Analyzing the voice of the customer is an important way to measure and improve satisfaction. In this blog we will demonstrate how you can turn a large volume of textual comments into actionable information to better serve your customers.
Let’s talk about a use case regarding a satisfaction survey run by a French sport association that is top 6 in terms of number of licensees. The license enables members to practice in affiliated clubs and to take part in official competitions. The association asked its members to evaluate their satisfaction around the licensing process.
The driving factors behind this licensing process improvement project were to reach out to potential members more effectively, and to encourage members to renew their license. Sports federations are financed by institutional subsidies and sponsorship, proportional to the number of members. Maintaining or even increasing the volume of licenses is vital, and directly influenced by the satisfaction around the licensing process.
Three main problems were identified that gave the impression the licensing process was not satisfactory:
The number of 50,000 unlicensed players was growing
Other sport associations seemed to have more modern online tools for licensees
Affiliated clubs found the processing of license applications tedious
The sport association decided to launch an improvement project to tackle these challenges.
The Improvement Project Begins
This Gantt chart in Minitab Workspace detailed the steps and timeline of the project
The sport association consulted regional and local associations and focused on several elements like:
The licensing process of other sport associations
Legal obligations
Pricing
Segmentation (characteristics of the player such as age, gender, geography, type of sport practice, etc.)
The online platform for licensing
The licensees’ opinions
In order to gather the licensees’ opinions, the association created a survey to get feedback with the following criteria:
The survey should take no longer than 6 minutes to complete
The analysis of answers should provide metrics like ratings, scorings, and suggestions for improvement
The sample of responses should be representative of the different categories of licensees
A questionnaire to collect opinions and ideas
19,921 questionnaires were returned out of 200,000 sent out.
The survey platform provided some rudimentary descriptive statistics, enabling an initial assessment of the results. The breakdown of responses to demographics questions proved that the sample was a fair representation of all categories of licensees. The association found that the majority of members had not encountered difficulties with the license renewal process. A third of potential licensees asked their club for help with the process, another third consulted the online help.
The basic graphs and statistics provided at this stage did not consider the variation of responses among the categories of licensees. Did geography, age, gender, seniority, or practice impact opinion?
In addition, textual comments were ignored from this interpretation of the survey results. The question then was ‘What valuable insights could come out of these comments?’
To take the study further, Minitab Statistical Software was brought in to provide data analysis.
The output of Principal Component Analysis (PCA) was explored to identify discriminatory criteria. The team realized the geographical area of respondents had no impact, so this parameter was excluded from later analysis. The contribution diagram helped the team to visualize clusters, reporting similar complaints or needs that could justify changes to procedures, or customization.
The loading plot from the Multivariate Analysis menu in Minitab Statistical Software uncovered several trends
This analysis revealed that young players and first-timers were interested in practice during school holidays. Adult players wanted a mobile app, tournaments, tickets for competitions and goods from the online store. Regardless of age, players wanted to have an online account on the sport association website and to benefit from additional services.
Text mining to better detect suggestions
Eight hundred out of 19,921 participants answered the open-ended questions, so the team thought it would be useful to use text mining to analyze these comments.
Semantic analysis and clustering of words, discovering phrases and main themes drew out some interesting insights.
Most prevalent words (using WordStat)
Similar words commenting on the license application steps were repeated multiple times, such as the documentation requested, the procedure complexity, the necessary simplification, and the breakdown of the cost.
Most prevalent phrases
Several iterations of phrases referred to requested certificates, cost, online payment, and signature.
Seven meaningful themes were identified
The theme clusters that emerged were the price breakdown, the electronic signature, the simplification, the managers’ license model, and the requested certificates.
Recommendations based on an in-depth analysis of opinions expressed
Thanks to this further analysis, the team was able to suggest improvements. The possibility to sign the application form electronically was recommended, as well as a simplification of the online platform, the elimination of superfluous documents to validate the license of volunteers, and better communications around the free services available.
The process and the offer were optimized before the next round of license subscriptions.
Opinion surveys help to measure customer satisfaction. The invaluable textual feedback is often overlooked. Without a semantic study, considering each line of text is tedious. The challenge is to distinguish opinions expressed by certain customer profiles. These can be key to decisive decision-making to improve customer satisfaction.
Interested in Improving Customer Experience Thanks to the Power of Predictive Analytics and 8D?