Inf Syst E-Bus Manage (2012) 10:19–42

DOI 10.1007/s10257-010-0151-3

ORIGINAL ARTICLE

A tale of two pricing systems for services

Kelly Lyons • Paul R. Messinger • Run H. Niu •

Eleni Stroulia

Received: 28 December 2009 / Revised: 19 May 2010 / Accepted: 27 July 2010 /

Published online: 26 November 2010
 Springer-Verlag 2010

Abstract Due to advances in technology and the rapid growth of online service
offerings, various innovative web-based service models and delivery methods have
appeared—including several free services. It is not always clear whether and how
these emerging mechanisms for online service delivery will result in profitable
businesses. In this paper, with an eye towards beginning to understand the issues
involved, we present an analytical model of rational customer choice between
available service plans. In particular, our model predicts how a monopoly service
provider should devise its plans, if it understands such customer behavior. We then
describe how this model would need to be extended in order to reflect increasingly
inexpensive and even free service offerings.

Keywords Business models  Online profit models  Social media  Free services 
Freemium  Service pricing

K. Lyons (&)
Faculty of Information, University of Toronto, 45 Willcocks St., Toronto, ON M5S 1C7, Canada
e-mail: [email protected]

P. R. Messinger
School of Business, University of Alberta, Edmonton, AB T6G 2R6, Canada
e-mail: [email protected]

R. H. Niu
School of Business and Technology, Webster University, 470 East Lockwood Avenue,
St., Louis, MO 63119-3194, USA
e-mail: [email protected]

E. Stroulia
Department of Computing Science, University of Alberta, 221 Athabasca Hall,
Edmonton, AB T6G 2E8, Canada
e-mail: [email protected]

123
20 K. Lyons et al.

1 Introduction

Web-based service offerings are making inroads on more traditional service-

delivery mechanisms. The phenomenon raises questions about (a) how the emerging
mechanisms for online service delivery will result in viable and profitable
businesses, and (b) how the availability of web-based service offerings will co-
exist with traditional service-delivery models. In Lyons et al. (2009), an analysis of
business models is presented resulting in a categorization of offerings into four
classes: (1) computational processing and database service offerings, provided as
traditional utilities; (2) content-based service offerings by providers from the old
media (gathered by news teams and shared through newswires) and new media
(gathered from the Internet or created by online communities); (3) transactional
service offerings for physical products and packaged software information, or media
products; and (4) brokerage or affiliate service offerings that help bring partners
together to make their own transactions or barter. For each class of offering, a
description of how value is exchanged in a variety of specific instances is presented
(see Table 1). The list of business models (column 2) largely follows Rappa (2008).
The fee-structures column describes ways to monetize the value created by the
business model; however, increasingly, value in these business models is realized
through non-monetary means. This demonstrates how the typical roles of provider
and customer are changing in the context of emerging online service offerings and
in light of a move from a goods-dominant to a service-dominant world (Vargo and
Lusch 2004). Additional third-party entities are also key stakeholders in these
offerings and co-creation of value takes place among many actors in the online
service offerings.
The analysis of Lyons et al. (2009) considers the co-creation of value and
involvement of third parties (especially in the case of advertising), but does not
examine in depth the phenomenon of ‘‘free’’ offerings. In many current on-line
businesses, several users receive significant services for free and the resulting
business models rely on the fact that at least some users (or some third parties)
are willing to pay for service offerings. In his latest book, Anderson (2009)
suggests that the new model of doing business means giving away much of an
offering for free by only charging for a fraction of the value created by the
offering. The prevalence of this business model is made possible in on-line
service offerings because, as Anderson argues, ‘‘atoms cost money, but bits are
free’’ (Anderson 2009). Accordingly, although users have been acculturated to
paying substantial prices for physical products, they often expect online service
offerings to be available for very little or free. This expectation is reinforced by
many companies (Google, Facebook, Twitter, Survey Monkey, LinkedIn, etc.)
giving away much of their service offerings—or at least starting their companies
by doing so.
Indeed, there exists an interesting contrast between two competing views of
business models for service offerings: one, wherein businesses, with some
monopoly power, rationally price their services to utility-maximizing customers
and the other, suggested by Anderson (2009), wherein many online service offerings
are provided free in order to engage customers who then may decide to pay for other

123
A tale of two pricing systems for services 21

Table 1 Service classes, business models, and fee structures (Lyons et al. 2009)
Service class Value exchange: business model Fee structures (with examples)

Computational processing and Utility model: User pays $ to Fee-for-Access: Various forms of
database services—offered as provider; provider provides SaaS.
old-style utilities service to user. Fee-for-Service: Salesforce.com,
Concur Technologies, Digital
Insight, Digital River,
Rightnow Technologies,
Rypple, Taleo, Ultimate
Software, WebEx,
WebSideStory, Workstream
Content providers in the old Advertising model: Third-Party Fee-for-Service: Google, Yahoo,
media (gathered by news teams (advertiser) pays $ to content Standard newspapers such as
and shared through wire provider; provider places the New York Times.
services) and new media advertising in media; end-user
(gathered from the Internet or receives services for free and is
created by online communities) exposed to advertising.
Subscription model: User pays $ Fee-for-Access: Standard
to provider; provider provides newspapers and cable TV, SaaS
service to user. applications.
Fixed Fees: World of Warcraft
Infomediary model: Third-party Fee-for-Service: Doubleclick,
service provider pays $ to info Cnet. (the only difference from
provider; info provider traditional media is the nature
consolidates list of service of the content).
providers; user selects service
provider; third-party provides
service to user. (User may also
co-create the service and
provide ratings of service
providers.)
Community model: Provider Free: Wikipedia, Facebook,
makes available service to user; Youtube, Amazon customer
users create content which review
attracts other users; third-party Fee-for-Access: Second Life,
pays $ to provider (advertising); LinkedIn, Cyworld,
user may pay $ to provider ClubPenguin, ActiveWorlds,
(subscription). World of Warcraft
Fee-for-Service: Facebook Ad,
Youtube Ad, Second Life Land,
ActiveWorlds Land, Webkinz
Toys (Ancillary objects), World
of Warcraft merchant (in-
world)
Transactional services for Merchant: User provides $ to Fee-for-Service: Most standard
physical products and packaged provider; provider makes eCommerce: 1800flowers.com,
software information, or media available products or services to Apple’s iTunes Store,
products. user; provider may create Borders.com, sears.com,
service or procure products or runningroom.com.
services from third parties for $.
Manufacturer (direct): User Fee-for-Service: Ikea, SaaS.
pays $ to provider; provider
sells product or service to user

123
22 K. Lyons et al.

Table 1 continued

Service class Value exchange: business model Fee structures (with examples)

Brokerage or affiliate models that Brokerage: User pays $ to Free: FriendFeed.

help bring partners together to broker; broker facilitates Fee-for-Service: eBay (auctions),
make their own transactions or match-up of users and service expedia.com (travel), Comfree
barter. providers (which may involve a (real-estate). Often commission
service exchange). based.
Affiliate: Users click through to Fee-for-Access: Google Affiliate
third-party for service; third Network
party pays $ to provider; user Fee-for-Service (per click):
pays $ to provider. Amazon Affiliate Program

offerings. In this paper, we examine exactly this contrast. We first develop an

analytical model of service pricing under the assumption that customers, served by
monopoly service providers, make rational choices towards maximizing their utility.
We then examine this model in light of Anderson’s (2009) views to show how such
an analytical model would need to be extended to take these emerging business
models into account. In some cases, this requires sketching out, in general terms,
possible underpinnings of models of service industries that may substantiate
Anderson’s conclusions. In the process, we lay out two very different perspectives
on the pricing of services. Probably the truth lies somewhere in between.
To follow this plan, in Sect. 2, we put forward a specific analytical model for
pricing services that supposes customer rationality, customer willingness to pay for
services, and a single monopoly profit-maximizing service provider. This analytical
model gives rise to a series of testable propositions. In Sect. 3, we then critique the
assumptions and conclusions of this analytical model in light of Anderson’s (2009)
views of online service industries as exhibiting proclivities, or biases, towards
providing free or low-priced services for many customers, together with examples
of on-line businesses that exemplify the various points. Finally, in Sect. 4, we
conclude with a summary of the knowledge contributions of this work and discuss
our plans for future research.

