CLV Roadmap for CPG
Measuring and Improving the Long-term Impact from
Marketing for Fast Moving Consumer Goods
The Market Research Event
Foresight ROI, formerly at ConAgra Foods,
Board Member Marketing Accountability Standards Board
MASB Copyright MASB 2009
CLV Roadmap for CPG, isn’t that a nice title . . . Isn’t that a nice use of acronyms? There are too
many acronyms and I start and end my title with one. Aren’t acronyms confusing ? -- because not
everyone knows the same acronyms. My wife is in health care, and they have their own set of them,
or the military. In the military CO can mean chief officer or company. Militarywords.com has a list of
150,000 acronyms. I googled CLV and got constant linear velocity. My point is that because I have
already used two acronyms, I should explain myself. CLV is customer lifetime value. It represents is
the present value of the future cash flows attributed to the customer relationship. In this case the
customer, is the end customer or consumer. So it’s the future value coming from consumers to the
retailer and to the manufacturer.
Measuring and Improving the Long-term Impact from Marketing for Fast Moving Consumer Goods.
Wow, isn’t that a long title? I had to practice saying it a few times. The point here is that we are
talking about the long-term impact from marketing. Customer lifetime value is a long-term metric. In
this case, it is the value, coming from the end-customer. The metric is expressed in net present
value, value to the P&L, value of both revenue and profit. We want to improve it, so that’s in the
title. To improve you have to know what is working, so you have to measure. Then I change it from
CPG to FMCG. They are the same in this case. So I don’t want to confuse you.
The point is that it’s the Fast Moving and the consumer goods that is the news here. CLV has
been around for quite a while -- banks and direct mail for example. Those industries are data rich,
which gives them the luxury of good metrics. In CPG, we now have more data than ever, and we
are starting to use it well – like Kroger and dunnhumby for example. Now the CPG industry has the
data. Then we just need the tools. The tools exist because CLV has been practiced and evolved for
quite some time. The tools, though, had to be modified for the Fast Moving part. So we are going to
talk about applying an old process to a new industry. We are talking about tools; we are talking
about the P&L value from the end customer, the consumer to the retailer and to the manufacturer.
Marketing Accountability Standards Board
Non-profit industry organization of CMOs,
CFOs, Researchers, Academics
Purpose is to raise the influence of
marketing in the board room
Conducts projects for industry
advancement and education
Some of you might be wondering what is MASB? The Marketing Accountability
Standards Board is a relatively new non-profit organization that was founded to increase
the status of marketing in the boardroom. The mission is to establish (issue, improve
and promote) marketing measurement and accountability standards across industry and
domain for continuous improvement in financial performance and for the guidance and
education of business decision makers and users of performance and financial
One of the projects MASB has sponsored is CLV for CPG. On this project, we have
some of the top academics, Nielsen, ConAgra and Kimberly Clark involved. This is the
tool that I will talk to you about today.
Customer Lifetime Value marketing can help with key
marketing issues for consumer packaged goods companies
How do we drive long-term growth with
How do we develop customer acquisition and
retention marketing strategies that are impactful?
How do we match the right offers to the most
MASB Copyright MASB 2009 3
In the era of marketing accountability, we are posing some tough questions to ourselves. One
of them is how do we drive long-term customer loyalty with our marketing? We measure the
short-term impact of our marketing well – which helps us manage the short-term results. But
our CFOs are asking for more. No longer are they satisfied with answers like: we are building
awareness, getting our message out, differentiating ourselves from the competition. This is all
important to the brand and the business, but they are not acceptable for accountability
standards. We need to link our marketing directly to the P&L – that is all of our marketing, not
just the short-term impact
For those of you in businesses that have formal relationships with your customers – like banks
and companies that sell subscriptions – you know who your customers are and who your
prospects are. Tell, me do you market to your current customers the same as you market to
your prospects? Of course not, when you have a customer, you already know something about
their needs and your job is to retain and build loyalty by satisfying their needs. You cross-sell,
up-sell or try to speed up frequency.
In the CPG world, almost all of the marketing impact work is measuring short-term lift because
those are the tools that we have. But the measurement doesn’t differentiate between an
acquisition strategy – getting new buyers – vs. a retention strategy. An example of an
acquisition strategy might be advertising in a low penetration market or product sampling.
Advertising, sampling and coupons are good tactics for acquisition strategies. The channels for
retention strategies are exploding allowing us to target who we want to reach – e-mail, direct
mail, and banner ads targeted to customers with specific purchasing behavior. In CPG, the
average brand looses 40% of its customers every year, so the upside of retention is huge. This
is one reason dunnhumby has been so successful at Kroger – by matching the right offers to the
right customers – many of which are current brand buyers.
