Root Cause Analysis

Getting to the bottom of what customers really want

Usually, when companies want to improve customer experience, service and even their products, their go-to place is the open-ended questions from surveys. They run a key driver analysis to understand which topics contribute to a positive NPS^® score and which don’t. The problem with this approach is that insights are too broad to inspire specific improvement actions.

When you run such an analysis on telecom customer open-ended questions, findings may just reconfirm well-known truths, without explaining the underlying reasons for them. For example, learning that customers are delighted when they encounter a helpful customer agent on the line, but are frustrated by high prices on the website, will not help you very much to improve your performance on the “call centre” and “e-shop” touchpoints.

The solution: root cause analysis

Root Cause Analysis is a 2-step model that allows CX and Insights Managers to uncover the deep, underlying reasons for customer satisfaction or complaints. Its main benefit is that enables the right actions – it allows you not only to target problem areas, but to understand exactly why and under what conditions they turn into issues for customers. For example, a lot of broadband Internet subscribers may share that the plan prices are too high. But when you dig deeper, it turns out that the price is judged subjectively against Internet reliability and speed. If you can’t deliver on your promise for a quick and reliable connection, then the fees you are asking for seem too high for the custom. To reiterate, a deeper analysis of satisfaction drivers will have the following benefits:

Business value

Understand strengths and weaknesses at every touchpoint.
Increase in customer satisfaction through targeted improvements at every touchpoint.
Increased retention due to implemented most efficient course of action.

Methodology – how to get to the bottom of things

STEP 1: MAIN DRIVERS OF NPS

First, we examine the link between NPS and the topics mentioned by customers in their open-ended survey answers in order to estimate the contribution of each topic to monthly NPS. We use a multiple regression model at respondent level (i.e. each observation corresponds to an individual survey response). The dependent variable is NPS and the explanatory variables are binary variables indicating the mentions for each topic with sentiment.

The contribution of each topic to NPS for a given month is obtained by multiplying the estimated regression coefficient with the share of mentions for that month (i.e. by fitting the estimated regression equation with the average values for the month). Separate models can be estimated based on any type of metadata attached to the data set (Segments, Product Lines, etc.).

STEP 2: DRIVERS OF DRIVER ANALYSIS

The Drivers of Drivers Analysis helps our customers gain a deeper understanding not only how drivers impact NPS, but also how drivers affect each other, in order to better focus CX improvement efforts. The analysis is based on partial correlations which indicate the correlation between two variables eliminating the influence of any confounding factors. Results from this model are represented in a tree-like graph where NPS is at the root of the tree and the topics are at the nodes. The closer a topic is to the root, the stronger the correlation with NPS. The thicker the edges, the stronger the correlation between the topics and NPS or the topics themselves.

A VERY IMPORTANT CONSIDERATION

We can’t stress this enough – you need a good topic framework. A good topic framework is one that takes into consideration all your business and organisational specifics:

Products and offerings
Departments and responsibilities – e.g. if customer service is split between call centre agents and digital community managers, be sure to reflect in your topic framework to improve responsibility and ownership of any uncovered issues
Any other important categorisations

Examples

Here is an example output of a root cause analysis project.

The empathy of customer service agents is the top driver that contributes to a positive experience, closely followed by the demonstrated knowledge. The availability of wide choice of TV channels exhibits strong importance as a negative factor. This means, that people who find available TV channels are not enough, are more disappointed by this than people who think TV channels are enough are happy by this fact.

The next step would be to understand which customers need more TV channels and what plan are they subscribed to in order to come up with possible solutions.

A deeper analysis reconfirms the importance of empathic customer service agents over knowledge and competence. Also, it uncovers that the dissatisfaction with the variety of TV
channels is caused by a lack of enough sports channels. Knowing this, we now have more actionable insights to work with:

We can do more empathy trainings for customer service reps
We can focus on the sports channel and maybe provide a special sports channels add-on or pay-per-view model