Profitable Marketing

Strange maps indeed

Here's a curious site that was featured in the Freakonomics blog: strange maps. Check it out, it has all sorts of interesting ways of displaying geographical data and some fabulous 19th century maps.

More on Customer Lifetime Value Modeling - Using LTV as a Marketing Strategy Tool

My two previous posts about LTV were more descriptive than prescriptive, let us take a look into how LTV can be used as a tool to drive Marketing Strategy at a high level.

Some notes:

• LTV is shared by Marketing, Finance and Treasury - Marketing is most likely owner of the metric and the one that will use it the most but Treasury can use it as a measure of risk and Finance is instrumental in calculating and updating the measure. Ron points out that Activity Based Costing should be used in the calculation of LTV, I agree but don't find it essential, a reasonably accurate knowledge of the margin associated with each sale is all that is required for implementing LTV.
• learn to live with losses - it is very likely that every company will always have some unprofitable customers; this means that you have to adjust your projections accordingly and:
• set your risk of losses to a realistic and objectively calculated LTV
• try and reduce the variability associated with the unprofitable customers

Looking at the plot of the previous post, we recognize that there are two areas under the bell curve: the one to the left of the y-axis that represents the probability of a customer being un-profitable and the one to the right that represents the likelihood of being profitable. Marketing Strategies strive to bring as much of the curve as possible into the positive domain (i.e. reduce the likelihood of having unprofitable customers), this can be accomplished by targeting the three customer LTV groups:

• target the unprofitable segment - the idea is to reduce un-profitability risk, and this can be accomplished in a variety of ways:
• change the product mix - eliminate configurations with lower margin as well as heavily discounted items, discontinue items with low inventory turns that are sold unbundled, etc.
• change the processes that support communication between the customers and the enterprise with a view to reducing the pool of unprofitable customers
• change the return policy - for instance, charge a re-stocking fee or reduce the return period
• change the policy for access to customer support - make it difficult for customers to "game the system", not to access your services and products
• target the profitable segment - this will shift the curve to the right by reducing the proportion of customers that fall in the unprofitable segment (i.e. shrink the ratio of unprofitable to profitable customers)
• acquire profitable customers - get more customers that actually drive your business to success, use clone modeling, any targeting process or gimmick but get more of these folks to buy from you
• change the product mix - same as above but try and cater to the tastes of these customers more that keeping an eye on the individual profitability of the products you sell
• increase the frequency and magnitude of purchases; for instance, increase up-sell and cross-sell - this can be as much a sale exercise as one of pricing, product configuration or even product re-design or service
• deploy a loyalty program that actively caters to this customer segment
• change you supply/value chain to allow for better product delivery and experience
• target marginally profitable customers - have them either fall in the un-profitable group or the profitable one

This list is not meant to be exhaustive but to give you an idea of how different tactics can be used to address the need to decrease the risk of customer unprofitability.

A Customer Lifetime Value Measure for Corporate Performance

One observation about LTV that I commonly find is that it cannot (or is difficult) to come up with an LTV metric that is "useful" at the corporate level. This may be true for cases in which LTV is calculated as a deterministic average across all customers, in this post I'll suggest an LTV metric at the corporate level that is related to VaR (Value at Risk) and that is a good representation of the effectiveness of sales and marketing in a company.

The easiest way to understand what this metric is about is to outline how it is calculated:

1. determine the LTV of each client ou have (or a significant sample thereof)
2. plot the histogram of those LTVs

Yes!!! That simple! You'll obtain a curve that will look like one of the following:

For the sake of simplicity, let's isolate the Normal curve and note that some customer LTVs are negative, these correspond to the unprofitable customers. The shaded area in the following diagram corresponds to the probability that a customer will be unprofitable, this is in essence a primary measure of "customer lifetime value at risk".

Given that some industries have a high likelihood that some customers will always be unprofitable, the threshold for customer value at risk can be set to the left of the y-axis at some other probability that is meaningful for that industry.

I trust that this metric of customer lifetime value at risk can become a useful metric for marketing and sales profitability. This metric represents the health of the customer pool of any firm.

Customer Lifetime Value Modeling

Following my previous post on dropping unprofitable customers and some posts on blogs that I follow (Kevin) (David) (Jim) (Ron), it's time I touch on customer lifetime value.

