Knowledge Services

It was a composite behavioral segmentation scheme aimed at establishing a proxy for customer value. The team had used a range of indicators covering trips, sales & profitability, but profitability calculated at individual customer as such was not coming out to be a reliable measure. We were debating whether to drop profit altogether as customer level profit measure was rather incorrrect.

In the debate, the leader remarked - "I would think an efficient customer segmentation scheme to be as simple as how airlines tier customers based on airmiles flown. Nothing but plain simple number of miles travelled, with a possible weight to premium (business) vs no-frills (economy)"

Though the final scheme we developed for this business was just not unidimensional sales based, we went with what reliable information was already available (dropped profit), than wait for information on every possible dimension to be available in the most reliable manner. The result: we were not very far-off in terms of achieving objectives that we started off with. A neat 10,20,30 & 40% (adding to 100%) split of customers into Platinum, Gold, Silver & Classic segments [this classification was achieved through a statistical model], with the 10% Platinum customers accounting for nearly 40% of sales and nearly half of all the profits (based on an approximate calculation logic). On the other end, we also had a segment that contributed very little on both dimensions.

This is partly because of a high correlation between sales & profits as such, but the lesson for us collectively was - don't get bogged down with what is not available. Proceed with what is already available (and that was not too little anyways), and improve the segmentation as reliable information on additional dimensions become available.

What this helps achieve is also to address the Organizational Change Management issue around getting used to a customer segmentation based decisioning process (for customer service, targeted campaigns etc..) than wait for everything you need to even get started - which would indefinitely delay adoption of an important CRM strategic lever like differentiated marketing treatments

This whole incident got me thinking on how to design simpler segmentation strategies, that would be easy to explain to the business & implement, and very importantly be impactful as well. My observation is that even very basic CRM-101 segmentation schemes based on classic Recency, Frequency, Monetary, and Cross-Shop dimensions work very well.

I remember the CRM director of one of our other customers (a Retailer) remark that their business had realized bottom line benefits worth multiple hundreds of thousands, and he attributed that to: benchmarking of customer behavior based on intuitive and straightforward RFMC score we had developed, clearly defined scale for each dimension of the RFMC, consistent implementation across different epochs of time (we had automated the logic to assign a RFMC score to each customer once-a-quarter to measure changes over time), adoption of RFMC segments in campaigns, and consistent monitoring & action on segment migrations.

Of course, today, this same business is talking about adopting analytics even deeper into their CRM strategies, but the initial taste of success came with using what was already there - Point of Sale transactions, and a proxy identifier to tag each transaction to a customer (conversely, map all transactions of a customer). That is pretty much what is required to get started.

So, just in case you dont have a customer segmentation scheme in place, what are you waiting for?