Table of Contents
Policies influencing the diffusion of instant messaging
Andreas Grofler and J orn-Henrik Thun
Mannheim University
Industrieseminar, Schloss
D-68131 Mannheim, Germany
Tel.: +49 621 181-1583/1584
Fax: +49 621 181-1579
E-Mail: agroe@ is.bwl.uni-mannheim.de
thun@is.bwl.uni-mannheim.de
Abstract
Many information and communication products show characteristics that are called
network effects or positive demand externalities, i.e. their utility for users depends on
how many other people also use the product. The paper investigates the diffusion of
such products. Instant messaging is taken as an example. Strategic options of the
players in the instant messaging market are presented and analyzed on the basis of a
system dynamics model.
Despite the bust of the so-called “New Economy” there can only be little doubt that in
today’s economies networks have become crucial. This statement can be understood in
two ways. Firstly, we observe a proliferation of computer networks giving the ability for
apparently indefinite chances to communicate and to access information. Secondly, we
realize the increased importance of informal networks constituted by users of a product
or technology that leads to an inter-dependency of these users (Shapiro and Varian
1999).
The trend towards an information society has emphasized the importance of goods
satisfying information and communication needs. Most of these products are said to
show positive demand externalities, also called “network externalities”. Network
extemalities are present when the number of consumers who purchase a particular good
is an important characteristic of that good, which affects the utility derived by
consumers either directly or indirectly (Katz and Shapiro 1985). Classical examples of
goods showing network externalities are fax machines, e-mail, or computer platforms.'
In a systemic view this can be described as a positive feedback: the more adopters
use a product (“installed base”), the higher the utility for others using this product will
be, hence, the more likely others will be attracted to its use, and so forth. New adopters
are attracted by communicating with current adopters. Ultimately, the number of
potential adopters of a product limits the growth process. A causal loop diagram
depicting these feedback loops is shown in Figure 1.
Ko Vw MY
utilization *3& 2g adopters zak communication 4-5 potential
4 a. contacts adopters
Figure 1: Causal loop diagram of diffusion with network effects
Many implications have been derived from the positive feedback structure, for
example the need to grow faster than the competition with the consequence of
accumulating high losses during the growth period (Kelly 1998, Evans and Wurster
2000). Also, concepts like compatibility and switching costs of customers have become
important (Shapiro and Varian 1999).
The next section contains a brief literature review on system dynamics studies
about product diffusion. After that, the domain of application discussed in this paper—
instant messaging—is described. The third section presents the structure of a system
dynamics model that is used for policy analyses in the instant messaging market. Such
analyses are conducted with the help of simulation experiments in the following section.
The last section of this paper deals with the transferability of insights to other markets
and with future research that needs to be done.
Diffusion models in the system dynamics literature
Diffusion models are one of the most prominent and wide-spread uses of system
dynamics modeling. For instance, one can find papers about diffusion processes in all
proceedings of the last five system dynamics conferences (1997-2001). Past issues of
the System Dynamics Review contain some articles about diffusion, for instance
Milling (1996) and Maier (1998). Areas of application discussed in the system
dynamics literature are, for example, the diffusion of computer chips, digital TV, special
drugs, online banking, etc.
One of the most popular diffusion models for conventional products is based on
the work of Bass (1969). It is the basis of most system dynamics diffusion models. As a
mixed-influence model, it integrates effects of mass and personal communication
(Mahajan and Peterson 1985). It distinguishes between two types of customers:
innovators and imitators. Innovators become customers because they are interested in
novelties. Imitators ground their decision to buy a product on the behavior of other
members of the population. The basic Bass diffusion model and many derivatives and
improvements are reviewed in Mahajan et al. (1990). Some further elaboration and
clarification on his model can be found in Bass et al. (1994).
Sterman (2000) gives a basic feedback- oriented interpretation of the Bass model
that consists of a balancing (advertising) and a reinforcing (word-of-mouth) feedback
loop. The model he develops is a first-order system (with two mutually dependent
stocks); the adoption rate is determined by the advertising and the word-of-mouth effect
and limited by market saturation. As an example, the diffusion of VAX computers in
the 80s is replicated with the help of the small system dynamics model.
Milling (1996) and Maier (1998) have employed and enhanced Bass’s ideas in
various system dynamics based analyses of product diffusion processes. Their usage of
system dynamics is motivated by shortcomings of the original concept that provides no
explanation (1) why diffusion actually occurs, and, (2) how the diffusion process can be
influenced by management (e.g., using different price strategies). Their modeling
approach includes the explicit consideration of competition, repeat purchases and
product substitution by newer product generations that are technically more advanced
and that cannibalize earlier product types.
Recently, Thun et al. (2000) have conducted a pilot system dynamics study on
how to influence the diffusion of products with network extemalities. With the help of a
relatively small system dynamics model, they investigated different strategies to secure
sales growth and market penetration of a single network product. As basic findings they
state the necessity to augment “the pool of interesting communication partners of every
user in the installed base. Ways to achieve this are
e marketing measures to make users communicate more with each other and also
with formerly unknown people (e.g., ‘communities’ ),
° technical advances that make it possible to use a network product in new ways.