2 An analytical model of service pricing

Service providers can utilize many possible fee structures. We refer to a fee-
structure regime as a service plan (or plan, for short). When a user selects a plan, a
contractual relationship is established between the user and the service provider that
permits the user to obtain the provider’s services at an agreed upon fee structure
(often for a pre-specified length of time).
Inspection of Table 1 suggests that most plans charge users directly in one of
three forms: a one-time fixed fee, periodic access fees, or fee-for-service schedules.
There are other plans, which make a service free to the user, but charge third parties,
such as advertisers, for the right to access the user network. We accordingly abstract

123
A tale of two pricing systems for services 23

the following essential features for modeling purposes: that a provider can charge:
(a) variable fees for particular services; (b) fixed fees for particular service bundles;
or (c) no user fees, but, instead, fees to third parties for access to the system and to
the service users.
We will examine such features of service plans from the perspective of customers
(i.e., users) and service providers. We ask two questions. First, how should rational
customers choose from among the available plans and, given a selected plan, how
much of each service should they utilize? Second, given the answer to the first
question, how should service providers design their service offerings and fee-
schedule plans? We examine these questions in the context of an analytical model.

2.1 Problem formulation

We begin by examining a model of a single service provider making available one

or more service plans. Customers then choose a plan (or no plan at all) and how
much of each service to consume under that plan. We assume that the service
provider knows the customer’s problem and optimally sets its plan(s) according to
how the customer will respond. We, thus, analyze this problem in reverse order
starting with the customer’s problem.

2.1.1 Customer’s choice of a service plan

We begin by supposing that customers can choose from among i = 1,…, I available
plans, each of which specifies a fee structure for k = 1,…, K services. Each plan
i specifies a fixed fee Fi that provides up to the quantity, sk,i, service units for each
service k = 1,…, K.1 An additional per unit charge of pk,i will be collected for each
unit used above a threshold sk,i for each service k = 1,…, K. We allow for another
feature of the plan, imposing hi hours of time spent with advertising or engaged in
self-service activities (a form of service co-generation). Customers are assumed to
have a total of T hours of time which can be consumed by spending hi hours
watching advertising and/or engaging in self-service (the assumptions apply equally
well for both interpretations) or which can be applied toward working at an hourly
wage rate, w, to support consumption of other services or goods. Table 2
summarizes our notation.
Note that this is a general structure that includes many forms of pricing that
we commonly see in the marketplace. When the fixed fee Fi, thresholds
sk,i, k = 1,…, K, and hi are all zero, we have standard per-unit pricing (pk,i is the
price for a unit of service k) for Plan i. When the per-unit charges pk,i, k = 1,…, K,
approach infinity (or are unreasonably high), we have pure bundled pricing (the
bundle (s1,i,…, sK,i) is being sold for a fee of Fi). When the thresholds sk,i approach
infinity, we have ‘‘all you can eat’’ pricing (for the fixed fee Fi). And when imposing

1
Note that, in addition to the base (e.g., monthly) fee, a further fixed fee could also be charged for the
cost of the initiation of the services (for example a one-time fee for a device or a set-up or registration
fee), which we may consider subsumed in Fi. One advantage of abstracting to a one-period model is that
the fixed fee can describe both a one-time initiation fee or a periodic fixed access fee.

123
24 K. Lyons et al.

Table 2 Model notation

Variable Description

Fi Fixed fee for plan i

sk;i Number of units of service k included in plan i
pk;i Price per unit of service k under plan i applicable after threshold sk;i
hi Hours used in self-service or watching advertising arising from plan i
qk Quantity of service k consumed
T Total time the customer has to work, self service, or watch advertisements
w Customer’s hourly wage
x Utility (and quantity) of everything else consumed (explained below).

a time cost hi (together with a lower direct monetary cost), we describe business
models that use advertising or a form of self-service.
There are myriad examples of all of these forms. The general formulation of fixed
fees, thresholds, and per-unit charges above the thresholds applies for cell phone
services which typically require a monthly fee for a given number of calls and text
messages, and, with use above this number, unit prices come into play. Electricity
and other utilities constitute a degenerate case with mostly variable fees. Cable TV
and premium membership fees for virtual worlds such as Second Life or World of
Warcraft constitute a different degenerate case with only fixed monthly access fees.
Infomediaries, such as cnn.com or yahoo.com, and search engines, such as Google,
constitute examples of plans with services free to users, but for which fees are
charged to third-party advertisers and such that users may be required to spend time
watching advertising. Assembling IKEA furniture is an example of co-generation of
value from the service of the furniture, which also consumes user’s time.

2.1.2 Customers’ choices of quantities of services consumed

Next, we assume that after having committed to a particular plan, the customer
chooses the quantity used, qk, of each of the k = 1,…, K available services.
Analysis of Customer’s Choice Problem: Following microeconomic theory (for
example, see Varian 1992), we begin by assuming that a customer maximizes his or
her utility function U, which we assume takes the following tractable form:
U ¼ x þ f ðq1 ; . . .; qK Þ; ð1Þ
where qk is the quantity consumed of service k (k = 1,2,…,K), f(q1,…, qK) is the
utility derived from consuming quantities q1,…, qK of the focal services, and x is the
utility arising from all other consumption activities. Concerning x, we are using a
Hicksian composite (often referred to in product-dominant microeconomics as an
‘‘outside good’’—e.g., see Varian 1992). We assume that x is scaled in such a way
that one unit of utility arises from consuming one unit of this composite. Thus,
x describes both the quantity consumed of this composite of everything else and the
utility derived from consuming this quantity. Furthermore, we assume that utility
x is scaled so that one unit of this composite has a price of $1. Generally, the

123
A tale of two pricing systems for services 25

intuition behind Eq. 1 is that overall utility is the sum of the utility derived from
consuming the various services associated with the chosen plan plus some
additional utility, which is garnered from all of the customer’s other consumption
activities unrelated to the services under examination.
We assume that the customer maximizes his/her utility by choosing a plan i and
quantities q1,…,qK of the focal services under this plan, subject to the following
budget constraint:
X
K
wðT  hi Þ ¼ x þ Fi þ pk;i max ðqk  sk;i ; 0Þ; ð2Þ
k¼1

where T is the total amount of time the user has to either work or be occupied with some
form of self-service or advertising, and w is the user’s hourly wage. Intuitively, the
customer uses up hi hours associated with plan i (either with advertising or self-service)
and has T - hi hours to work, which, at wage rate w, yields an income of w(T - hi). The
customer can spend this income to buy the outside utility or to pay for service plan i,
which involves the standard fee Fi plus the cost of the extra units he or she consumes of
each service, above the default quantities included in plan i.
Now we can state the customer’s choice problem as follows (substituting out
x from Eqs. 1 and 2):
max ½ max ½U subject to Eq: 2 ¼ max½ max ½f ðq1 ; . . .; qK Þ þ wT  whi  Fi
i q1 ;...; qK i q1 ;...; qK
X
K
 pk;i maxðqk  sk;i ; 0Þ ð3Þ
k¼1

