Measuring and Managing Customer Lifetime Value:: A CLV Scorecard and Cohort Analysis in A Subscription-Based Enterprise
Measuring and Managing Customer Lifetime Value:: A CLV Scorecard and Cohort Analysis in A Subscription-Based Enterprise
Measuring and Managing Customer Lifetime Value:: A CLV Scorecard and Cohort Analysis in A Subscription-Based Enterprise
Managing Customer
Lifetime Value:
A CLV Scorecard and
Cohort Analysis in a
Subscription-based Enterprise
BY MASSIMILIANO BONACCHI, PH.D., AND PA O L O P E R E G O , P H . D .
VALUE (CLV) AND CUSTOMER EQUITY (CE) BECAUSE OF THE LACK OF A PRACTICAL
CLV SCORECARD AND COHORT ANALYSIS ARE PRACTICAL TOOLS THAT CAN
he last few years, companies have increas- HomeAway, LinkedIn, Pandora, Skype, and Zynga.
Company.net
Mobile
Content Content & Media
Telcos
Owners Service Companies
Providers
the customers who have already been acquired by drivers are cost of customer acquisition (CoA) and life-
the end of the period.6 time value (LTV).
2. Total customer equity (CEtot), which is the sum
of the future profit margins generated from cur- Cost of Customer Acquisition (CoA)
rent (CEcur) and future (CEfut) customers of the CoA is a straightforward metric. In this setting, it can be
firm.7 estimated in a very precise way for every single cus-
tomer. Company.net uses search engine advertising
From a management accounting perspective, the that measures cost per click and cost per acquisition
challenge is not only to measure CLV—and thus CE as metrics—as opposed to banner advertising (measured
a summation of each customer’s CLV—but also to man- by a cost per impression model) or traditional channel
age its drivers. Looking at the relationships among advertising such as magazine and TV ads. Because CoA
these drivers is crucial because the variables of the CLV is influenced by the type of marketing policies, its
formula are interdependent. When a metric is pushed analysis is the first step for managerial actions that will
in one direction, it becomes more difficult to ensure impact the company’s attraction and conversion rate. In
that the other keeps the same pace. A managerial tool other words, it can lead to better targeting, better adver-
to consider the interaction among CLV and its drivers is tising, better landing pages, optimization of the flow
the CLV scorecard.8 Figure 2 shows the CLV scorecard through to checkout, more and better payment options,
Company.net used. As you can see, the two key CLV and so forth.
Measures
Average
Revenue
per User (ARPU)
CONTRIBUTION
Yield
MARGIN
Cost of
service
LIFETIME VALUE
(LTV)
Instant churn
LIFETIME
CUSTOMER Historic churn
LIFETIME VALUE
(CLV)
Attraction
COST OF
ACQUISITION Conversion
(CoA) Cost of contact
(cost per click,
cost per acquisition)
Feedback
COUNTRY
SUBSCRIPTION FLIGHT
TELECOM OPERATOR
Margin is the contribution margin per customer. It is customers in order to gauge the effectiveness of a sub-
equal to the average revenue per user (ARPU) minus scription campaign on a specific product or service. For
the cost of service per customer. For example, a $10 example, a customer acquired through Google AdWords
subscription to a mobile service (such as ringtones) at a could have a different churn rate than a customer
per-customer service cost for content delivery of $4 acquired through Yahoo! My Display Ads or Facebook.
would generate a margin of $6. Furthermore, customers who are acquired from a specif-
Yield is defined as the ratio between subscribers suc- ic mobile operator (such as AT&T, Sprint, or T-Mobile)
cessfully billed and customer candidates to be billed. could have different attitudes toward churn and differ-
For instance, if 1,000 customers signed up but only 550 ent yield.
actually pay, the yield would be 55%. These drivers are leading, forward-looking indicators
and must be monitored on a daily basis to ensure the
Monitoring LTV company is heading in the right direction. Company
Monitoring these metrics is useful for evaluating strate- presentations often show all customer metrics improv-
gic and tactical choices, including different advertising ing as they look toward the future. This is unlikely to
campaigns, the launch of a new product, the launch of play out in reality. For example, if a firm attempts to
an existing product in a new country, cross-sell and raise ARPU (price), it will necessarily increase churn.
upsell campaigns to increase ARPU and improve mar- Similarly, if a firm aims to grow faster by spending more
gin, and so forth. on marketing, CoA will likely rise. Churn may increase
The typical Company.net report on LTV drivers as well because a more aggressive marketing campaign
focuses on a product or service (e.g., music, ringtones, will likely capture customers of a lower quality.
and other value-added services) and includes country, The availability of customer data and the particular
subscription flight (i.e., the scheduling of advertising for nature of the business—in which all transactions are
a period of time), and telecom operator used to deliver made online and the log files of each transaction are
the service (see Table 1). Comparisons at the country constantly registered—is a prerequisite for timely
level are used to evaluate the success of a product that management control of these metrics. This allows
was already launched in other countries. The type of Company.net to judge the effectiveness of its marketing
marketing campaign is linked to the churn of acquired investments.
