How to Use Impact Analysis to Prioritize CX Projects in Telecom

Getting any major project off the ground has its hurdles. In the ever-evolving telecommunications industry, this is truer than ever before. RevGen Partners has worked with several telecom clients, all of whom are committed to improving their customer experience (CX). 

The question is always: How? 

With so many technologies available, a myriad of issues to address, and new innovations to put to work, it’s not easy to understand where to start. One way to navigate these tricky waters is using data science to understand the potential impact of any investment. 

Why Impact Analysis for CX and Telecom?

Impact analysis serves to illustrate urgency and aid prioritization of efforts. Prioritization can be incredibly important, especially when it comes to the following key CX concepts: 

• Customer Personas – segments of customers based on attributes 
• Customer Journeys – the experience of a customer in interacting with the company’s service, interface support, etc. 

Inside Customer Journeys, “pain points” are a central concept in CX analysis. These are when customers are prevented from successfully or efficiently completing some sort of interaction, whether via customer support or an app. 

In many cases, when it comes to optimizing the customer experience, there are disproportionately few customers who account for many resources allocated. For example, perhaps only 20% of customers using phone support are responsible for 80% of phone hours. This is especially true of large and ubiquitous utilities, such as telecommunications services. 

Thus, identifying these asymmetries’ existence and impact is essential to proper CX analysis. 

High-level impact analysis is particularly crucial to giant corporations where single decisions are highly leveraged to the point where decision-making is no less important than execution. Therefore, in a use case like telecom, curated prioritization helps align varying divisional data with executive decision-making. 

[Read More: Harnessing a Journey Map to Empower Your Customer Experience Strategy]

Organizational Engagement

Often, a CX-oriented data science solution must be justified to stakeholders with interests beyond CX. Getting key stakeholders to investigate or define CX problems to begin with is a common challenge, especially given the nature of big data, and the fact that data is often the responsibility of departments such as IT, which are further removed from the end-user experience. 

Mega corporations, especially those in the telecom industry, often follow a siloed structure in multiple aspects. Frequently, this stems from building a company through regional acquisitions, which is very common in the telecom space. This decentralized structure often leads to siloed data, where the root cause is siloed organizational protocol. 

This isn’t always detrimental to data science efforts. A decentralized approach is often required when analytical comparisons are made within divisions rather than across them. While this structure is often intentional and effective, it may conflict with top-down initiatives if the understanding of processes between personnel needs to be aligned. 

Generating Buy-in with Impact Analysis

High-level stakeholders of CX initiatives should consider analyzing potential decisions by using data science to systematically and quantitatively evaluate impact. This method of impact analysis may be applied across different groupings of data, and helps other, more indirect stakeholders understand the justification for investing in such initiatives. 

The impact may be evaluated top-down within the confines of a predefined scope to have a sufficient but not overwhelming level of breadth. 

First, key metrics should be defined based on stakeholder goals. Potential impact on these metrics based on outcomes of decisions are then inferred or predicted, with a level of rigor depending on the nature and stage of the analysis. The level of granularity is also dependent on the context of the analysis. 

It should be stressed to target stakeholders that initial steps serve as a baseline for analysis rather than as a complete methodology. Later iterations of analysis may rely on more details specific to the company’s unique business situation. This is critical in the telecom space, as utility services often require more competitive differentiation. 

[Read More: Top 5 Challenges Faced when Building A Customer Experience Program]

Defining and Selecting Key Metrics for Models

Getting stakeholders’ attention with systemized analysis requires defining clear key metrics. Ideally, these are used in modeling as a dependent variable, a concise financial number indicating profitability. 

Numerical examples may include: 

• Total payments per customer 
• Customer satisfaction survey results 

Categorical examples may include: 

• Active vs. Inactive customer (churn status) 

The characterizations in the above examples are up to interpretation, as there are ways to interpret numerical variables categorically or vice versa. 

Special care should be taken to ensure that metrics are not improperly compared across different corporate divisions. If key metrics (for example, subscriber satisfaction) are calculated differently across other divisions which is common, the business may need to consider analyzing impacts within divisions rather than across divisions. 

Furthermore, how key metrics are calculated and interpreted may change based on new findings from initial analysis. It’s important to allow data processes to be sufficiently flexible in anticipation of potential changes. 

Curating, Evaluating, and Prioritizing Big Data

Initial “quick hit” analysis templates also serve as a more robust feature engineering starting point. This serves to help further prioritize “slices” of data for further evaluation of insights and modeling. 

We can see this frequently within the telecommunications space. Given the sheer volume of customer attributes, a systemized method of curating customer data is often crucial. For example, a company may have hundreds of categories of customer service requests and numerous customer payment interactions. In this case, manual averages and regression metrics inspection would be inefficient, especially considering short timelines. 

One may elect to prioritize customer attributes by automating the evaluation of explained variance (e.g., principal component analysis). A more straightforward option may be simply prioritizing attributes by individual totals (e.g., Total number of payments executed via mailed cheque vs. via an app). 

Prioritization via curated data is also essential for many categories of customers within any given attribute rather than many attributes to analyze. For example, manually comparing 100 different geographic locations of customers may be inefficient. Counting customers to size categories may help illuminate the next steps in such situations. 

Conclusion 

Bringing data science forward to help telecom companies prioritize investments, especially for customer experience initiatives can feel like a large workload. However, when done systematically, the payoff – better, more informed decision-making – is huge, keeping strategic priorities aligned with data-driven actionable insights.  

To learn more about how we put data science to work for our clients, visit our Analytics & Insights page.