Marketing on the Internet by Mr A. Goldstuck (http://www.worldwideworx.com/)
Steps to implement marketing optimisation by a local bank (Just for clarity I used to work for this local bank)
This discussion is all about Marketing optimisation. The theory goes that if you combine the probability of selling a product and NPV of that product for many products on a single customer then you can maximise the value creation to the enterprise. So to get this right you need:
- The NPV of all your products
- The Probabilities of selling the product for a specific customer
This for me is where it all comes together, world of statistics meet world of financial modeling. To understand the title of the blog I need to spend some time in the world of statistics, you see to get the probabilities we need to build models that say customer X has Y probability of taking this product. These models will use any customer attribute that helps predict a sale and that makes logical sense. These attributes change over time so a customer can have one probability this month and another the next month. Furthermore these probabilities can be different for the different products on offer for a given customer.
However the NPV's of the products don't really move around month to month, so since the probabilities are changing the best product to sell to a customer is changing month to month. The marketing optimisation will attempt to maximise value for the organisation by picking the product that has the highest potential value for that customer, which is the right decision for the enterprise.
Now enter division Acme of the enterprise. Acme has a manager who needs to manage the sales of her division, to do this she needs to budget a number of products sold and the estimated value that creates. Acme believes that when it comes to getting leads to sell on the Enterprise will do that in a fair manner. What the manager doesn't know is her product has the third lowest NPV out of the ten sold in the Company.
To maximise value the Company starts its path down the route of marketing optimisation starting a division who's only purpose is to optimise marketing. In the first month all leads are allocated according to probability and NPV, as Acme has a low NPV they get zero to no leads.....there goes "fair manner" In the second month all the probibilities have changed and the top NPV product is stuck with too much sales capicity as all the leads are allocated to other higher value creating products, this makes managing something like a call centre almost impossible.
This is a pretty simple example but a very real one, and one with no easy solution. The spirit of optimisation is correct and in the long run will maximise value but how does each division manage sales or set budgets? Is the best solution to forget about budgets and just trust the optimisation to maximise value? I think this is a very good one given you have very good management of marketing execution, but I fear it would require a whole new business management model and consolidated marketing channels (think one call centre to sell them all)
I look forward to hearing from my vast number of readers on any solutions you might have.
Marc
I have no experience in this - only an intellectual interest...
ReplyDeleteSurely, we are missing a third important component here - the NPV of a customer, or customer segment?
If leads are allocated predominantly on the basis of product NPV, surely the allocation system will fail to match each lead with the product that each customer is MOST likely to buy? On this basis, a model that allocates leads to products on the basis of customer NPV would be more valuable. This would create a scheduling problem in the channels but an 80/20 type prioritisation system may do the trick...
Let me know if I've oversimplified the problem here....
Paul
Hi Paul
ReplyDeleteFirstly thanks for the comment, first one so kind of exciting.
Essentially if you consistently match probabilities to product NPV's and select the one with the maximum value for that customer you will over time maximise each customers NPV. The NPV of the customer is a sum of the product parts plus some other effects like lower attrition or churn.
I think customer segments play a big role in affecting the probabilities, at least for fixed income products like insurance. However when it comes to something like a credit card, whole new ball game.
Your 80/20 solution is spot on, I don't think there is an executive out there who will say maximising the companies NPV is a bad thing so simply getting them in a room and letting them schedule according to value is a big step in the right direction.
Thanks for the feedback
Marc