Six Sigma Marketing
# 49: 9/2004
Drilling Down - Turning Customer
Data into Profits with a Spreadsheet
Customer Valuation, Retention, Loyalty, Defection
Get the Drilling Down Book!
In This Issue:
# Topics Overview
# Best Customer Retention Articles
# Measuring Customer Retention
Hi again folks, Jim Novo here.
There are 3 reasons the September newsletter
is being delivered in
My apologies, and hopefully the following article links on advanced
customer retention techniques and the detailed answer to the evergreen
question of measuring customer retention are useful to you.
There are almost 2 months left in Hurricane Season; while the
"eye is passing over", Let's do some Drillin'!
Best Customer Retention Articles
Survival Data Mining for Customer Insight
August 23, 2004 Intelligent Enterprise
A lot of the questions in database marketing really start with this simple
concept: when is a customer no longer a customer? You have to be able to
define the probable end of the Customer LifeCycle
because this sets up the context for every other decision you are likely to make
- if profitability is the prime concern. Survival analysis is a first step
if the end of LifeCycle is clear; if not, you'll need to use one of the LifeCycle
Show Them How Much You Care
September 4, 2004 Target Marketing
One of the most powerful ROI generators in the retention
marketing toolkit is the simple Thank You contact. Done with some
imagination and the right timing, Thank You's are huge incremental sales
generators with best customers. Don't coupon your best, most active
customers; that's a sure way to lose money through subsidy
costs. Instead, thank them and watch your profits soar.
Questions from Fellow Drillers
Measuring Customer Retention
Q: How do most companies measure customer retention?
Is there a formula?
A: The short answer is not many companies outside of
specific industries measure it - yet. It is most commonly used
in telecommunications, financial services (including insurance),
direct marketing (catalogs / web sites, etc.), subscriptions /
publishing, and the travel industry.
The reason for this concentration: these industries have
traditionally collected detailed data on customer interactions as part
of the business model. Now that many other industries are
collecting data on customer interactions, the lessons learned in these
"lead" industries are proving quite valuable for industries
new to direct customer interaction.
A "standard" way to measure it, if you are looking to
align your metrics with Wall Street and your financial statements for
example, is "12 month active". Any customer you have
had contact with in the past 12 months is still a customer, any
customer with no contact in the past 12 months is a defected customer.
This is a retail / mail order oriented view, and if you sell
products, then "contact" means "purchase".
If you are in the services business, it could be any contact - phone
call, e-mail, sales call, download. Divide the number of 12
month active customers by the total number of customers and you have
your retention rate.
There is no reason you can't use "24 month active" or
"36 month active" or "5 year active". The
point is to define what retention is for your particular business and
stick with it. Get agreement on what makes sense for a measuring
stick and try to improve. Often your own data will tell you what
the best "no activity cutoff" is for your business.
Retention is really a "continuum", and retention rate is
always "relative" to your perspective. If you use a
very "tight" definition like "12 month active",
you will lower your retention rate. As you expand the time
period, your retention rate rises. The problem with most
companies is they expand this cutoff time period to infinity, meaning
every customer is still a customer unless they notify you they are
It makes sense for retailers to use "12 month active"
because seasonality and holidays drive many shopping opportunities
over a year's time; if the customer doesn't purchase at least once
across all these opportunities, they're probably not a customer
anymore. If you sell cars, you probably go with something more
like "5 year active". Computers, maybe "3 year
active". Defining retention rate this way is certainly
better than having no definition of retention at all, but it's not
You might notice that using this type of definition, your retention
rate will naturally fall over time as the base of "all
customers" grows. You can tighten this up a bit by defining
a standard base, as in "of all customers who had at least 1
contact in the past 36 months (36 month active), what percent are 12
month active? This way truly defected customers are not included
in the "total customers" base used to calculate retention
rate. I mean, if they are not customers, they should not be
included in the base of "all customers", right? Or you
can track multiple retention rates - 12 month active, 24 month active,
and 36 month active.
Here is another way to look at retention:
Pick a month when your business is heaviest. Find all the New
customers generated during that month and flag them in your database
(for reporting convenience). Each month, run a report to see how
many of those customers have had a transaction with you since the very
first one; include all months since the first transaction for
analysis. Over time, you will see a downward sloping curve like
Of course, if you have customer history, you can go back 2 or 3
years and run the same type of analysis from the perspective of
"today": of the New customers I generated 3 years ago, what
percent have had a transaction with us in the past year? This
percentage is your retention rate for that group of new customers.
But here's the thing. As I tell many people, you can't put a
"retention rate" in the bank, it's really not very
actionable because you can define it however you like. Ideally,
you want to study your customer base for the most actionable
retention measure, and use that measure to drive increased
profitability. Another, perhaps more tangible and actionable way to look at
retention is to make customers "prove" they are still
customers. This may not work for every business, but it is worth
mentioning because some cultures prefer it.
This is where the idea of retention as a "relative"
measure really comes into play. The question is not whether the
customer is "retained" or "defected", but what is
the likelihood the customer is retained, what is the economic
potential of the customer to the business relative to other
customers? If the potential is high, it is worth spending money
to ensure the customer is retained. If the potential is low,
it's not worth taking action.
This likelihood is like the "odds" in betting, and you
can use these likelihoods to optimize marketing budgets. If I
have $1 to "bet" on a customer using a marketing program or
contact, I want to place that bet where the likelihood of
"winning" (making a profit) is highest. I can rank
customers by their potential to be winners, and place bets on all of
them down to where my likelihood to win is low or negative.