Short-term ROI marketing mix assessment covers the
smallest and least profitable sales volume
Sources of Sales Volume
S-T ROI measures
60% of sales
CPG Industry Average
MASB Copyright MASB 2009 4
What is a baseline sale and where does it come from? (ASK)
The typical definition is the sales that we would have gotten without this year’s marketing. There
is a lot of interest in what is driving base volume because it is big and the most profitable piece
of the business. Then where does it come from?
Is base volume a direct result of all of your previous marketing? -- Distribution, brand equity
from advertising, and product satisfaction. These previous marketing efforts have gotten the
brand to where it is today. It’s your brand equity, your loyal buyers who come back to the brand
even if marketing was cut temporarily. It is an accumulation of all past marketing, your
distribution, your product satisfaction and WOM. Base sales also represent the long-term health
of your brand, because these are sales that you don’t have to incent your customers to get.
Measuring and managing base sales is one of the two components of CLV we will discuss.
Some of the sources of base sales we have some understanding – like distribution increases or
new skus and judgmentally we know if we increase advertising, we will see a positive base
sales impact. But we don’t know enough to be predictive or be able to manage base sales
effectively – especially when we have to cut spending. We cut spending without knowing the
long-term impact on the business.
In an MMA presentation for one brand, we saw an unexplained increase in base sales. We
investigated every possible cause – additional competitive pullback, consumer trends, better
copy – we traced it back to the lag effects from an increase in advertising last year.
So baseline sales are the long-term impact from previous marketing.
Sources of Growth
1. Acquire new customers
2. Retain more customers
3. Increase purchase size
MASB Copyright MASB 2009 5
One way to look at growth is from the three sources. You can either gain new
customers, retain more of your current customers – plugging the leaky bucket, or
increase the amount each customer purchases. We will examine these metrics
because they are the basis of CLV. When you are able to predict customer behavior in
this way you will know the future value of each of your customers. I use the term
customers in the broadest sense – in B2C, your end customer is the consumer and that
is the behavior you will want to understand how your marketing affects consumer
Looks like these businesses are growing
driven by new customers
% of 60%
Healthy Maxwell PG Tips Toyota in NBC Soap
Choice House Pyramid UK (next News Opras
(annual) Coffee Tea Bags purchase) (weekly) (weekly)
(annual) in UK
MASB Copyright MASB 2009 6
Customer churn is a problem
It’s good that these businesses are attracting new customers and at first the results look
impressive. The truth is that all of these businesses are basically flat. Yes they
attracted 30 to 60% more customers to the brand franchises, but they also lost the
same amount of their current customers. So, each brand had to acquire a substantial
number of new buyers, just to stay flat.
The new customers only replaced the lost
ones instead of driving growth
New Customers = Lost Customers Retained Customers
% of 60%
Healthy Maxwell PG Tips Toyota in NBC Soap
Choice House Pyramid UK (next News Opras
(annual) Coffee Tea Bags purchase) (weekly) (weekly)
(annual) in UK
MASB Copyright MASB 2009 7
The new customers gained only replaced those that were lost – this is customer churn.
There are a few ways to look at this data. One way that these businesses are very
successful at attracting new customers and they just need to attract even more new
customers to grow. That takes much more marketing effort than retaining your new
customers – up to 13X.
Or they can plug the leaky bucket. Of course we all want to do that– it’s a matter of how
not whether we should or not.
The point here is that we need to start measuring marketing impact on customer
behavior instead of on brand sales.
Driving the right B2C customer behavior is key to
Impact of a one-week 30% price reduction
Heinz Ketchup Hunt’s Ketchup
Short-term CLV Lift Short-term CLV Lift
Customer (consumer) behavior impact
Own customer purchase Increased purchase size
acceleration and category usage
Source: Dominique M. Hanssens and Shijn Yoo (2008)
MASB Copyright MASB 2009 8
Here are some results from a customer lifetime value study for some CPG brands. The
results are fascinating – especially when you look at how marketing is influencing the
purchasing dynamics. Here we are looking at the impact from a one-week discount of
30% for two ketchup brands. Both brands got a good short-term lift over the next 13
weeks. While Heinz had a smaller percentage lift, the actual volume lift for Heinz was
much greater than Hunt’s because it is a much bigger brand.
