • Survey data analysis • Call Center data analysis • Repair notes analysis • Incident report analysis • Claims notes analysis • E-mail target routing • Competitive intelligence • Fraud detection • Intellectual property research © 2007 Megaputer Intelligence
Select Customers Government Insurance Financial High Tech Pharmaceutical Marketing Manufacturing © 2007 Megaputer Intelligence
Next Steps Call (812) 330 -0110 or write sananyan@megaputer. com 120 W Seventh Street, Suite 314 Bloomington, IN 47404 USA © 2007 Megaputer Intelligence
- Making connections
- Listening strategies
- Compression tackle
- Text analytics and text mining
- Text analytics and text mining
- "amplitude" analytics or "product analytics"
- Letter to viceroy lord irwin answers
- What is the horror genre by sharon russell answer key
- The tell tale heart assessment answers
- Against nature essay
- Bonne annee story timeline
- Without a title poem
- Write the answers in your notebook
- Power bi text analytics
- Text analytics summit
- Idol text analytics
- Text analytics world
- Text analytics forum 2019
- Wziu hub
- Jmp text explorer
- 51 ppt
- Text analytics unipi
- Voc ölçümü nerelerde yapılır
- Tech voc definition
- What is voc
- Skular voc
- Pembuatan pabrik senjata di semarang dan surabaya
- Eu gosto de voc
- "technology process management" tpm