Getting past the definition that customer lifetime models measure the present value of future cash flows that are expected from customers, here are some characteristics of a workable model:

• measure lifetime value at the customer level - unlike most approaches, I believe that a workable model will not be built at an aggregate level, (for instance as an average of present value measured across all clients) but rather a model that is built for each customer
• the future is more of the same - although you'll be projecting cash flows, our best bet is to bese them on known, historical ones instead of making up some ideal ones
• include costs and revenues - the calculation for prospective value should address margin, thus you'll need to account for renue as well as costs... namely, marketing costs.
• no magical numbers - although CLV is a single number, you'll need to be able to understand and group purchase patterns, so that you understand how there may be several paths to profitability; therefore, also score your customers in terms of frequency, magnitude and tenure

Then, divide your customers into three groups:

• profitable - the name says it all
• marginal - either marginally profitable or not, or margin-neutral
• unprofitable - essentially costs to the enterprise

Once you've built your model you will be able to better design your acquisition and retention efforts. In a nutshell, you want to:

• acquire new customers that look the most like your profitable ones - a clone or response model can be based on an augmented data set that contains these customers
• retain and cross sell to your profitable customers - see what insight you can gain from grouping (clustering, segmenting, whatever you want to call it) profitable customers per pattern if possible or relevant, this should give you an idea of how different purchase patterns are and how you can position products to foster up-selling and cross selling
• increase the profitability of marginal or break-even customers - again, you want to foster purchase patterns that mimic those of the profitable customers. This is where great brand planning and advertising comes in.
• drop unprofitable customers - have them pay for support, charge re-stocking fees, insulate them from discounts and special offers, etc.

All in all, targeting will always follow a targeting process that is oriented towards identifying profitable customers prior to designing strategies or tactics, for instance, if these customers are highly responsive to DM in-home one week after airing a TV30 spot, do not send them a catalog (unless this proposition has been tested and proved positive), also, if you're developing a responder model, use a data set that is composed of profitable customers rather than just "customers".

The Loyalty of Non-profitable Customers - Harrah's Entertainment Bars Richard Brodie From its Properties

The recent misadventures of Richard "Quiet Lion" Brodie, one of the fathers of Microsoft Word, with Harrah's Caesars Palace property in Vegas remind us of why it is important to manage the customer pool and cull unprofitable customers.

According to Brodie's description, he went through a spell of good luck at the poker machines, Harrah's analytical systems flagged him as a non-profitable customer and he was subsequently sent a registered letter informing him that he had become persona-non-grata at Harrah's properties.

Although there must be some pain associated with receiving a letter such as that, the reality is that Harrah's did what most businesses should do: drop unprofitable customers. In fact, Harrah's followed a sensible process in that it took over a year to track Brodie's behavior and arrive at a conclusion. Being in the business that it is in, Harrah's might not have another course of action in avoiding non-profitable customers than sending them a letter, informing them that they are not to come back. It's harsh but that may actually be the only option.

The lessons for other industries is simple:

• Monitor the profitability of your business closely - in particular, monitor the profitability of your customers if you can; this is a classic: target your profitable and marginally profitable customers and drop the non-profitable
• non-profitable customers are simply not customers - you can try and migrate them to a profitable segment (probably not a worthwhile exercise) or find polite ways to drop them
• forget:

• loyalty effects - there is no point in having loyal customers if they are not profitable
• the word-of-mouth effect - do you really want non profitable customers to generate more non-profitable business for you?
• market share considerations - companies exist in order to generate profit, if you give your wares away free, you'll have nearly 100% market share... for as long as your business exists.

I'd like to side with Brodie, after all his post comes across as the writing of someone that feels victimized, but I have to agree with Harrah's in that profit is their raison d'etre and that they cannot cater to unprofitable customers.

Customer Surveys for Target - When Customer Intimacy is Just too Intimate

Target Corp. has just pulled the plug on an online survey that asked questions that were getting a bit too personal. Good as their intent may have been, the questions seemed to personal for most people and a closer reading of the few questions indicated in the article doesn't shed much light into the possible use of this line of questioning.

Attitudinal surveys and models tend to be way too opaque for most marketers but Target's questions seem to defy interpretation... what would be the use of a segmentation that has a factor the depth of a customers' neurosis?