However, the short-term lift for Heinz did not last so it had no impact on the CLV of the
brand. What we found was that the incremental purchases were from current Heinz
customers who moved up a planned purchase to take advantage of the deal. Their
customers are trained to know when Heinz is on sale.
The discount for Hunt’s did generate a long-term improvement in the brand’s CLV
because customers stocked up on ketchup. Pantry-loading in this case translated to a
faster use-up rate which actually grew the category and the brand.
Customer Lifetime Value for
MASB Copyright MASB 2009 9
Getting to CLV with metrics and processes
Long- CPG Aspiration
term Customer CLV
MMA Churn Target
Processes Management Marketing
CPG Currently Auto Manufacturers
Kroger with Dunnhumby
MMA Mass Target
MASB Copyright MASB 2009 10
Two things we have to do differently to get us from our current state of ROI marketing to
• Start measuring our marketing with customer metrics so we can learn how our
marketing works among our target segments. Product metrics, for example, are:
units sold, share, sales lift, ROI. Customer metrics are: who bought, whether an
acquisition or retention purchase, purchase frequency, and amount a customer buys.
These metrics help us to a better understanding of how our customers are
responding to our marketing and allows us the better meet their needs with our
• Start measuring and managing both the short-term and the long-term impact from
marketing. Managing the long-term impact is managing customer loyalty – it’s
measuring the baseline sales.
ROI Mass Marketing is model most of us in CPG are following. We’ve got darn good
product metrics. We know if we have a sales plan gap, we know how much trade
funds or advertising is needed to fill it. For example, if we were short $.5MM in
revenue, one Kroger trade deal can fill that gap. Our decisions tend to be more
short-term focus, because that is what we can predict – the short term effect. What
we don’t know is how long the sales lift will last, whether we are attracting new
customers or servicing the needs of our core target.
The CLV targeting world may look more like this: our brand sales are expected to be
soft next period because our new advertising is not working as well with our core
target. We have some P&L leeway because we are projecting to over deliver on our
EPS to Wall Street. So we can add some more advertising targeted specifically to
our core target. This fills the gap next period and sets us up with higher base
volume in the future.
A different situation may lead to a different solution. If the P&L was tight and short-term
efficiency was important, we might choose to do a trade deal might be more efficient
in the long run and sacrifice the long-term effect.
This is the type of control we want from our marketing impact.
Choices for marketing long-term impact
Distributed Lag Impulse Response Function
Markov Brand Equity
There are many ways to measure the long-term impact of marketing – all of them have
some advantages and some shortcomings. One of the most popular with the modelers
is a distributed lag, which is an attempt to figure out how the lag effects of advertising
perform relative to the short-term effects. The distributed lags are an input into the
model, not an output. The typical example of a distributed lag is an advertising half-life -
- a two week half life would mean that 50% of the advertising effect is still present after
two weeks and 25% is left after four weeks. The method works pretty good at
estimating the medium-term effects since it usually gets to be pretty low after 13-weeks.
Markov is a brand switching approach, which can predict the long-term share of a brand
by predicting consumer brand switching behavior. These models have been accurate at
predicting the long-run impact of advertising and pricing but it doesn’t tell you how long
it will take or what programs to get there.
There are many “brand equity” metrics which can tell us what consumers are thinking
about our brand and decompose brand equity into many useful components. Clearly
these metrics are related to long-term brand health and thus are predictive. The
predictive capabilities can be a little weak on the cause & effect, which is important for
marketing planning. We have to know what levers work and how well they work.
An impulse-response function is an output of a dynamic system as a result of a change
or impulse. The term dynamic is important because the one change may cause a
chain-reaction of changes. When we run a feature ad in a grocery store a number of
effects happen. The store manager may realize that demand will increase and more
inventory is needed. So he installed displays to store that inventory and provide some
help to make the feature ad work harder. The competitor is locked out of the feature ad
and display that week causing lost share and sales gap in their business plan. That
competitor intervenes with an incremental trade deal.
This impulse response function is a typical shape of a trade deal. A big sales spike the
first two weeks, then sales ease as more consumers make a purchase taking them out
of the market until the next time they need the product. Then sales return to normal or
sometimes below normal when purchases are pulled forward like with Heinz ketchup.
Then we will see a small bump as the new buyers we acquired return to our brand the
next purchase occasion.
