Cell/Mobile
number becomes a part of a person’s identity. When concept of number
portability was not there, it caused some degree of loyalty to the telecom
operator even if customer was not fully happy with the services or pricing as
changing telephone operator meant changing the number which can be a strenuous task,
as it comes with an additional responsibility of informing the acquaintances
about the new number. After number portability, this element of ‘forced loyalty’
got diluted significantly. If unhappy with the telecom operator, the user can
change the telephone operator without changing the number.
As per
telecom literature, cost of acquiring a new customer is at least 5 times of
cost of retaining an existing customer. With penetration reaching
saturation in many markets, telecom operators are eyeing the same pie, making acquisition all the more difficult, thereby increasing the focus on retention programs. Retention programs are the actions to retain
the existing customers. However, retention/CRM managers find that they get less
than acceptable returns on their retention programs. This is mostly because
retention programs are not targeted sharply enough. A large proportion of
customers targeted with retention programs are often customers that would not
have churned in the first place.
To bring
sharpness in targeting of customers, for optimal utilization of
marketing budget, predictive analytics is deployed. Analytics at work for churn
management, in form of Regression models/Decision trees are build with an aim
of knowing in advance “Who will churn?”, for optimal targeting, i.e. allocating
resources on the basis of probability of attrition (maybe along with
profitability of the customer or other factors). Basic data requirements of
churn analysis are:
- Demographic data from customer information file like age, sex, zip code etc
- Contractual data from service account file such as pricing plan, activation data, contract identification etc.
- Usage & Payment data from billing system such as number of calls, airtime, fixed line time, total amount spent, no. of times calls made to customer care center, change in price plan etc
However, in
the process of building single customer view, handling many variables sometimes
causes dilution in focus on price elasticity aspect of telecom services. Looking at different price-quantity coordinates and drawing
insights from the change in trends/patterns is an important exercise which often
gets overlooked. Also, time to
expiry of contract is an important variable diligently tracked by the telecom
operators. Out of leaving customers, majority leave after their contract
expires. Quite a few times because of this reasoning Survival analysis, an analysis which tries
to answer both questions “Who will churn?” and “When will he/she churn?”, gets ignored.
Importantly, none of these methods answer the question "Why does a customer leave?". Answer (to a limited extent) to this question calls for a serious inclusion of telecom CRM analytics into telecom CRM strategies, rather than seeing it like any other number crunching exercise.
2 comments:
Nice read
Lokesh, thanks for a nice post. Predictive analytics is good, but it's a toy for big companies with petabytes of data. What should small companies do? :)
There is another way to anticipate customer churn. It is based on setting triggers that will inform company employees if it is a risk of losing a customer. Here is even an article on this topic that I hope will be interesting to you: http://bit.ly/TYgbJB
Post a Comment