Subscribe to DSC Newsletter

I'd like to hear some thoughts on measuring customer retention; the business I work in (wholesale financial services distribution) creates some wrinkles that makes straightforward measurement difficult.

Our customers are third party financial advisors (Merrill Lynch, Smith Barney, Wachovia, Edward Jones, etc.). The advisor recommends our products (life insurance, annuities, mutual funds) to their clients. Our sales people call on the advisor just the same as pharmaceutical sales reps call on doctors. The advisor can use our product, or any one of hundreds of competitors products. They do not have any contractual obligation or relationship to us. Additionally, the frequency with which they use our products with their clients is consistent in some cases, sporadic in others.

My question is this: how do we know we are being effective at retaining customers if we don't know the point at which a customer has stopped using our product?

Pharmaceutical sales is analagous, as is the catalog sales business. Any thoughts?

Tags: customer, metrics, retention

Views: 156

Replies to This Discussion

Pharmaceutical sales do have interesting parallels to the situation you mentioned. In many countries, what the doctor does after a sales rep call is often not known by the pharmaceutical company. However, aggregated prescribing data is generally available and broken down into distinct geographic territories. If something similar is available to you for third party advisors' use of products, you would at least have a top-down proxy for the effectiveness of sales rep /marketing activities in regions. You could also survey advisors on a regular basis and ask them about their product usage. Granted, what they say they do may not be the same as what they actually do, but you would now have a bottom-up proxy for estimating customer retention.
Do you have a big enough pool of customers to segment them according to their patterns of usage and employ different strategies depending on which cluster they fall into. That way you can employ completely different approach between regular customer and those known to be sporadic. You could also analyse the differences between the customers in each cluster and look to move customers from sporadic to regular.
Thanks Iain - that's more or less the path we're starting to pursue now. We'll segment customers by historical purchase frequency into a handful of groups and see where that takes us. We'll probably have better results with the high frequency purchase group than the moderate or low frequency.

The other idea we're kicking around is focusing less on a retention measurement and more on some sort of recency and frequency measures for different customer segments. For example, if we segment customers into 3 groups based on their trailing twelve months (TTM) total sales, we would look at how many in each group have purchased from us within the last 30 days, 90 days, 180 days, whatever.

Any thoughts on using recency and frequency this way?
How many customers do you have?

It sounds like your sales reps have a fairly personal relationship with each one of your customers. I take it since you're asking the question, asking the sales reps about customer attrition isn't working.

I've got two suggestions:
1) How often do sales reps make a call and get told their contact isn't working in the business any more -- literal customer attrition?
2) Since you're selling to people that then sell to people, I'm guessing that the vast majority of your customers have fairly regular buying patterns. This is in contrast to most stores, where a custoemr can visit once a year and still be a customer. How regular are the buying patterns?
Thanks Emund, answers to your questions below; and yes, this is a very relationship focused business where personal selling plays a large role.

1) Literal customer attrition is not that common, minor factor.

2) A variety of factors influence the buying patterns. The financial advisor can have new clients "coming in the door" meaning new money to invest, or they can be making transactions with their existing customer base. Also, soem products such as mutual funds are broadly applicable to many client types and situations, where variable annuities are less broadly applicable and suitable in specific situations. As Iain hit on in the post above, an area we are exploring is segmenting advisors by purchase frequency and going from there.
The best advice I can give you is to read this article. It is a symposium that addresses the exact issues you raised of how to analyze wholesale financial service sales response data in depth.

I browsed though the article since I was very busy, (I would otherwise have liked to read more of it before referring), but what I read looked like it would be extremely helpful.

If his idea is too costly, and you are willing to accept more vague figures, you could do focus groups and develop surveys both with the financial advisors and past customers. Get a large enough group of people together (even after the fact) and pay them to tell you the truth about their reasons for why they left and you'll start putting together some very useful data.

Let me know if I can help out in any other way. Good luck...
Thanks Chana - I'll check that out.
First, you might want to consider who your customers really are. You have an agency problem, because the advisors are really your customers, at least from the perspective of sales.

Second, you might want to consider if you even really want to know whether the end-customers are planning on stopping. If you identify such groups -- and the advisors find out -- they might actually increase churn by directing these customers to competitors. My guess is that people change financial products more often than they change independent financial advisors, although I cannot point to specific statistics.

Third, why would you not know when a customer stops a particular financial product? Wouldn't the account relationship end? Or, are the agents responsible for servicing the product as well as selling it?

As a note, there are big differences between what you are doing and the pharmaceutical industry. In general, you are correct that pharmaceutical companies directly market to doctors and do not know who the end user is (in the United States) due to privacy restrictions falling under a set of laws called HIPAA. However, pharmaceutical companies know something that is not available in other industries. They know all the products (their own and competitors) prescribed by each doctor. This information is available, for a price, by companies such as IMS, Verispan, and Wolters Kluwer. In my experience, people in other industries would love to have such competitive information at such a granular level.

In addition, they can get access to some amount of patient longitudinal data, where the patients are deidentified. In effect, you have a stream of prescriptions over time assigned to a specific individual. But, you don't know who the individual is.

From what you say, you need to develop a strong relationship with the advisors and find some mechanism for developing a relationship with their customers. If you really know nothing about the end-users, then your situation may be more analogous to packaged goods manufacturers selling through large retailers.

May be I am asking a dumb question.
Are there any industry sources or sponsored surveys of end level customers that are designed to track their use of your products vi-a-vis competitors' products. If so, you might be able to get testimate the retention or brand switching behavior of customers, without worrying about the advisors who are "middle men", although your promotional efforts will have to be targeted at both the advisors and the end customers.
Some off-the-cuff suggestions ...
Since we are talking about customer segmentation as a solution, won't it be a worthwhile idea to apply 2 layers of segmentation - 1. Customer Segmentation where the customer would be "Advisors" and 2. Customer segmentation where the customer would be "End Users".

The latter could be delegated to the Advisors whereas the former could be done by the company itself. But remember that the criteria of segmentation for 1 & 2 need not be the same.

Also it pays to have mandatory KYC (Know Your Customer) norm in place on both sides - at the company and at the advisors'.


On Data Science Central

© 2019 is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service