A Strategic Approach to Improve Customer Satisfaction with Data Science
Before the digital era, it was comparatively easier for businesses to keep their customers happy with occasional discounts or goodies. But today’s digitally empowered customers, with endless information and options at their fingertips, not only look for better deals but also for better experiences; therefore, are becoming hard to please.
In the wake of this shift in consumer behavior, customer satisfaction has become the biggest challenge and a priority for businesses of all sizes and across industries. In fact, the likes of Amazon and Alibaba are spending millions to improve their customers’ satisfaction level even marginally.
However, it doesn’t mean that you have to spend similar amounts to improve your customers’ experience. A strategic approach, backed by Data Science, can do the trick for you.
Before we elaborate on that, let’s see how customer satisfaction is measured.
Customer Satisfaction Measurement
The standard industry practice is to take surveys. These surveys are typically in questionnaire form, with optional comment section for suggestions and feedback. These surveys capture customers’ response to measure their level of satisfaction on various metrics. Based on the budget, demographics, and nature of business, these surveys can be send out to customers via multiple channels such as email, IVRS, snail mail, SMS, etc. There are million ways to write a survey questionnaire, but the most reliable are the ones designed to capture the Net Promoter Score (NPS).
Net Promoter Score
NPS is an 11-point scale (0-10) that captures customers’ level of satisfaction based on their response to each question.
- 0-6: Dissatisfied customers
- 7-8: Neutral customers
- 9-10: Satisfied customers
NPS is measured as “% Promoters – % Detractors” and is calculated in percentage units, for instance NPS for Amazon and Apple linger between 50-60%. You can easily measure NPS by maintaining a spreadsheet for customers’ response to your survey.
But there’s catch!
The NPS gives you a good enough idea where your firm stands on customer satisfaction, but it doesn’t tell you why the unhappy customer is unhappy. Besides, it is likely that the questions you asked are focused entirely on the product/service. So, there are very thin consumer insights you get out of it. Essentially, you have come to know that there is a problem, but you have no clue how to address it or where to begin.
Remember that optional comment section in your survey? The answer to all these questions could be lying there.
Suggested Read: Customer Experience Goals and Technology Roadmap to Achieve Them
Identifying the Problem Areas
Let’s break down the process into steps for simplicity’s sake:
Create a separate sheet/table and migrate all the records where customers gave a score of 0-6 to that new sheet. This is your detractor base. Here, note that a 2% reduction in detractor base, will automatically shoot up your NPS by the same percentage without even giving any extra doles to the promoter base.
Mine the Text
The comment section we just talked about, which has the key to the problem areas, is nothing but text until you extract valuable information from it. There are various text mining tools available for the purpose. Mining the comment text will help you find the service/product features that are making customers unhappy – Is it the registration, complaint resolution, after sales service, spare parts – you will come to know.
Categorize the comments using text-mining products like IBM SPSS modeler or python based open source text mining modules. Feed the comments to your text mining module and it will categorize the comments in a few seconds or minutes. Work in an iterative manner to identify and map keywords and categories that are most closely related to your business process/product features.
If you are a service provider, categorize your business process; if you are a product seller, categorize the features of your product.
Visualize Problem Areas with Association Chart
Next, generate association charts and category weights to decipher the problem area. A typical association chart will look like this:
As to how to read association charts – the size of blue circles signifies number of detractors complaining about a particular business process or product feature. And the connecting grey lines signify same detractor complaining about the two connected processes/features.
For instance, in the chart above, the biggest problem areas are payment issues and inaccurate information. So, now, it is apparent that taking care of just these two issues can help reduce the detractor percentage significantly.
In this age of hyper-adoption and hyper-abandonment, business trends are primarily driven by consumer behavior and their expectations. So, to stay ahead in the race, it is important that businesses continuously strive to come up with new ways to improve the level of customer satisfaction associated with their products and services. For instance, while text mining & identifying major problem areas is a good method to start with, more advanced techniques like Principal Component Analysis can provide insights that have more impact on the detractor reduction.
We will be covering how to perform Principal Component Analysis in our upcoming post, so stay tuned in to learn more.