Lastly, we note that a utility-maximizing customer will only choose to purchase

the optimal plan i* (that maximizes (3)) if the utility from plan i* is at least as large
as the total cost of the plan; that is, if
X
K
f ðq1 ; . . .; qK Þ  Fi þ whi þ pk;i maxðqk  sk;i ; 0Þ ð4Þ
k¼1

Otherwise, the customer will choose no service plan at all. This follows from
examination of Eqs. 2 and 1. Intuitively, the right side of (4) is the total cost to the
customer of plan i*, which from Eq. 2 is the amount, x, of the outside composite
consumption activity that must be given up to pay for plan i*. From Eq. 1, this is also the
amount of outside utility that must be given up for plan i*. If (4) does not hold,
the amount of foregone outside utility exceeds the amount of utility arising from the
service plan i*, itself, and the customer will choose not to purchase plan i* at all (or any
other service plan, for that matter, since i* is better than any other available plan for the
customer). The relation (4) is an example of a ‘‘participation constraint’’ for the
customer, since the customer will not participate in a plan at all unless (4) holds.
Overall, this is a useful formulation of the customer’s choice problem for services
with potential applicability in some service applications. We explore some simple
applications below.
Two Preliminary Results: We begin by noting that the above formulation gives
rise to the following two straightforward propositions.

123
26 K. Lyons et al.

Proposition 1 Suppose that plan i* is chosen and quantities q*1,…, q*k are
consumed by a customer. If the marginal utility of a service k is positive, then
q*k C sk,i*. Proof: Inspection of (3).
Intuitively, this states that so long as customers continue to value additional units
of a service, they will endeavor to utilize at least the default quantity stipulated by a
chosen plan i*.
Proposition 2 If the only differences between two available plans are fixed cost,
Fi, and the imposition of time hi associated with self-service or reading of
advertisements, then a customer will choose the option that minimizes his/her
whi ? Fi. As a result, customers with low wage rates will be relatively more willing
to give up time hi in exchange for a lower fixed cost Fi and customers with high
wage rates will be less willing to do so. Proof: Inspection of (3).
Intuitively, this states that customers who have a high value of their time will be
less willing to give up their time to engage in self-service co-generation or to watch
advertisements, and, instead, will be more willing to pay the fixed fee.
We now turn to more complex applications of this formulation.

2.2 Service provider’s problem

We consider the problem facing a single service provider operating a local

monopoly. We consider two scenarios involving selling to homogenous and
heterogeneous customers. When customers are homogenous, they all share the
same preferences, described by the same utility function. When customers are
heterogeneous, they have different preferences for the available services,
described by different utility functions. We begin with the case of homogenous
customers.
Homogenous customers: Without loss of generality, we consider a group of
homogenous customers as a single customer (i.e., we normalize the market size—
the number of customers—to one). Suppose the service provider incurs fixed
production costs FC to create and offer services to customers (this is distinct from
the fixed fee Fi that the service provider charges to customers for plan i). In
addition, the provider is assumed to incur a variable cost of ck, for each unit of
service k provided to customers. With these assumptions, we can write the profit P
of offering plan i as a function of parameters (Fi, s1,i,…, sK,i, p1,i,…, pK,i), as
follows2:
X
K X
K
PðFi ; s1;i ; . . .; sK;i ; p1;i ; . . .; pK;i Þ ¼ Fi þ pk;i maxðqk  sk;i ; 0Þ  FC  qk c k
k¼1 k¼1
ð5Þ

2
In principle, the provider’s costs might be decreasing functions of hi when we describe self-service, and
the revenues might be increasing functions of hi (in some way) when we describe advertising. We omit
such considerations here, and leave these possibilities to future research. In addition, for the remainder of
this section (for homogenous and heterogeneous customers), we set hi = 0.

123
A tale of two pricing systems for services 27

The service provider’s problem is to design and make available the plan (or
plans) that maximize the service provider’s profits. In the case of a monopoly
service provider facing homogeneous customers, the provider need only provide one
plan, since all customers can maximize their utility by choosing the same plan
(according to (3) above), in which case any other plans the provider might offer
would be moot. For this reason, in the case of homogeneous customers, we suppose
the provider offers a single plan, and we drop the subscript i. We describe the
^ s^1 ; . . .; ^sK ; p^1 ; . . .; p^K Þ that solve
provider profit-maximizing plan as the values of ðF;
the following problem:
" #
XK X K
Max P¼ Max Fþ pk maxðqk  sk ; 0Þ  FC  qk c k
F;s1 ;...;sK ;p1 ;...;pK F;s1 ;...;sK ;p1 ;...;pK
k¼1 k¼1

ð6Þ
Now, note that any profit-maximizing plan for the service provider must meet the
customer’s participation constraint (4). Otherwise the service provider would have
no customers, which is generally not optimal. We further argue that a monopoly
service provider optimizing the problem (6) will raise the fees, such that (4) holds as
an equality, which we write as
X
K
f ðq1 ; . . .; qK Þ ¼ F þ pk maxðqk  sk ; 0Þ ð7Þ
k¼1

(We note again that we are assuming hi = 0 for this section. This generalizes
when hi is not zero, but the derivations are more complicated.) Substituting (7) into
(6) and suitably adjusting the domain of maximization yields:
" #
XK
Max P ¼ Max f ðq1 ; . . .; qK Þ  FC  qk c k ð8Þ
q1 ;...;qK q1 ;...;qK
k¼1

Intuitively, in economic terms, (8) says that the monopoly service provider
will appropriate and maximize the total market surplus (customer utility less
service provider’s cost). To do this, the service provider needs to set a plan
^ s^1 ; . . .; ^
ðF; sK ; p^1 ; . . .; p^K Þ that induces the customer (via (3)) to consume quantities
q1 ; . . .; q^K ) that maximize (8). Assuming an interior solution, the optimal quantities
(^
q1 ; . . .; q^K ) from the service provider’s perspective are such that
(^
of ð^ qk ; . . .; q^K Þ
q1 ; . . .^
¼ ck ; k ¼ 1; . . .; K ð9Þ
oqk
For the monopoly service-provider problem, the following proposition estab-
lishes two possible optimal plans.
Proposition 3 A monopoly service provider has two profit-maximizing plans (that
solve (6)). One plan described by (10) emphasizes fee-for-service equal to the
marginal cost of providing the service. (This plan assumes an interior solution.)
Another plan (bundled pricing) described by (11) emphasizes a fixed fee that
extracts all the customers’ possible surplus utility.

123
28 K. Lyons et al.

