Last month, we asked you if your patients are happy and looked at what you ought to consider when collecting feedback. However, as you might expect, the all-important aspect of feedback collection begins once feedback can be analysed and translated into actionable insights.
Actionable insights mean taking your analysis beyond graphs and charts. Whilst tick boxes may be easy to count and analyse, looking at the numbers alone won’t tell you much beyond the basics. In order to access deeper insight, various businesses instead choose to use free text questions. These are able to give companies the best indication of what their customers are feeling, but come with the disadvantage that they can be difficult to analyse. When it’s easy to spend hours sifting through texts in an attempt to grasp some deeper understanding, how can you streamline the process so it is faster and translates into actual improvements?
Automated language processing?
The traditional method of analysing free data is to manually separate the information under different themes. A great amount of time and effort goes into this, and human judgement on themes can lead to overlapping and/or incorrect theming. As well as these challenges, it could be argued that the analysis is not really qualitative; effectively, words are first counted and then being placed within a theme, which is in fact quantitative analysis.
In contrast, we propose a solution that raises the bar for analysis of free text. You may be familiar with Natural Language Processing. Let’s consider the approach taken by Clarabridge, the leading provider of intelligent Customer Experience Management solutions.
Collecting customer feedback
To help illustrate the power of Natural Language Processing (NLP), let’s take a closer look at how Document Capture Company (DCC) and Clarabridge collaborated on a specific VOC implementation. For one client, a leading British multinational retailer, Clarabridge collected free text customer feedback via A5 customer comment cards, along with other sources of customer feedback data. Once the written comment cards were received, DCC designed the verification and 100% accurate extraction of the data including verbatim free text, through a rigorous, fully managed process. DCC uploaded the data to the Clarabridge solution. Clarabridge’s NLP engine then analysed the customer feedback text and sentiment. The actionable insights drawn out from the collected feedback data were then viewed and explored through a series of reports and dashboard. In essence, this is a single screen interface that displays all analysis as a dynamic visual representation. Within the dashboard the client is able to access multiple methods of displaying the information, such as in graphs, heat maps, bubble charts and word clouds.
Before the dashboard is created, all free text comments are processed using Clarabridge’s text and sentiment analytics software. The software is able to measure sentiment and emotions from within the customer feedback data and rate it on a sophisticated 11-point sentiment scale, one end of the spectrum being extremely positive and the other being extremely negative. This facilitates an in-depth and robust means of analysis:
Without delving into the technical mechanics, Clarabridge’s NLP engine rates the customer feedback comment on an 11 point sentiment scale that can understand context and identify grammatical modifiers, for example:
- “The phone service is good”: gives a positive rating, for example the a rating of +1
- “The phone service is really good”: the “really” is a modifier amplifying the positive sentiment, it is given a higher rating, for example the rating of +3.
Having mentioned dashboards in passing, we can now take a closer look at how data might translate into greater insight. Below is a screenshot of a Clarabridge dashboard. Please note that this is a test dashboard and does not represent real data:
This dashboard visualises individual, dynamic and interactive reports. The reports shown in this Clarabridge dashboard include:
- Top left quadrant is a bar chart measuring customer feedback categories against data volumes, which allow the business user to understand the most heavily talked about topics. Top right is the word relationship cloud. Rather than forming the word cloud from a count of words, the cloud shows the relationship between words for added accuracy. The colour of the word relationships depicts sentiment, while the size of the word relationships reflects volume.
- Bottom right quadrant is a double-axis chart, measuring volume and sentiment for each customer feedback category.
- Bottom left is a heat map showing different categories identified by analysing and categorising the comment cards and other VOC data the client collects. Colour depicts sentiment and size depicts volume.
Processing insight into action
For this multinational retailer, Clarabridge’s sophisticated text and sentiment analysis, dashboard and reporting capabilities are enhanced by DCC’s 100% accurate verification and extraction process of the customer comment cards. Clarabridges’ reporting enabled the dashboard to be exported to all relevant locations and to the relevant people, allowing them to instantly identify the strengths and weaknesses about their particular area of concern, without having to spend considerable time reading pages of reports.
The software is truly unique. The same process can be applied through email enquiries, phone calls and even comments made about your business on social media websites on a real-time analysis basis. As you might expect, it can be applied to any form of feedback or free-form text requiring agile, consistent and insightful analysis.
We would be more than happy to offer you a free educational workshop. Give us a call on 0208 90305432 or drop us a line through our contact page!