Analysis of VOC (Voice of Customers)
Recently customers’ interest in and demand for use of consulting record data, analysis of VOC combined with big data, compliance with the Financial Consumer Protection Rules, and evaluation of banks according to the receipt of civil complaints by the Financial Supervisory Service are increasing. Companies are preparing for the Sense & Response Management System, which takes the lead to find the entire customer experience and makes a company-wide and rapid response to customers’ potential demand. In addition, it is becoming more difficult to effectively respond to and integrally understand customers’ opinions and complaints about products and services by daily using the Internet, mobile, social media, etc. In particular, evaluation and complaints of their products and services that are distributed through the Internet and social media but are not received by companies are becoming future risks that cannot easily be predicted.
Saltlux’s VOC Analysis Solution collects and integrates in real time VOC from various channels, such as consulting memos at the Customer Center, email, various social media, and portal bulletin boards. It helps secure a more intelligent and proactive leadership for customers, such as the development of new products, reputation management of companies and services, response to VVIP, and maximizing customer satisfaction by providing an early detection system of abnormalities and real-time response system, and analyzing in-depth customer complaints and evaluation.
< Integrated VOC Analysis Web Services >
It is becoming more difficult to effectively and reasonably respond to and integrally understand VOC (complaints and evaluation) distributed fragmentarily through multiple channels.
It enables new business planning, real-time risk management, reputation management, and maximizing of customer satisfaction by analyzing in-depth data automatically collected in real time from various channels according to types of products, services, customer groups, and complaints.
It provides automatic generation of visualization/dashboard, automatic alarming, detection of issues, analysis of trends, emotion analysis, automatic reputation, search, automatic classification, real-time collection and convergence of all data to analyze VOC such as consulting memos at a call center, social media, portals’ bulletin boards, internal DB, and reports.
DISCOVERY, Interworking TORNADO and RAINBOW
KT, Korea Expressway Corporation, Hanwha Group (10 Group Companies), National Health Insurance Service, Industrial Bank of Korea, KEB Hana Bank, Shinhan Bank, Busan Bank, etc.
VOC Definition of VOC System
VOC (Voice Of Customer) refers to various queries, complaints, suggestions, etc. that reflect customer reactions to companies’ services for their management activities. The data comprising the VOC has traditionally been consultants’ consulting memos with customers to obtain through CRM’s customer support system used mainly at call centers and bulletin boards for customer support, though now it has grown to include comprehensive data such as customers’ reaction to services and products to obtain through various channels including blogs, twitter, community sites, and other forms.
Saltlux’s VOC Analysis Solution consists of the Information Collection System to collect internal/external informal VOC information, the VOC Analysis System to analyze the collected information and the web service to provide the analyzed information on the web. The solution builds successful consulting record analysis systems by using data collection/analysis solution such as text mining and related information searches of which functions are optimized. Performance is proven through various building cases to analyze informal data (VOC and consulting records).
< VOC Process by Phase >
VOC Necessity for Analysis of VOC
Large-volume customer consulting memos have accumulated on systems by building large-scale call center infrastructure including CRM systems in the 2000s. In addition, consumers express opinions actively due to the proliferation of the digital environment and the facilitation of the Internet. Especially due to the emergence of the Web 2.0 paradigm, prosumers who express actively opinions as consumers through blogs and communities on the Internet have appeared. They actively share their opinions with other users, and by doing so they exert influence on corporate management. Clearly understanding various customer needs and problems and henceforth reflecting them in products/services has a direct relation to the survival of businesses in an increasingly competitive environment.
Main Functions of VOC Analysis System
1. Related Information Analysis/Trends Analysis
By applying topic-ranking technology, the system connects all internal and external related knowledge and suggests the correlation. It deeply understands customer VOC trends by conducting regular statistical analysis of interest keywords in VOC.
< Informal VOC Data-Related Information Analysis >
- By applying topic clustering technology, it connects all internal and external related knowledge and discloses the correlation to provide users with insights.
- It verifies related issues occurred around issues by showing correlations and expressing major related keywords by radar charts.