8
>
> p^k ¼ ck ; k ¼ 1; . . .; K
<^sk ¼ 0; k ¼ 1; . . .; K
Plan 1 P
K ð10Þ
> F^ ¼ f ð^
> q1 ; . . .; q^K Þ  ck q^k
:
k¼1
8
< p^k ¼ 1; k ¼ 1; . . .; K
Plan 2 ^s ¼ q^k ; k ¼ 1; . . .; K ð11Þ
: ^k
q1 ; . . .; q^K Þ
F ¼ f ð^
Proof. See Appendix 1.
We can interpret these optimal plans as follows. For Plan 1, the service provider
does not offer any lump-sum service level; instead, it charges a unit price equal to
its marginal cost. Intuitively, by charging a unit price equal to its own marginal
costs, the service provider is able to align the customer’s incentives perfectly with
its own, and the customer chooses what is optimal for the service provider. The
fixed fee is then set to extract any remaining customer surplus (of utility less
variable fees paid). For Plan 2, the service provider uses lump-sum service levels
and very high additional charges that motivate the customer to choose the quantity
that the service provider wants (as a boundary solution). Again the fixed fee is set to
extract any remaining customer surplus and maximize the monopoly profits.
Heterogeneous customers: In this scenario, we assume heterogeneous customers
can be divided into two market segments and customers in each segment are
homogeneous. We assume that there are two segments, 1 and 2, of equal size, and
we normalize the size of each market segment—the number of customers in each
group—to one. We further assume that the difference between these two segments
is in the value that customers in each segment place on service k such that Segment
1 values service k more than Segment 2; that is, f1 ðq1 ; . . .; qK Þ  f2 ðq1 ; . . .; qK Þ and
of1 ðq1 ;...;qk ;...;qK Þ
oqk [ of2 ðq1 ;...;q
oqk
k ;...;qK Þ
; where f1 ðq1 ; . . .; qK Þ is Segment 1’s utility function
and f2 ðq1 ; . . .; qK Þ is Segment 2’s utility function. (Note that we did not have any
subscript for the utility function in the previous discussion because we considered
only one type of customer—one homogeneous market segment.)
According to principle-agent theory (Holmstrom 1979; Eisenhardt 1989), the
service provider will design two different payment plans, in order to differentiate
the two customer types and maximize its total profit. Assume that Plan 1 is designed
for Segment 1 and Plan 2 for Segment 2. Then two types of constraints should apply
for each customer segment in order for the plans to differentiate these two segments.
First, individual rationality constraints (sometimes also called participation
constraints) provide that each segment has non-negative net utility from buying
Plans 1 and 2, respectively:
X
K
f1 ðq11 ; . . .; qK1 Þ  F1  pk;1 maxðqk1  sk;1 ; 0Þ  0; and ðSegment1Þ
k¼1

X
K
f2 ðq12 ; . . .; qK2 Þ  F2  pk;2 maxðqk2  sk;2 ; 0Þ  0; ðSegment2Þ
k¼1

123
A tale of two pricing systems for services 29

where qkj is the quantity of service k that Segment j will choose to use when
choosing Plan j (j = 1,2).
This constraint guarantees that participating in the most preferred plan is better
than buying no plan at all. This is the same constraint we applied in our analysis of
homogeneous customers in (4).
Second, incentive compatibility constraints (Varian, 1992) provide that Segment
1 prefers Plan 1 and Segment 2 prefers Plan 2. In particular, we have
X
K
f1 ðq11 ; . . .; qK1 Þ  F1  pk;1 maxðqk1  sk;1 ; 0Þ  f1 ðq112 ; . . .; qK12 Þ  F2
k¼1
X
K
 pk;2 maxðqk12  sk;2 ; 0Þ and
k¼1

X
K
f2 ðq12 ; . . .; qK2 Þ  F2  pk;2 maxðqk2  sk;2 ; 0Þ  f2 ðq121 ; . . .; qK21 Þ  F1
k¼1
X
K
 pk;1 maxðqk21  sk;1 ; 0Þ;
k¼1

where qk12 is the quantity of service k that Segment 1 will choose to use when
choosing Plan 2, and qk21 is the quantity of service k that Segment 2 will choose to
use when choosing Plan 1.
The service provider would then design two payment plans
(F1 ; pk;1 ; sk;1 ; F2 ; pk;2 ; sk;2 ) to differentiate the two segments that maximize its profit
as follows:
X
K X
K
Max F1 þ pk;1 maxðqk1  sk;1 ; 0Þ þ F2 þ pk;2 maxðqk2  sk;2 ; 0Þ
F1 ;pk;1 ;sk;1 ;F2 ;pk;2 ;sk;2
k¼1 k¼1
X
K
 ðqk1 þ qk2 Þck  FC ð12Þ
k¼1

Here qk1 ; k ¼ 1; . . .; K; are the utility-maximizing quantities for Segment 1

(and qk2 for Segment 2), assuming that individual rationality and incentive
compatibility constraints apply for both customer segments.
Now we can state the solution to the above problem as follows:
Proposition 4 A monopoly service provider has two profit-maximizing options
(that solve (12) and differentiate the two customer segments).
• One option consists of (13) and (14) involving fixed fees for both plans (bundled
pricing). The other option consists of (15) and (16) involving a fee-for-service
equal to the marginal cost of providing the service for Plan 1 and a fixed fee
(bundled pricing) for Plan 2.
• The service provider can use either option to extract all of Segment 2’s possible
surplus utility (note that Plan 2 is the same in the two options). However,
in order to differentiate the two segments, the service provider cannot extract
all of Segment 1’s possible surplus utility. The service provider has to leave

123
30 K. Lyons et al.

M ¼ f1 ð^
q12 ; . . .; q^K2 Þ  f2 ð^
q12 ; . . .; q^K2 Þ to Segment 1 in order to motivate
Segment 1 to choose Plan 1.

Option 1 :
8
>
> p^k;1 ¼ 1; k ¼ 1; . . .; K
< ð13Þ
Plan 1 : ^sk;1 ¼ q^k1 ; k ¼ 1; . . .; K
>
>
:^
F1 ¼ f1 ð^q11 ; . . .; q^K1 Þ  M
8
< p^k;2 ¼ 1; k ¼ 1; . . .; K
>
Plan 2 : s^k;2 ¼ q^k2 ; k ¼ 1; . . .; K ð14Þ
>
:^
F2 ¼ f2 ð^q12 ; . . .; q^K2 Þ
Option 2 :
8
>
> p^k;1 ¼ ck ; k ¼ 1; . . .; K
>
>
< s^ ¼ 0; k ¼ 1; . . .; K ð15Þ
k;1
Plan 1 :
>
> P
K
>
>
: F^1 ¼ f1 ð^
q11 ; . . .; q^K1 Þ  ck q^k1  M
k¼1
8
< p^k;2 ¼ 1; k ¼ 1; . . .; K
>
Plan 2 : ^sk;2 ¼ q^k2 ; k ¼ 1; . . .; K ð16Þ
>
:^
F2 ¼ f2 ð^q12 ; . . .; q^K2 Þ

where of1 ð^q11 ;...^

oqk1
qK1 Þ
qk1 ;...;^
¼ck ; of2 ð^q12 ;...^ qK2 Þ
qk2 ;...;^
oqk2 ¼ck ; k ¼1;...;K; and M ¼f1 ð^
q12 ;...; q^K2 Þ
f2 ð^
q12 ;...; q^K2 Þ
Proof. See Appendix 1.
We can interpret this result as follows. Note that since the main concern is to
motivate Segment 1 to choose Plan 1, it does not matter whether a fee-for-service or
a fixed fee is emphasized in Plan 1. As long as Plan 2 is a fixed fee payment option,
the service provider can differentiate the two customer segments using either option.
Also it can be shown that it is not possible to obtain solutions to the problem by
replacing (14) and (16) with fee for service structures, because Plans 1 and 2 of each
option could no longer differentiate the two segments. Therefore, the service
provider should always offer the fixed-fee structure to the segment with lower
valuation. Both fixed fee and fee-for-service structures work for the segment with
higher valuation on the services.
The question remains, however, whether the service provider always benefits
from differentiating between the two segments. We have the following proposition.
Proposition 5 (A) A monopoly service provider benefits from differentiating the
two segments only when the following condition holds:
X
K X
K
f1 ð^
q11 ; . . .; q^K1 Þ  ck q^k1  f1 ð^
q12 ; . . .; q^K2 Þ  ck q^k2 ð17Þ
k¼1 k¼1

