Customer Churn – by Roopam

When they leave, they leave footprints for others to follow  

Customer Churn

A couple of months ago, I closed down my savings account with a leading bank in India. I had this account for more than a decade. Since I have worked with the banking and financial services industry, I know I was a reasonably high-profit customer for this bank. I rarely used their ATMs or branches to incur any costs for them. Additionally, my account always had a fairly high amount engaged in high revenue instruments for the bank. Hence, I was a desirable customer for them by all means. If they look back in their data, my account had a steady cash flow with a healthy average quarterly balance (AQB) since the beginning. However, the trend reversed in the last couple of years with more debit transactions (money moving out) from my account than credit. I gave them enough hints by transferring large sums of money from my account with them to my other account with their competitor bank. However, I believe they never tracked the transactions in my account or if they did they never acted upon it.

The interesting part is this bank is one of the pioneers in terms of implementation of analytics in India. I know for sure they have a large team of analysts monitoring accounts in an effort to predict customer churn. Before we try to understand what might have gone wrong with this bank let’s try to learn the relevance of customers churns for companies.

Customer Churn in the New Economy

Customer retention is an important aspect for companies since in most developed economies the market size has almost stabilized. For instance, the population of the United States is more or less constant for the last decade. Population growth rate for most countries shows a similar trend of stability. In essence, there are very few new customers added to the economy. Additionally, easy access to information technology has made the customers extremely informed about the availability of choices. This new economy has ignited a new height of competition among the B2C companies.

Some industries are more susceptible to customer churn than others. For instance, insurance, where policies need to be renewed every year, has a higher probability of churn than banking. However, no industry is immune to customer attrition. Often, customer churn is defined as customers’ voluntary termination of usage of service / product with a company. This often results in a gain for their competitors since in most B2C scenarios usage of services or products is a zero sum game i.e. one provider’s loss is another’s gain. Additionally, acquiring a new customer is much more expensive than retaining the existing customer hence customer churn for the entire industry is a delta negative game rather than a zero sum game. This is evident for airline and telecom industries where excessive price competition and resulting customer attrition has made these industries struggle.

Customer Churn is Disrupting Status-Quo

All natural systems have the tendency to be in a state of equilibrium. Customer or human behaviour is no different. It takes a fair bit of energy to disrupt equilibrium or shake the status-quo. There has to be enough incentive or push/pull for customers to make an effort to move from one service provider to another. The source of this energy is either:

1. Positive incentives from the new service provider (positive pull)
2. Negative perception about the current service provider (negative push)

This brings us back to the point to identifying customer churn upfront and acting upon it. The idea behind customer churn analytics is to identify factors linked to the above forces i.e positive pull and negative push. If one monitors customer related data carefully it is not hard to find a manifestation of these forces in customer behaviour. Some of the common sources of data to register these signals are:

• Customer transactions: pattern in usage of service / product over a period of time (usage velocity)
• Call center: customer interaction data (measurement and quantification of callers voice,  pitch, tone, and text)
• Emails: text mining emails from customer to understand their overall level of satisfaction
• Social media: quantification of customer sentiments

Actionable Insights – Operationalize Analytics

Once the predictive model is ready to identify customer churn, the next and crucial step is to integrate this insight into operations of the company. This is an equally, if not more, creative process compared to the development of high-quality predictive models. I seriously believe this is where the bank I described at the beginning of the article failed. Their staff on the field was possibly not aware of the insights generated through predictive models. Moreover, they were not training to use this information and convert that into positive actions for the bank. This entire process requires a clever team to connect the dots with the apt usage of technology.

Many times expensive actions are not required. For example, human behaviour is  funny, at times companies could prevent high-value customer churn by simple actions like birthday wishes. The next step for data science team is to identify low cost-high conversion actions against detrimental actions and integrate these positive actions in business processes and operations.

Sign-off Note

When they leave, they leave footprints for others to follow. This is true for customers churn as well, the question is whether companies will trace the footprint of churned customers to prevent next lot of mass customer attrition.