Data360 - Another Great, Open Site for Data Aggregation in Swivel's vein

Would this be the beginning of a trend? The folks at Webster Pacific LLC have started a data aggregation site in Swivel's vein: Data360. They're stated aim for this site (still in beta):

Data360 has been created with two purposes:  First, we hope that Data360 will help provide clear context to important cultural, environmental, social and economic issues.  Second, Data360 is a tool for any organization to report on their internal performance with slideshow graphs for key business metrics. 

The site does not have the same community building flavor of Swivel but does a great job of aggregating data and  allowing  users to access it. Navigation is also easy.

All in all it's a great site to help you nurture your inner geek.

LinkedIn Customer Service - Something Good About It

Like it or now, use it more or less, LinkedIn has made itself into another useful business tool. As I started to blog and network, the number of my connections in the network started to increase significantly, my use of the tool also became more aggressive and after a change in the rules of use, my account was locked.

Regardless of whether I made frequent use of the account, it was an irritant not to be able to use the account as I previously had... and the fact that I use a free account did not diminish my sense of frustration. But this is where great customer service came in: LinkedIn took a bit to respond to my inquiries but the reps' enthusiasm and ability to resolve my issue was quite admirable, more so given that I use a free account!

I can't speak for other people's experiences but to me LinkedIn has displayed a remarkable sense of customer service even considering that I am not a profitable customer... here's a lesson to learn.

Oh, and, should you wonder,  I am not paid to endorse LinkedIn.

Delinquent Accounts - Know When to Lose and Lose with Grace

Here's a quick case: would you rather take delinquent customers to court or sell their debt to a collection agency?

Certainly there is no clear cut answer as that would have to do with the amount owed, the margins involved and so on, but this hypothetical problem can shed some light into the decision process around collections.

In general terms, is it preferable for a retailer with a significant number of delinquent accounts and to take them to court to collect, or sell them for pennies on the dollar to a collection agency? In the first case, there is a significant cost associated with prosecuting every delinquency as well as a risk of losing the case or of taking too long to collect in case of a win as well as the possibility that only a fraction of the debt will be recovered. That is, the decision to prosecute entails a high level of risk and if there is a significant number of delinquencies on the sheet of receivables, there may be more risk on attaining quarterly targets than what's desirable.

On the other hand, selling the collection rights to an outside firm for a fraction of the actual debt, eliminates the risk from the process.

In any case, good targeting, sales and customer care should avoid a significant level of risk in collections.

Gateway Customer Service - What Not to Do and The Power of PR

An article in the Sacramento Bee about a stubborn PC owner that decided to bring Gateway to small claims court illustrates several good points about bad customer service, bad approach to business and to PR. A customer that received a defective computer, couldn't get it fixed, says he didn't receive a replacement machine and decided to take Gateway to small claims court instead of arbitration. Gateway now scrambles to bury the customer in legal costs and complications... all typical legal maneuvering and bad management.

This is a great example of how bad customer service tends to snowball into a costly problem: had Gateway reps or managers had had the power or the foresight to replace the computer, had been savvy enough to ship the new machine with receipt/signature confirmation, the problem would have been contained. Instead, the primary economies of the problem are all wrong now: imagine that the machine cost $1,000, and there are two options: send a new machine or follow the course that Gateway has chosen. In the first case, Gateway would have been out $1,800 ($2,000 minus the corresponding margin - say 20%), in the second case, assume 5, 45 minute calls at about $50 a call which comes to $250... eliminate the margin on the machine. Then throw in 5 hours of legal work at $300 an hour... by the time Gateway "wins" the case, it will be deeply in the red on this sale.

Add to this the erosion that the brand suffers from a case like this and you get a sense for how bad this whole idea of pursuing the case really was.

And on another vein... what's with this arbitration craze? Sure it is cheaper to put a case through arbitration than through the courts but I am not sure I see how that benefits a firm!? From a Machiavellian perspective it seems that the costs of a court case would deter most customer complaints, from a customer perspective it seems that arbitration is akin to resorting to a parallel legal system that is skewed towards the companies (even if it isn't, remember, image and perception are also areas that need to be managed), there is a fear of class action suits but, please, are most products really so bad that they warrant this fear?

All in all we go back to some basic parameters: superior products backed up by superior customer service go a long way to protect against risk; also, customers support managers must be equipped with the knowledge and power to know when to yield to a customer complaint.