123
A tale of two pricing systems for services 31

and (B) when condition (17) holds, customers who value a service more are willing
to use more and pay more for the service when a monopoly service provider
differentiates its customers.
Proof. See Appendix 1.
The left side of condition (17) is the gross market surplus of Segment 1 (customer
utility less service provider’s variable cost) that the provider can extract when it
differentiates the two segments. The right side of condition (17) is the gross market
surplus of Segment 1 that the provider can extract when it does not differentiate the
two segments. The service provider benefits from the differentiation only when it
can extract more surplus, which is to say the left side is no less than the right side of
(17).
Overall, we presented an analytical model of rational customers, possibly
heterogeneous, each charged the value of the services they receive from monopoly
service providers in a static context. Pricing is done through fixed fees for lump-sum
quantities, and variable ‘‘add-on’’ fees, and the service provider has the ability to
craft different plans targeted to different user segments. Customers have the ability
to choose between plans, giving them some power to avoid being completely
dictated to, but firms also have the ability to craft different plans targeted at different
market segments. Note, however that, in our analysis, we have not modeled the
competition among service providers. Modeling the impact of competition on
consumer choice is quite detailed when there is more than one competitor and each
competitor has more than one offering, but conceptually including such competition
is a straightforward extension of our analysis (which we omit for reasons of
tractability). Modeling the impact of competition on product offerings of
competitors is far more subtle, involving treatment of competition as a form of
non-cooperative game and determining the equilibria of such a game. The model
development along these lines is recommended as a subject for future research. In
Sect. 3, we assess our analytical model in light of Anderson’s views (Anderson
2009). Then, in Sect. 4, we discuss possible extensions to our model to include some
special features of modern service applications.

3 Anderson’s view of ‘‘free’’ services

The problem of defining service plans that result in profitability for service
providers has grown increasingly complex in the presence of extreme competition,
low cost to market entry, growing social networks, and increased opportunity for
advertising-based revenues through larger audiences. As an illustration, Twitter is a
very popular web-based service offering that, after almost four years of growth, has
yet to define its model for profitability (Liedtke 2009a).3 As we saw in Sect. 2,
creating analytical models to help determine profitable service plans that will attract
heterogeneous, rational customers facing a monopoly service provider in a static

3
More recently, Twitter announced a revenue-generating relationship that will give Microsoft and
Google the rights to index Twitter data (Liedtke 2009b) with users continuing to participate in the service
for free.

123
32 K. Lyons et al.

context is sufficiently complex. In this section, we discuss the added complications

brought about by a world of ‘‘Free’’ offerings as described by Anderson (2009)—
with extreme competition, increasing reliance on advertising as a business model,
and a dynamic environment made up of user-generated content and growing social
networks. These factors influence our ability to understand and model users’ choices
and service providers’ offerings.
Most of the business models listed in Table 1 rely on the fact that (some)
customers are willing to pay a fee for their service use. At the same time, we are
witnessing an increasing number of online web-based services becoming available,
in some (possibly limited) form, for free. In his book, Anderson (2009) describes
two trends: some free things are cross subsidized in the sense that ‘‘you get one
thing free if you buy another, or if you pay for a service’’ where others are free
because their cost, ‘‘based in silicon’’, is falling fast. The economic importance of
free products and services lies in the fact that, from the customer’s perspective,
there is a huge psychological difference between very cheap and free: a free service
can become viral, in a way that seems impossible when it costs even a cent. Such
viral services enable, even cause, the creation of large ecosystems of multiple
parties, providers, customers and others, some of which do exchange money,
although not necessarily in exchange for a free service. Anderson (2009) lists six
business models involving the exchange of free products and services:
1. ‘‘Freemium’’ model: Content, services, and software are available in multiple
tiers of use, including a basic free tier. In this model, perhaps only 1% of the
customers pay; and, since the actual cost of the product/service is low, the
paying customers bring in enough to cover the costs for the 99% of customers
who use the free version. (The term freemium was coined by venture capitalist
Fred Wilson.)
2. Free-through-advertising model: Content, services and software are offered for
free because the advertisers (third parties) are willing to pay for access to
customer communities with distinct interests, exemplified by their use of the
offerings.
3. Traditional cross-subsidized model: Products and services are offered for free
and sold as loss-leaders because they entice customers to pay for something
else.
4. Low-cost model: Some things are free because the marginal cost of production
and distribution is zero. In these cases, the service or content may become a
marketing vehicle to promote the sale of something else (and the motivation of
distribution is nearly the same as the cross-subsidization model). This may be
the case when free music is distributed to encourage attendance to concerts by
the same artist, for example. If it turns out to be never possible to charge for the
content (in this case, the music), these activities may become hobbies for
content providers (as is the case for some musicians).
5. Labor-exchange model: Services become free because the customers, through
their use of the service, add value to a network of users, by adding content (as
with Facebook) or simply by expanding the viewer base. This may be

123
A tale of two pricing systems for services 33

subsequently monetizable through different means such that the service

provider receives value from offering the service for free.
6. Gift-economy model: Some things become free because the providers gain some
non-monetary value out of the process. For example, Wikipedia contributors
provide their services as gifts to posterity presumably for the intrinsic value of
self-expression and making a contribution. Somewhat similarly, developers of
open-source software may gain satisfaction from the reputation they build
among their peers (in this case, they may be able to monetize their reputation at
a later time as well).
Many on-line service offerings today employ a combination of the above
business models. In Table 3, we associate the business models presented in Table 1
with Anderson’s (2009) taxonomy described above. In order to understand how
businesses can adopt these new ‘‘free’’ business models and remain viable and
profitable, analyses, similar to those presented in Sect. 2, are needed that can also
incorporate the complexities inherent in the environment of ‘‘free’’ in which these
business models exist. This requires an understanding of the circumstances in which
customers are willing to pay, in each of these models. Anderson suggests that
customers are motivated to pay for an offering under the following circumstances:
(a) to save time, (b) to lower risk, (c) for things they love, (d) for status, or (e) once
they are hooked on the offering. In the remainder of this section we present specific
illustrative examples for each of Anderson’s business models and discuss how
analytical models, such as the ones presented in Sect. 2, would need to be extended
in order to be useful in setting pricing and profit models in the complex web-based
environment that Anderson (2009) describes.

3.1 The Freemium model (Anderson 1)

In principle there are at least two useful metrics upon which the success of the
Freemium model may depend: (a) the actual cost of the service; and (b) the relative
sizes and usage patterns of the free and paying users. Related issues critical to
success concern the pattern of conversion from free to paying users, the length of
the users’ access to the service, and the usage patterns of the two communities (free
and paying users). As shown in Table 3, we can find examples of the Freemium
model in three of the four classes of online services identified in Lyons et al. (2009).
Consider Skype for example, a recognized success of the Freemium model.
Thanks to their peer-to-peer infrastructure,4 the cost of their service is quite low.
Even more importantly, the ratio of the free Skype-to-Skype minutes vs. paid
SkypeOut minutes has been hovering between 7 and 8.5 while other companies are
lucky to get 20 or 100 (see http://www.skypejournal.com/2009/07/skype-sets-new-
performance-records.html).
Online game companies aim to structure their costs so they can break even if as
little as 5-10% of the users pay. More specifically, in Club Penguin 25% of users
pay a monthly fee ($5/mo per paying user); Habbo has 10% monthly paying players

4
http://www.skype.com/help/guides/p2pexplained/.

123
34 K. Lyons et al.