< Informal VOC Data Trends Analysis >
- It provides services by conducting regular statistical analysis of interest terms within the relevant domain through trend analyses.
- Extraction of popular terms to extract frequently mentioned words in analysis cycle and extraction of sharply increasing terms reflected by the rising tendency of the frequency mentioned in the past based on the analysis time.
2. Visualization of Statistical Information
It helps correctly understand causes of issues and problems and establish countermeasures by quantifying the possibility of issue occurrence, providing visualization and dashboards, and analyzing property information by classification of collected consulting information of customers in various statistical methods (frequency, development and multi-dimensional analysis).
3. General Dashboard
It can view customer demands at once by providing recent trends of VOC for various types such as period, age, sex, etc.
< External Dashboard > < Internal Dashboard > < VOC Keywords Ranking >
4. Comprehensive Ranking of Keywords
It can verify continuous VOC issues through comprehensive rankings, understand new emerging issues through sharply increasing rankings, and provide interest keyword rankings that can see ranking of the information users want.
Inquiry for issue keywords by type (customer/demand/service)
Development of related keywords about question keywords.
Inquiry for the best issue keyword results during the relevant period
Inquiry for the search results of question keywords
< Keyword Ranking No.1 > < Keyword Ranking No.2 >
5. Report Function
It provides various types of reports and statistics with high visibility regarding analyzed VOC.
It inquiries for the current situation including keywords by type such as customer demand type, service type, treatment result type, etc.
Enter multiple keywords when inquiring about types
Enter exclusion keywords while searching the results
Inquiry for setting the period
< Customer Type > < Demand Type >
< Service Type > < Treatment Results Type >
6. Analysis of Related Issues
It inquiries for related issues about internal and external collected information (twitter, blogs, cafes, etc.) and provides trends analysis.
Inquiry for ranking of keywords related to interest keywords
Status of occurrence of related keywords by period
Ranking by related keyword
Inquiry for the status of collected data information
< Keyword Ranking for External Information Analysis > < Related Analysis >
Features of VOC Data
VOC data has generally the following characteristics of contents and needs to apply an analysis technique suitable for each characteristic.
- Short text compared to general documents
A consulting memo is written by short contents with average of 100 characters. In some cases, it is not a sentence, but rather an enumeration of keywords about main issues (complaints, defects, etc.)
- Subordination of product/service domain
VOC data is generated from customer-facing management activities such as sales of products and services. Therefore, VOC data includes feedback contents including customers’ various complaints, suggestions, opinions, and so forth about products and services. It needs to secure and build keywords such as product names, service names, customer classification, etc. to express domain knowledge about products and services.
- May include sensitive information
When describing consulting contents with customers as they are, it includes various customer information such as name, address, resident registration number, telephone number, card number, etc. Such private information is not exposed to the analysis results, and is instead filtered through a preprocessing process.
- Description with abbreviation and a colloquial style
VOC data built by counselors at call centers is prepared during a short consulting time with customers. As a result, it is often described with abbreviation and colloquial styles according to the counselors’ convenience. It is needed to consider that such abbreviation and colloquial styles can be effectively processed in the phase of language analysis.
- Applying partial standardization
Due to efforts to use contents within call centers, partially standardized patterns can be stored. However, it is realistically difficult to conduct pattern-based analysis without exception, as there are many data errors due to various customer feedback and rapid replacement of personnel at call centers. Therefore, it should process complexly parts described with patterns and other data.
Analysis Technique of VOC Data
VOC Data Analysis Considerations
Verification of Data Quality Level
It is necessary to verify VOC data to be analyzed in advance and share it with customers. Though customers recognize that VOC data quality is not good, they may easily ignore contents of quality they wanted after performing projects.
Verification of Major Domain Issues
It is needed to prepare the scenario of detailed customer demand of which information is ultimately extracted from VOC to analyze. Without this, risk increases due to the increase of time to discuss a project.
- TF (Term Frequency): Measurement of the frequency of occurrence of certain keywords in documents
- Topic Rank: Saltlux’s vocabulary related analysis algorithms, or improvement of Text Rank algorithms