How to start a customer behaviour analysis?

UPDATE: A new blog related to how to conduct a customer behaviour analysis has been posted in the following link:

We are going to start supporting the Business intelligence team of the marketing department of a Mobile operator trying to analyze the customer behaviour impacting on a revenue drop because of the MOU’s decrease.

As most of the emerging operators in which we are working, there is no Customer centric culture and therefore, no customer value management is done. Being asked by our clients on the objectives and methodology to be applied in the next assignment, we have defined an end-to-end approach to enable a decision-making tool based on the customer economics.

This approach will handle the following analysis:

1. Top-down analysis: based on the detailed assessment of monthly performance, the main factors contributing to top-line deterioration (e.g., market evolution, competitive dynamics, regulatory compliance, operator performance, etc.) will be identified and their relative quantitative impact will be allocated

2. Bottom-up analysis: by looking at the evolution of detailed operational KPIs, we will be able to draw multi-faceted conclusions to better pinpoint those areas for improvement. This will allow to further break down the extent of the operational drivers of revenue shortfall. Different dimensions need to be taken into account and crossed:a. Traffic volumes and patterns evolution, by typology (on-net / off-net, peak / off-peak, billed / not billed, outgoing / incoming, etc.); b.Shift of minutes and/or subs to the competitors by analyzing incoming traffic patterns; c. Revenue loss analysis by customer segments; d. Regional subs & traffic distribution, by sales channels; e. Recharges volumes and patterns evolution

For this work, we will start designing the customer behaviour datamart, based on current datawarehouse and complementing it with additional sources of information, considering the different needs for the data (end-user can be several departments, different “historical data” needs, different product structures…) and that a Datamart is a “living tool”: it should evolve as needs evolve (customer segments, products, analytical requirements, etc.)

In the particular case of time-series data, it is key to consider not only what happened in the past but also on what is needed to be tracked on time, for how long, and with what level of granularity. This is the first step for defining a customer datamart and to obtain customer value analysis. The faster we get to the industrialization of the analysis, the better our decision-making processes will be.

Will keep you posted on how this project evolves. Best regards. CVA

Update May 2010: The second part of this post has been published here. Enjoy it!

About this entry