Customer experience is at the centre of almost every business – companies are actively trying to listen and act on customer feedback and some even predict customers’ next actions. But are businesses taking advantage of all customer feedback they have at hand? In this blog, we’ll show you the uncovered ways for businesses to gain the true power of insights by analysing customer data successfully. Text data is everywhere. From survey responses to emails, complaint forms, call centre agent notes, chatbox conversations and unprompted, genuine feedback coming from social media, forums and reviews. The granularity of insights coming from text data is unprecedented but often underutilised.
What’s missing are the tools to turn unstructured customer opinions and survey answers into an insightful, actionable plan. The actionability aspect is important. It goes beyond vague terms and provides direct insight that shows you where to invest (i.e. training your customer service agents to be more empathetic) and what your customers find incredibly valuable (i.e. good value for money product).To utilise text analysis for the benefit of your business, it is important to learn the mechanics behind.
TABLE OF CONTENTS
Utilise all available sources
Don’t just analyse open-ended responses from surveys; to analysing customer data successfully include sources as account manager meeting notes, social media and forums.
Make a very granular topic categorisation framework
To analyse customer data successfully use three-level topic frameworks as a standard. For example, when we are talking about delivery and installation touchpoints we would like to understand more about topics like self-installation, cost of installation and technicians. When talking about technicians we can go even deeper with the correctly structured topic framework and assess their attitude, helpfulness, duration of visit, issue resolution and etc.
Fine-tune the sentiment to account for any business or industry-specific language
The tone of voice of our customers doesn’t always follow the specific logic we set out in our questionnaires and doesn’t abide by the rules of business terminology. If you ask in a survey: What can we improve? And someone answers “Nothing” – this is indeed a very positive statement. However, if we don’t fine-tune the sentiment it would automatically be read as something negative.
Represent results in an intuitive way so you can engage your entire business
Dashboards should be fine-tuned so they can be used by your entire organisation: CX, Marketing and Global Data & Analytics Departments. One of our biggest clients in insurance has successfully engaged 17 different stakeholders from markets and corporate departments respectively.
Let’s examine how this works in practice to help you start analysing customer data successfully. We’ll look at two very impactful and practical applications of text analytics in two completely different industries – telecom and retail.
The telecom industry is one of the most data-rich and data-intensive industries. Text feedback per year can easily reach some 5 million pieces of individual feedback collected via different channels: transactional and relationship NPS surveys, call agent notes, feedback forms and many others. In-class text analytics is the art of analysing all different types of feedback at hand and coming up with granular insights that can serve every touchpoint and department within your business. This means that no matter how your customer experience program is structured you can analyse customer data successfully across every touchpoint.
To create a real, tangible impact for this data-rich industry we created over 200 topics with an accuracy of categorisation above 85%. Having text analysis set and ongoing led to a 10% improvement across different touchpoints.
The next success story comes from the retail sector and is about a large multinational retail chain that had a very specific issue at hand. In a particular region of Germany, their shops started to underperform regularly, visible through a drop in their sales.
There was no rational explanation and no indication of the reasons based on pre-defined CX metrics. We turned to the unstructured customer text feedback to look for drivers of dissatisfaction that were specific to these locations. Through the creation of a custom-specific topic framework based on mentions and sentiment analysis, we discovered an impactful negative sentiment topic formed around “changing rooms” in the underperforming shops.
Further customer data analysis allowed us to conclude that the length of the curtains in the changing rooms was the main reason for reduced sales. The impact was immediate and a month after the issue with the curtains was addressed, sale volumes across the region successfully return to the expected levels.
The next logical step for them was linking CX improvements to sales in shops and understanding which were the most important drivers which impacted sales within each branch location. Thus via the help of Text Analysis, they were able to understand the value of NPS across branch locations, as well as drivers which influence specific customer groups- ex. women vs men, various age groups, demographics, etc.