Approached this way, retention is not an arbitrary idea, it can be
pinned down to profit. At the point where the company can no
longer make any profit from the customer, where the likelihood of
"winning" my marketing bet is low, the customer is
considered defected. Period.
Sound complicated? Here's a simple way to get started with
this general idea of measuring retention:
Take a small random sample of your customer base and organize it by
monthly Recency buckets - last transaction < 1 month ago, last
transaction 1 - 2 months ago, last transaction 2 - 3 months ago, last
transaction 3 - 4 months ago, etc. At the tail end you can use
"Last transaction greater than 36 months ago" or a similar
idea. If you don't have contact with your customers very often,
you can use quarters or (gasp!) years.
Then create an irresistible offer or other reason to contact the
customer in line with your business, with the objective of moving the
customer to take some kind of action. Deliver the offer or
contact and look at response rate by Recency bucket. You will
see a downward sloping response curve as the Recency buckets get
"older" which eventually approaches zero.
This is your "likelihood" curve, these are your
"odds" at winning a marketing bet on a particular group of
customers segmented by Recency. Now, I would argue that in many
businesses, if the customer is not generating transactions / contacts,
they are no longer a customer. They can't just "sit
there" with "potential". They have to be in the
game generating activity. The longer it has been since the last
contact with the customer, the less likely it is they are still a
customer, and you will see this in your response curve. The
"potential value" of the customer falls as the time since
last contact rises. At some point the potential value of the
customer will approach zero.
The question for you when looking at this curve is this: where
along this downward sloping curve does it make the most sense to
attack the defection? In many businesses, this can be
demonstrated in a purely economic sense. At some point along
this curve, the cost to reach out and try to slow the defection of a
customer by reselling, upselling, or cross-selling exceeds the value
generated by retaining the customer. At some point along this
curve, the cost to make an offer to the customers exceeds the profits
generated by those customers who respond to it. As the Recency
buckets get "older", you will get to the one bucket where
your efforts to reach out generate economic losses.
That point is "economic defection", not because the
customer won't respond anymore at all or even respond ever, but
because you can't make any money trying to get a response /
retain them, there is no longer positive economic value in the
customer, relative to customers in the more Recent buckets. If
the point this happens is at the 18 month Recency bucket, then a
customer is defected when there has been no contact / purchase in 18
months. Period, end of story. A hard, economic cutoff.
Salespeople are intimately familiar with this approach to managing
resources. In their language, the "lead goes
cold". This does not mean the lead will never buy, but it does
mean it is not worth expending resources at this time to try and close
the prospect. The same applies to customers. If it makes
no economic sense to retain the customer, why waste the money?
Just mark the customer "defected" and spend where profit
likelihood is higher given your resources.
Under this approach, your "retention rate" would then be
the number of 18 month active customers divided by the total number of
customers. But now you have a much more powerful idea driving
this metric, because it is defined by economic defection rather than
some arbitrary cutoff. Do you follow the logic?
There are some products and services where this "potential
value curve" is distorted by external, often predictable cyclical
forces that create "flatlines" in economic behavior.
Car purchase cycles come to mind. If it is known by my dealer
I trade cars every 35 months, then the downward sloping potential
value curve doesn't start to kick in until just past the point I
become very likely to trade in my car, at about 35 months. My
potential value is a "flatline" in years 1 and 2.
At some point near the 35 month mark, my potential value to the
dealer who originally sold me the car starts to rise as the time to
trade comes closer. If my 35 month anniversary passes without
this dealer seeing me, my downward sloping potential value curve kicks
in with a vengeance, dropping very rapidly over the next few months to
zero. Why? Because after this 35th month, it is highly
likely I have already traded with another dealer. This type of
curve also has implications for marketing "bets" and when
economic defection occurs. Trying to "chase" me at
this point has very low likelihood of economic reward for the dealer.
What you really need to do is get agreement on how to measure your
retention rate, and then just stick with it. The way you define
it is not really the issue; the point of defining retention is to
create a measuring stick that drives a plan of action. You can
attack customer defection anywhere along the curve, but there are
always periods where it is most profitable to attack. To
find these periods, you have to test your programs / offers / sales
techniques along the entire curve to see where you can drive the
highest profitability given the resources applied.
As you might expect, different customer segments have different
slopes to their potential value curves, often depending on:
* Media source of the new customer
* Offer made to acquire the new customer
* First product / service / contact type
Once you define defection for all customers, you can then start
segmenting and define defection curves for specific customer segments.
This in effect creates a series of new potential value curves that
tilt the odds for winning your marketing bets even more highly in your
favor by segment.
I hope the above helps rather than providing too much information.
Let me know if you have any questions!
If you are a consultant, agency, or software developer with clients
needing action-oriented customer intelligence or High ROI Customer
Marketing program designs, click
That's it for this month's edition of the Drilling Down Newsletter. If you like the newsletter, please forward it to a friend - why don't you do this now while you are thinking of it? Subscription instructions are at the top and bottom of the newsletter for their convenience when subscribing.
Any comments on the newsletter (it's too long, too short, topic suggestions, etc.) please send them
right along to me, along with any other questions on customer Valuation,
Retention, Loyalty, and Defection right here.
'Til next time, keep Drilling Down!
- Jim Novo
What would you like to
the book with Free customer scoring software at:
Out Specifically What is in the Book
Marketing Models and Metrics (site article