Table 3 Anderson’s taxonomy as related to service classes from Table 1

Service class Value exchange: business model Relationship to anderson’s
taxonomy

Computational processing and Utility model: User pays $ to Anderson 1 (Freemium):

database services—offered as old- provider; provider provides service Can provide tiered levels
style utilities to user. of service
Content providers in the old media Advertising model: Third-Party Anderson 2 (Advertising)
(gathered by news teams and (advertiser) pays $ to content
shared through wire services) and provider; provider places
new media (gathered from the advertising in media; end-user
Internet or created by online receives services for free and is
communities) exposed to advertising.
Subscription model: User pays $ to Anderson 1 (Freemium):
provider; provider provides service Can provide tiered levels
to user. of service
Infomediary model: Third-party Anderson 3 (Cross-
service provider pays $ to info subsidized)
provider; info provider consolidates
list of service providers; user
selects service provider; third-party
provides service to user. (User may
also co-create the service and
provide ratings of service
providers.)
Community model: Provider makes Anderson 6 (Gift-
available service to user; users economy)
create content which attracts other Anderson 5 (Labor
users; third-party pays $ to provider exchange)
(advertising); user may pay $ to
provider (subscription).
Transactional services for physical Merchant: User provides $ to Anderson 1 (Freemium):
products and packaged software provider; provider makes available Can provide some
information, or media products. products or services to user; services or products for
provider may create service or free.
procure products or services from Anderson 4 (Low cost)
third parties for $.
Manufacturer (Direct): User pays $ Anderson 1 (Freemium):
to provider; provider sells product Can provide some
or service to user services or products for
free.
Anderson 4 (Low cost)
Brokerage or affiliate models that Brokerage: User pays $ to broker; Anderson 3 (Cross-
help bring partners together to broker facilitates match-up of users subsidized)
make their own transactions or and service providers (which may
barter. involve a service exchange).
Affiliate: Users click through to Anderson 3 (Cross-
third-party for service; third party subsidized)
pays $ to provider; user pays $ to
provider.

123
A tale of two pricing systems for services 35

(at $10.30/mo per paying user); Runescape reports 16.6% monthly users pay (at $5/
mo per paying user) and Puzzle Pirates has 22% monthly paying players (at $7.95/
mo per paying user). It is estimated that 5–10% of free Flickr users convert to paid
Flickr Pro and Ning reports that 3% of its 500,000 social network creators pay for
the premium version. In contrast, 2% of the casual downloadable-game users pay,
and many free trial web startups get about 3–5% paying users and shareware
software programs often see less than 0.5% of users paying. (for background, see
http://www.longtail.com/the_long_tail/2008/11/freemium-math-w.html).
At the other end of the spectrum, Intuit offers basic TurboTax Online free for
federal taxes, but charges for the state version. As a result 70% of users opt to pay
for that version, since most people have to pay both federal and state taxes. Tax
software is also an example where the development of the content/service is rather
expensive, since substantial accounting expertise is embedded in it. In these cases,
without a substantial conversion rate, Freemium becomes unviable as a business
model. This is likely the reason why Babbel and the Wall Street Journal have
recently turned away from the Freemium model (http://www.fastcompany.com/
blog/chris-dannen/techwatch/can-freemium-work?partner=rss). Babbel, a language-
learning site, moved to a subscription model, as it came out of beta with the intent of
developing more and better content, maintaining a better look-and-feel for the web
site without unsightly ads, and encouraging members to stick with the service and
not move to free competitors, once they have committed to becoming members. On
the other hand, the potential exclusion of free access through Google by the Wall
Street Journal (WSJ) online appears to be less well thought out. HitWise estimates
that 25% of WSJ online traffic is funneled through Google News, and that the
number is increasing. When they do not have a specific task in mind when visiting
WSJ but rather visit to learn about a topic, users are reliant on aggregators and
portals like Google News or Yahoo to let them know that news is breaking
(http://www.weblogs.hitwise.com/bill-tancer/2009/11/newscorp_googleless.html).
The analytical model in Sect. 2 considered that rational customers choose a
payment plan between a fixed fee and no charge for extra use (Merchant model from
Tables 1 and 3) and no fixed fee with a charge for each use (Utility model from
Tables 1 and 3). These plans resemble the Freemium model if the fixed fee is set to
zero and there is no charge for extras for some customers. The analytical model in
Sect. 2 would have to be extended to incorporate the fact that a considerable
majority would choose zero fixed fee and zero charge for extras. Indeed, the model
would have to explicitly incorporate the increased value of the network to those who
would be willing to pay for the premium version of the service.

3.2 Free-through-advertising model (Anderson 2)

The advertising model is, likely, the most prevalent business model on the web. It
almost always accompanies the Freemium model with users of the free service
being subjected to advertisements so that the cost of their service usage to the
provider is covered, at least partially by the paying advertisers. Some common
forms of advertising include Yahoo’s pay-per-pageview banners, Google’s pay-per-
click text ads, and Amazon’s pay-per-transaction ‘‘affiliate ads.’’ Also relevant is

123
36 K. Lyons et al.

paid inclusion in search results, paid listing in information services, lead generation,
and pay-per-connection means of monetization used by social networks like
Facebook.
The analytical model in Sect. 2 considers time spent viewing ads but does not
take into account partnerships with advertisers, the large numbers of customers
reached by the ads, the low cost of ads, and the fact that some customers find the
targeted ads useful. The model treats advertisements simply as a ‘‘price’’ that the
customers have to pay (by wasting time which could otherwise have been spent
earning some other utility). On the other hand, the advertising business model treats
them as a ‘‘value-producing service’’ for which the providers can charge the
advertisers. This distinction highlights the most interesting limitation of the
analytical model in Sect. 2, namely that it circumscribes its analysis to a two party
system, including providers and customers, where most of the business models in
Table 3 involve multiple types of partners. For the advertising model, one should
realize the benefits to advertisers of increased market awareness, trial rates, and
even repeat purchase behavior of consumers.

3.3 Traditional cross-subsidized model (Anderson 3)

Web beacons (i.e., transparent 1 9 1 pixel images embedded in HTML documents

to track users’ visits through to a set of web sites) were an early mechanism for
establishing networks of affiliate service providers. Through the monitoring of web
users’ navigation patterns, the group of affiliates could point users to each other’s
web sites. Today there are more and more complex mechanisms to support such
service-provider consortia.
For example, Google’s AdSense product enables any web site to become
affiliates to other providers who wish to advertise their products and services. With
AdSense for Content, the affiliate can place text, image or link ads in its web sites,
provided they follow Google’s policies. With AdSense for Search, the affiliate
places a search box in their web site, which leads to a results page that hosts more
pay-per-click ads. Finally, with Google Referrals the affiliate refers visitors to use a
Google product, like AdSense, AdWords, the Google Toolbar and other Google
software.
The proliferation of content and services on the web has also brought forward the
need for aggregators (who collect content of interest to their users, filtering out
uninteresting information) and infomediaries (who match users and content
providers). NetZero was one early example of a third-party infomediary. They
offered 40 free hours of monthly Internet access to more than 8 million customers,
who were required to allow a special browser, the ZeroPort, to remain on their
screen while online. The ZeroPort displayed ads that, based on the marketing
information the customers provided to NetZero, were likely to interest them.
ZeroPort also allowed small businesses, subscribing to NetZero, to reach local
customers and view the daily results of their online ad campaigns.
The infomediary model is a specialization of the affiliate model, where the
original service provider is the infomediary who forwards the customer to select
affiliates. The analytical model presented in Sect. 2 assumes one provider only and

123
A tale of two pricing systems for services 37

rational customers and would have to be extended to incorporate affiliates and other
third party participants. In order to properly model and analyze affiliate relation-
ships, one would have to examine users’ navigation patterns in order to understand
the potential value of an affiliate relation. Clearly, not all affiliations will be
productive; the ‘‘original’’ service provider and its affiliate cannot be direct
competitors, neither can they be completely unrelated. The nature of the relation has
to be synergistic in a way that makes users who visit the original provider likely to
visit the affiliate. Modeling complexities such as these are nontrivial.