Vector Auto Regressive approach represents the
dynamics of marketing effects
Developed in the 1980’s as an economic
application to account for:
Co integration of variable drivers
Interdependencies of independent
Dynamic lag effects and variable
Vector Autoregression is a specialized type of regression model which helps to estimate
these impulse response functions from changes in marketing or changes in the
environment. Regression models are the most popular market mix assessment tools
used today. They were adopted from economics in the 1980s who used regression
models successfully since the 1940’s. As marketers were discovering regression,
economists were advancing their methods. They realized that the economy is much
more dynamic than can be represented in a single regression equation. What was
missing was synergistic effects from multiple drivers simultaneously. For example, what
happens when unemployment and interest rates rise? The two together produce a
different outcome than each of them separately. They realized that there are chain
reactions – when consumers slow their spending, we need to produce less and fewer
workers are needed.
In January 1980, Dr. Christopher Simms published "Macroeconomics and Reality“,
which introduced the world to Vector AutoRegression. Now it has replaced single
equation regression analysis in economics and is the main cause and effect tool for the
Federal Research Board to manage the economy.
Marketers are now discovering its benefits also.
What is VAR?
Systems of equations
Explains the evolution of a set of variables
as a function of their past evolution
Branch of regression analysis where
independent variables are all previous
values in a time series (Autoregression)
Vector autoregression is a special branch of regression modeling that uses a system of
equations to capture the dynamics between the driver variables and over time. While a
typical single-equation regression model may predict sales as a function of base sales,
advertising sales, promotion sales and discounting. VAR predicts sales as a function of
all of the driver variables combined working together over time. Variables and
coefficients are allowed to evolve as a result of their own evolution and cause & effect
relationships between other variables. It uses all of the independent variables across
the time-series simultaneously as the variables to predict sales.
The bottom line is that the approach is more reflective of how a marketing system works
just like it is more reflective of how an economy works.
VAR system reports marketing results in terms
of consumer behavior
VAR Inputs Variable Dynamics
Marketing Drivers Variable synergies and causalities
Distribution e.g. Coupons + Merchandising
Advertising e.g. Advertising effects on dist.
Coupons e.g. Lag effect of advertising
Discounts e.g. Seasonality–marketing
Other tactics All other synergies and causalities
Competitive VAR Results
Weather 1 2 3 4 ..n
Other factors Purchase Results
Here is an example of what a VAR model looks like. The marketing and environmental
variables are combined to account for the synergistic effect. In this case, we have
multiple dependent variables. So instead of just predicting sales, we can predict future
acquisition, retention and purchase size – which are the major components to CLV. It
also tells you how your marketing is working. Are you attracting new buyers or just
pulling purchases forward that you would have gotten anyway.
Applications: Long-term growth
Cut advertising in 2005
Marketing mix analysis
predicted volume loss for
Unexpected sales losses in
2006 and 2007
Ok, let’s move off the modeling speak and on to practical applications. Here is a
business situation I was in a few years ago. We were able to witness the long-term
consequence of our decision. In 2005, management did not like the advertising copy so
advertising was cut. Our models accurately predicted the lost sales for one year, but
the business took three years to recover because of the long-term effect from the
advertising that we could not measure. Had, we known, perhaps we would have made
a different decision.
Applications: Customer Retention
Heavy buyers eat 56 steaks
Indispensible ad campaign
Growth turnaround and
Ogilvy advertising award
Here is an example of customer retention marketing. When Kraft Foods bought the A1
Steak Sauce brand, whose sales have been sliding for years. Consumer research
revealed that consumers either like A1 and use it or don’t like it, so we felt it was more
prudent to go after our users who like the product. The heavy users ate an average of
56 steaks per year but did not always use A1. Since we felt we couldn’t get them to eat
more steaks, we developed the “Yeah it’s that important” campaign to communicate the
indispensability of A1 – that you can’t have steak without it. The business turned
Kroger stock down 20%
2001 - 2004
system implemented in
Kroger stock up 40% 2004
– 2009, though WMT flat
The best practice of targeting in CPG is at Kroger with the help of dunnhumby, who
uses e-mail and direct marketing to send the right offers to the right consumers. The
result has been that Kroger has outperformed every other food or mass merchandise
retailer on nearly every metric including stock returns since it was implemented widely in
Manage long-term and short-term growth
More control over acquisition, retention
and purchase size strategies
Develop targeted integrated programs
that create synergistic effects
More tools are becoming available to help us better control the impact we get from our
marketing. The CLV approach and tools we just discussed is one of many approaches
available to marketers today. It provides you with a method to control the long-term
impact from marketing to develop more loyal customers. It allows you to implement
successful acquisition and retention strategies and better understand how to layer
integrated marketing elements to create synergies.
MASB is educating the industry with these types of tools. I would encourage you also to
do so. MASB is at www.theMAB.org.