For the calculation of MpK, it is necessary to complete the scenario with ARPU as the input variable. The example
covered here uses an ARPU of $10, which is considered to be booked at the beginning of the contract period, and a
cost of service of $3.50, which makes a contribution margin of $6.50. The customer acquisition cost is $11 per customer
(this figure is estimated looking at the ratio of marketing cost/gross addition).
At the end of May, managers raise the following questions:
1. All else being equal, what is the expected contribution of the next 1,000 customers we will acquire?
2. What is the marketing investment’s payback period?
3. If we stop investing in this market or service, what is the expected residual value of the customers currently on the
firm’s books (i.e., customer equity of current customer base)?
To answer these questions, a report can be developed that follows the MpK rationale. The first step is to transform
billing events in monetary terms by multiplying billing events by margin and scaling the result by 1,000 customers
(Table 3). For example, the January figure ($2,238) has been obtained by multiplying 8,614 (January’s billing) by $6.50
(Margin). This is then multiplied by (1,000/25,016) in order to scale by 1,000 customers.
2 $3,908 $567 - 8%
of past
data
7 $1,650 - 16.1%
decrease
Forecast
t … …
35 $123 - 3.8%
36 $118 - 3.6%
A typical pattern found in Company.net is that, following an initial period of time, MpK tends to level off month by
month. With such a pattern, the company can extrapolate forward using the same month-by-month decrease across
several months. To forecast the evolution of MpK for the cohort for which data is not available (Cohort 6 and beyond in
this example), the following algorithm is applied: The drop from the previous month is multiplied by the percentage,
an x% of decrease, to account for a decrease in churn rate (in this example, the percentage of decrease is 95% for each
month). This example includes five months of data. That is extrapolated forward using the same month-by-month
decrease (on the basis of previous experience) for the subsequent 31 months. Stated differently, MpK of period 6
(1,988) is obtained as follows: 2,420.61 x (- 19% x 95%). This method allows Company.net to project a cohort’s MpK in
the future, estimating the distribution’s tail (as exhibited on the right side of Figure 3). At Company.net, this estimation
procedure usually does not exceed 36 months when forecasting MpK.
Table 4 provides an answer to the managers’ first two questions. Namely, 1,000 customers acquired today would
produce a margin of $31,121 ($2,906 + $3,908 +…+ $123 + $118) during the next 36 months. This result is obtained with
∑ MpK
the following formula:
36
MpK36 months = t
t=1
$4,500
$4,000
$3,500
$3,000
$2,500
$2,000
$1,500
$1,000
actual future
$500
$0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
GMpK
Note: The graph plots the distribution of the monthly margin of the “next” 1,000 customers being acquired.
The vertical line at the fifth month separates the estimation based on actual data (first five months) from the
projection of the tail applying the previous monthly change and a certain rate of decrease (e.g., 95%).
INPUT
OUTPUT
MpK6months $17,805
MpK12months $24,621
MpK36months $31,121
Cost of Acquisition $11,000
Note: The report highlights the relation between CoA and CLV (estimated using MpK). These kinds
of reports are crucial in order to judge the contribution of new subscribers to value creation. The
value of future customers and the payback period alone, however, are not sufficient to obtain a full
picture. The other piece of information required to make a rational decision in this business setting
is customer equity (CE).
Customer Customer
equity equity of
scaled per current
Gross Margin X # customers billed scaled 1,000 customer
Addition by 1,000 customers customers base
Note: Customer equity scaled per 1,000 customers is the value at the end of May of 1,000 customers
acquired in each of the five cohorts. Customer equity of current customer base is obtained from the sum of
the unscaled customer equity of each cohort.
Because of the normalization of the data, the performance measurement system of Company.net is also able to sim-
ulate the effect of other factors that are not under the control of the company, including competitor actions, regulation
changes, or technological discontinuities—any of which may affect the consumer behavior in both conversions and
retention. This is particularly important when it is necessary to neutralize the effect of these factors from the metric
utilized as the base for managers’ bonuses. For example, consider the case of a problem in the log files transmitted
from the telecommunications carriers to Company.net for a specific cohort.