3.4 Low-cost model (Anderson 4)

In some cases, on-line service offerings will charge users for the service but provide
low-cost offerings (such as screen savers, images, and other features) for free to
promote or help market the online service. For example, software developers
frequently offer a limited version of a new product for free, in order to entice the
trial users to buy it. The analytical model in Sect. 2 could capture the choice of free
service offerings by setting the price per unit to zero but cannot currently model the
return value of marketing and promotion.

3.5 Labor-exchange model (Anderson 5)

The proliferation of on-line communities for professional networking, gaming, and

web-based and virtual-world social networking has given rise to many variants of
the community model, and these communities, in turn, have made possible an easy
application of Anderson’s labor-exchange business model. The presence of such
communities also brings tremendous dynamics to the web environment.
Twitter, although not yet profitable, is continuously increasing in membership
(thus, growing its information exchange community) and this has led several third
parties to monetize its community (http://www.mashable.com/2009/03/23/twitter-
business-model-2/). Mashable launched the Twitter Brand Sponsors that syndicates
a limited number of brands into the Mashable sidebar, so that interested visitors can
choose to connect with those brands on Twitter (http://www.mashable.com/2009/
03/05/twitter-brand-sponsors/). The ad network Federated Media launched Exec-
Tweets that aggregates Tweets from business executives, so that Twitter users can
follow threaded version of the executives’ tweets. In the former case, it is the brands
that pay Mashable in order to be featured in its sidebar. ExecTweets is currently
sponsored by Microsoft but the intent is to have people who are interested in the
executives’ opinions pay to subscribe to the service. The community of Twitter has
also been the subject of substantial analysis by Sysomos (http://www.sysomos.
com/insidetwitter/), a leading Media Platform Analytics company, interested in
mining tweets to extract information relevant to businesses that are the Sysomos’
paying customers.5 Twitter itself appears to be considering the idea of charging
users to read tweets of key users (see http://www.techradar.com/news/internet/

5
In July 2010, Sysomos was acquired by Marketwire–http://www.marketwire.com/press-release/
Marketwire-Acquires-Sysomos-1286185.htm.

123
38 K. Lyons et al.

twitter-to-charge-for-reading-tweets-next-year-654608?src=rss&attr=all and http://

www.media.asia/DigitalMedia/newsarticle/2009_11/Twitter-Japan-to-introduce-
payment-model/38057). It is likely that some tweets will be available for free, with
valuable content—images, video or original research—being available only to
paying subscribers. All of these examples demonstrate the potential value generated
by large communities and the exchange that takes place within them, in terms of
aggregation of interesting information, business intelligence or, simply, original
content. As Anderson (2009) notes, ‘‘In each case, the act of using the service
creates something of value, either improving the service itself or creating infor-
mation that can be useful somewhere else.’’
Service offerings become attractive to users because of what other users
provide. The analytical model in Sect. 2 does not take this notion into account
when modeling a user’s utility simply as the sum of the utility gained by the
provider’s service and the external utility. Instead, the community business model
postulates that the simple existence of a community, possibly created around a
free service, may give rise to other utilities that can become the basis for other
service offerings. This scenario also implies the need for a model that extends the
one presented in Sect. 2 to enable consideration of a multiparty system: namely
the provider of a (potentially free) service enabling the creation of a community
of customers whose collaboration enables new service offerings by the original or
new providers.

3.6 Gift economy (Anderson 6)

In gift economies, valuable goods and services are exchanged without any explicit
immediate or future rewards. This is similar to the communities of information
exchange we are witnessing today where people share their expertise and
knowledge for free. Consider, for example, the case of web sites, where software
developers can post and answer questions. Information is a non-trivial good that can
be gifted at practically no cost. The free-software community is another example of
an information gift economy. Programmers make their source code available,
allowing anyone to copy and modify or improve the code. These downstream
developers, depending on the software license attached, may be prevented from
even monetizing their improvements or they may have to share their gains with the
developers of the original software. In any case, these exchanges are diametri-
cally opposed to market-based trades made in a market economy, as described by
Sect. 2. (Also see http://www.en.wikipedia.org/wiki/Gift_economy;…/free-soft-
ware;.../community; and …/source_code).
Overall, the six business models described by Anderson (2009) and presented in
this section can only be profitable as long as some people (or organizations) are
willing to pay, at least in some circumstances. These models are not such that most
customers are the ones who pay. Generally, Anderson’s (2009) perspective departs
significantly from the dyadic view of services (service provider/customer) described
in Sect. 2 where the service provider receives fees from customers who pay for the
service plans they use.

123
A tale of two pricing systems for services 39

4 Conclusions and conjectures

The analytical model presented in Sect. 2 describes how customers are willing to
pay for some services. The model also includes the tradeoff that customers
encounter between paying a lower price and having to spend more time (with
advertising and self-service). The model is applicable to the pricing activities of
traditional and some newer service offerings, such as cell-phones, telecommuni-
cations generally, Internet connectivity, cable TV, residential electricity, and water.
The model, however, falls short in its ability to describe many of the types of service
offerings described by Anderson (2009) and discussed in Sect. 3.
To more fully describe modern service applications (that are online, collabora-
tive, including a community of users, and often free) significant extensions to the
model presented in Sect. 2 are needed, including the following features:
1. Extension 1: Heterogeneity of User Preferences: There would have to be some
explicit heterogeneity of preferences represented in the model, whereby a
majority of consumers would prefer the free service, but a minority of
consumers would be willing to pay fees for the premium version of the service.
This minority of people would be willing to pay to save time; to reduce risk; for
things they love; for status; and, because they are hooked, according to
Anderson (2009). The fees they pay would cover the costs of providing the
service and allow for a majority of users to access parts of the service for free.
Indeed, more complex versions of the model should acknowledge the
tremendous diversity of users with different needs, desires, and usage patterns.
2. Extension 2: Inclusion of Third Parties: A key source of revenue to service
providers in many cases is from third-party entities, including advertisers and
providers of related services. The model presented in Sect. 2 assumes only a
provider/customer scenario whereby extensions would have to recognize salient
features of advertising environments, including the number of customers
reached by ads, the low cost of ads, and the fact that some customers find the
targeted ads useful. The benefits to advertisers would need to be represented in
these models such as increased market awareness, trial rates, and even repeat
purchase behavior of customers. Third party revenues could also arise from
other affiliate relationships whereby clicks or referrals generate revenues to the
managers of the virtual community. Provisions for each of the third-party
entities would need to be included in the model.
3. Extension 3: Value Brought by Users: The service users, by virtue of the fact
that they use the service, can augment the content available to all the users. This
is a content externality distinct from the well-known network externality (Katz
and Shapiro 1985). The network externality states that having more passive
users on a network enhances the value of the network because more people can
be reached on the network. The content externality that we emphasize states
that users are active in co-creating the content in the network, and that having
more users co-create the content leads to economies of scale benefiting all
community participants. Accordingly, the community business model postu-
lates that the simple existence of a community, possibly created around a free

123
40 K. Lyons et al.

service, may give rise to other utilities that can become the basis for other
service offerings. The complexity of modeling content and other value brought
by the community of users would need to be included in the model.
Each of these proposed model extensions describe community co-generation of
value with a community facilitator sharing some (but not all) of the value that the
community creates. Such a perspective differs markedly from the perspective of the
dyadic view of the relationship between a service provider and the customer
described in Sect. 2 wherein the service provider receives fees from customers who
pay for the service plans they use.
Future pricing models for online and other informational services, thus, need
different perspectives than past pricing models for products and more traditional
services. These future models should build upon collaborative co-generation of
value among a community of users and not just co-generation of value between
service providers and its customers. We need pricing models that describe the
economic relationships between the various participants in a collaborative
ecosystem, whereby pricing and property rights are designed to induce productive
coordination of activities in ways that dramatically diverge from pricing systems of
neoclassical economics.