(NOTE: A copy of the Excel file used in the analyses is available from the authors by request.)
average across, so the potential for error increases. metrics are computed across different time
Nevertheless, it is still a useful exercise to get an ranges:
idea of future margin associated with the acquisi- ● MpK1month = MpK1: expected margin of the next
tion of 1,000 customers. How far into the future 1,000 customers in the first month.
MpK6months = ∑ t=1MpKt : expected margin of the
6
the estimation can be extended depends on the ●
type of business. At Company.net, the following next 1,000 customers after six months.
TA K E A W AY E N D N OT E S
1 V. Kumar and Bharath Rajan, “Profitable Customer Manage-
The goal of this case study was to provide management
ment: Measuring and Maximizing Customer Lifetime Value,”
accountants with a better understanding of how to Management Accounting Quarterly, Spring 2009, pp. 1-18.
ascertain the preferences of new and latent customers 2 Peter S. Fader, Customer Centricity: Focus on the Right Customers
for Strategic Advantage, Wharton Digital Press, New York, N.Y.,
in a typical subscription-based business model. It 2012.
involves directly observing the customer’s purchase 3 Paid search advertising entails advertisers competing for top
behavior and subsequently linking this data to customer listing position by bidding in ongoing auctions and then pay-
ing when users click on the advertisements, making paid
value and firm performance. search a flexible and accountable form of advertising (see Des
Calculating CLV (and, by extension, CE) is not Laffey, “Paid Search: The Innovation that Changed the Web,”
Business Horizons, 2007, pp. 211-218). Pay-per-click ads are one
enough because managers need to examine CLV dri- of the most cost-effective advertisement tools. They are the
vers using a framework that makes cause-and-effect “sponsored ads” that are displayed at the top and right of the
relationships between managerial actions and key cus- search results on Google, Facebook, Yahoo!, Bing, and similar
websites. Payment for these ads occurs whenever someone
tomer metrics visible. There is no such thing as an aver- actually clicks on an ad, not when it gets displayed. The cost
age customer lifetime. Because the typical survival per click can range anywhere from a few cents to several dol-
lars, depending on the type of industry and keywords.
curve drops quickly and then levels off, better-informed 4 Christopher S. Chapman, Controlling Strategy: Management,
decisions can be made if customer groups are seg- Accounting, and Performance Measurement, Oxford University
mented based on a “join date.” Failing to do so will Press, New York, N.Y., 2005; Robert H. Chenhall, “Manage-
ment Control Systems Design within Its Organizational Con-
undervalue a firm’s CE. text: Findings from Contingency-based Research and
For every decision, such as the launch of a service in Directions for the Future,” Accounting, Organizations and Soci-
ety, 2003, pp. 127-168; Jean-François Henri, “The Periodic
a different country, a new advertising campaign, and so Review of Performance Indicators: An Empirical Investigation
forth, managers need to rely on reports showing how of the Dynamism of Performance Measurement Systems,”
that decision will impact future CLV and CE. This will European Accounting Review, 2010, pp. 73-96; Ralph Kober,
Juliana Ng, and Byron J. Paul, “The Interrelationship
allow decision makers to better answer questions about Between Management Control Mechanisms and Strategy,”
the types of customers to retain, grow, or acquire, as Management Accounting Research, 2007, pp. 425-452.
5 Kumar and Rajan, Spring 2009.
well as how much should be invested to do so. The use 6 Julian Villanueva and Dominique M. Hanssens, “Customer
of CLV and CE also helps managers understand which Equity: Measurement, Management, and Research Oppor-
tunities,” Foundations and Trends in Marketing, 2007,
advertising channels are more effective and efficient
pp. 1-95.
and how to increase the value of the customer base. ■ 7 CEcur can be considered a special case of CEtot, where the
acquisition of future customers is a project with a net present
value of zero, as seen in Massimilliano Bonacchi, Kalin Kolev,
Massimiliano Bonacchi, Ph.D., is an associate professor at and Baruch Lev, Customer Franchise—A Hidden, Yet Crucial
the University of Naples “Parthenope” and an adjunct pro- Asset, Canadian Academic Accounting Association, Ontario,
Canada, 2012. See also John E. Hogan, Donald R. Lehmann,
fessor of accounting at NYU Stern School of Business. He
Maria Merino, Rajendra K. Srivastava, Jacquelyn S. Thomas,
can be reached at massimiliano.bonacchi@uniparthenope.it. and Peter C. Verhoef, “Linking Customer Assets to Financial
Performance,” Journal of Service Research, 2002, pp. 26-38; and