Acknowledgments The authors acknowledge the constructive comments of participants at a workshop

held at the Centers for Advanced Studies Conference in October 2008 (CASCON, 2008), including input
and examples from Paul Sorenson (University of Alberta), Sasha Chua (IBM), Timo Ewalds (founder of
Nexopia.com), Henry Kim (York University), and Stephen Perelgut (IBM). The authors’ research has
been supported by NSERC, the IBM Centers for Advanced Studies, iCORE, Alberta Advanced Education
and Technology, the Social Science and Humanities Council of Canada, and the University of Alberta
School of Retailing. The authors contributed equally to this paper.

Appendix 1

Proof of Proposition 3: With Plan 1, substituting p^k ¼ ck and s^k ¼ 0 into (3) yields a
maximization problem that has first order conditions described by (9). So Plan 1
P
provides an (interior) solution for (8). Then F^ ¼ f ð^ q1 ; . . .; q^k Þ  Kk¼1 ck q^k also
makes it so that (7) holds. Therefore, Plan 1 will also solve (6).
With Plan 2, if the marginal utility of ð^q1 ;...^ qK Þ
qk ;...;^
oqk is positive, then ^sk ¼ q^k and
p^k ¼ 1 will guarantee a corner solution with optimal quantity choices q^k ; k ¼
1; . . .; K; described by (8). Then F^ ¼ f ð^
q1 ; . . .; q^k Þ also makes it so that (7) holds.
Therefore, Plan 2 will also solve (6).
Proof of Proposition 4: With Plan 1, because of1 ð^q11 ;...^ qK1 Þ
qk1 ;...;^
oqk1 ¼ ck ; of2 ð^q12 ;...^ qK2 Þ
qk2 ;...;^
oqk2 ¼
ck and of1 ðq1 ;...;q
oqk
k ...;qK Þ
[ of2 ðq1 ;...;q
oqk
k ...;qK Þ
; s^k;1 ¼ q^k1 [ s^k;2 ¼ q^k2 : And it is true that the
lump-sum quantities in the two options are the optimal quantities that the customers
should choose no matter who chooses which payment option. The reason is the
really large additional unit prices. So if a customer in Segment 1 chooses option 2,
he or she will choose s^k;2 ¼ q^k2 : And if a customer in Segment 2 chooses option 1,
he or she will choose s^k;1 ¼ q^k1 : If the service provider wants to extract all the

123
A tale of two pricing systems for services 41

surplus from customers in Segment 1, i.e. M = 0, customers in Segment 1 would

choose option 2 instead of option 1 because such customers can get a positive net
utility from option 2, which is M ¼ f1 ð^ q12 ; . . .; q^K2 Þ  f2 ð^ q12 ; . . .; q^K2 Þ: Thus, if the
service provider wants customers in Segment 1 to choose option 2, the service
provider needs to give up what such customers 1 can obtain from choosing option 2.
That is why we have the M term in option 1. Segment 2 customers will not choose
option 1 because they will obtain negative net utility with option 1).
With Plan 2, since option 2 is the same as in Plan 1, the service provider has to
give up same amount in the fixed fee in option 1 to incent customer 1 to choose
option 1, which is M ¼ f1 ð^ q12 ; . . .; q^K2 Þ  f2 ð^ q12 ; . . .; q^K2 Þ: All the rest of the
analysis in solution 1 applies to this solution.
Since the main concern is to incent customer 1 to choose option 1, it does not
matter if fee-for-service or a fixed fee is emphasized in option 1. As long as option 2
is a fixed fee payment option, the service provider can differentiate the two
segments using either Plan 1 or Plan 2.
Proof of Proposition 5A: When the service provider incent customer 1 to choose
PK
option 1 instead of option 2, it incurs an additional variable cost, ck q^k1 
k¼1
PK
k¼1 ck q ^k2 because customer 1 chooses the higher quantity q^k1 . Thus, the service
provider wants to differentiate the two customers only when the gain achieved from
incenting customer 1 is no less than the cost of incenting, i.e. f1 ð^ q11 ; . . .; q^K1 Þ 
PK PK
f1 ð^
q12 ; . . .; q^K2 Þ  k¼1 ck q^k1  k¼1 ck q^k2 : The condition can also be written as
P P
q11 ; . . .; q^K1 Þ  Kk¼1 ck q^k1  f1 ð^
f1 ð^ q12 ; . . .; q^K2 Þ  Kk¼1 ck q^k2 : Then the left side is
the gross market surplus that the service provider receives from customer 1 when
the two customers are differentiated and the right side is the gross market surplus
that the service provider receives when the two customers are not differentiated.
Proof of Proposition 5B: With Plans 1 and 2, customer 1 (who values the services
more than customer 2) chooses quantity q^k1 ; that is, more than what customer 2
chooses, q^k2 : Customer 1 pays f1 ð^ q11 ; . . .; q^K1 Þ  f1 ð^ q12 ; . . .; q^K2 Þ þ f2 ð^ q12 ; . . .; q^K2 Þ
in both plans. Customer 2 pays f2 ð^ q12 ; . . .; qK2 Þ in both plans. Since
^
f1 ð^
q11 ; . . .; q^K1 Þ  f1 ð^
q12 ; . . .; q^K2 Þ þ f2 ð^ q12 ; . . .; q^K2 Þ [ f2 ð^ q12 ; . . .; q^K2 Þ; customer 1
pays more than customer 2.

References

Anderson C (2009) Free: the future of a radical price. Hyperion, New York
Eisenhardt K (1989) Agency theory: an assessment and review. Acad Manag Rev 14(1):57–74
Holmstrom B (1979) Moral Hazard and Observability. Bell J Econ 10(1):74–91
Katz ML, Shapiro C (1985) Network externalities, competition, and compatibility. Am Econ Rev
75(3):424–440
Liedtke M (2009a) Something to Tweet About: Twitter Valued at $1B. Associated Press http://
abcnews.go.com/Technology/AheadoftheCurve/wireStory?id=8673657. Accessed 23 April 2010
Liedtke M (2009b) Google, Microsoft now mutual friends of Twitter. Associated Press http://
www.thestar.com/news/sciencetech/technology/twitter/article/714187–google-microsoft-now-
mutual-friends-of-twitter. Accessed 23 April 2010

123
42 K. Lyons et al.

Lyons K, Playford C, Messinger PR, Niu RH, Stroulia E (2009) Business Models in Emerging Online
Services. In: Nelson ML, Shaw MJ, Strader TJ (eds) Value Creation in E-Business Management,
15th Americas Conference on Information Systems, AMCIS 2009, SIGeBIZ track, San Francisco,
California, August 2009, Selected Papers, pp 44-55
Rappa M (2008) Business Models on the Web. http://digitalenterprise.org/models/models.html. Accessed
23 April 2010
Vargo SL, Lusch RF (2004) Evolving to a New Dominant Logic for Marketing. J Mark 68:1–17
Varian HR (1992) Microeconomic Analysis, 3rd edn. WW Norton and Co, New York

123