This would primarily increase the utility of the product but— in a next step— could
also make users communicate with new people (e.g., ‘SMS’). Furthermore,
communication with more than one partner could be made possible (e.g.,
‘conferencing’ ), and
e extending the installed base indirectly by creating compatibility to other
products.” (Thun et al. 2000)
However, their approach is limited to the examination of a hypothetical network
product. Furthermore, they do not consider dependencies of competing network
products. This paper tries to build on the work of Thun et al. (2000) by applying a
system dynamics model of similar scale and scope to the real world diffusion of instant
messaging, which is discussed in more detail in the next section.
The model in this paper resembles Sterman’s (2000, p. 393) model of network
effects. The differences compared to his approach are the threefold: (1) the number of
competitors is extended to three because of the instant messaging market's
characteristic, (2) a product's attractiveness is described and explained in more detail,
and (3) discards and repeat purchases are possible.
Instant Messaging
Instant messaging (IM) is an Internet application that allows communicating directly
and synchronously with partners all over the world using not only text (like Unix chat
programs do for twenty years now) but also other media, for instance, graphics, sound
and video. All data is transmitted in real-time. Other than e-mail, instant messages— as
the name implies— can be seen instantly on the screen of the communication partner.
Users just need to start the application once which then allows to chat, exchange
pictures or send files.
A precondition for communication, however, is that communication partners use
the same software program because most of the widely used clients are not compatible
with each other. They use different, proprietary protocols for data exchange and have
different user interfaces.’ Furthermore, users need to exchange their IM addresses
beforehand and they must mutually accept each other as a potential communication
partner within the software (“buddy list”). If users log in they immediately get the
information which of their buddies is online, and, vice versa, they are displayed as being
online as well to their partners. One can exchange messages with one specific
communication partner or with a group of partners.
Technically, three modes exist how IM works: (1) a centralized connection, (2) a
peer-to-peer connection, or (3) a combination of both. In a centralized connection all
users are connected via a network of servers, which handle all data transmission
activities. In a peer-to-peer connection only log in information is managed by a central
server. A fter establishing a communication path between two (or more) IM clients, data
is transmitted directly between these clients. In a combination solution, small (text-
based) messages are send via servers, bigger files (like voice or graphics) are
transmitted directly from client to client.
The first successful instant message client was ICQ (“I seek you”) from Mirabilis,
launched in 1996. In the first two years the program was downloaded more than 10 Mio.
times. Shortly after the success of ICQ became evident AOL launched a similar product
called AIM (AOL Instant Messenger), which allowed easy communication between all
AOL users. In 1998 AOL acquired Mirabilis including ICQ. Microsoft entered the
instant messaging market in 1997 with its product Microsoft Messenger and has acted
as the basic competitor of AOL since. Furthermore, there exist smaller providers of
instant messaging services and applications, for instance, Y ahoo, Sonork, Odigo, Gaim
and Fire.
All IM clients can be downloaded for free. The providers use IM as public
relation tools and try to create synergies to their original businesses. For instance, AOL
hopes to attract users as an Intemet provider and Microsoft wants to tie instant
messaging to the sales of PC operation systems. Additionally, usage data of users of IM
can be used for marketing purposes. The display of (personalized) banners on the
program’s user interface is possible. Another potential business area lies in the sales of
content, for example, selling pictures that users want to exchange. Around 200 Mio.
people are expected to use instant messaging in 2004.
In order to derive some estimations for parameters used in the model we
conducted a small survey among our students at Mannheim University. We asked 200
undergraduate and graduate business administration students about their knowledge
about and their usage pattems of instant messaging. 89,5 % claim to know what instant
messaging is, 65% posses an IM user account. Further results of this survey are
presented in the next section along with the discussion of the structure and
parameterization we used to develop the system dynamics model.
Model description
Asa basis for the simulation experiments we conducted, a system dynamics model was
created that builds on the work described in Sterman (2000) and Thun et al. (2000). The
basic structure of this model is depicted in Figure 2. As with most diffusion models
there is an untapped market (potential adopters). Potential adopters occasionally
become adopters of the IM client of one of the three players in the market (increasing
the installed base): for simplification reasons we combine AIM and ICQ as AOL,
Microsoft (MSN), and all other instant messaging providers. Some adopters quit using
the product, either ultimately or they become again potential users and adopt once more
in the future. Furthermore, users switch between the three installed bases when they are
unsatisfied with the utility their current application provides. They then change to an IM
provider that promises better utility.
installed bases
\ dedoption
others
to)
Al
potential adoption
adopters
L
MSN 4
repeating potential adopters
Figure 2: Block diagram of diffusion model
Of course, the actual simulation model is more complex than Figure 2. The
stock/flow diagram for one instant messaging provider (AOL) is depicted in Figure 3.
The structure for MSN and the other instant messaging programs is equivalent. Thus,
basically the actual model is three times the size of the one depicted in the figure.
Equations for the complete model can be found in the appendix.
Basically, the structure is similar to the one that was presented in Thun et al.
(2000). The adoption of an instant messaging client is laid down by the effect of two
groups of users: innovators and imitators (Bass 1969). How many potential adopters
become innovators or imitators is tied to the so-called innovation coefficients alpha and
beta. In the model used in this study we have endogenously calculated beta (i.e. set its
value within a feedback loop), but left alpha exogenous.
Coefficient alpha, which determines the fraction of potential adopters that become
innovators, is exogenously set as a parameter. For clarification purposes we have split
the variable into two parts: the degree of advertisements (ADV) placed to support the
product and the effect of reports in magazines and other media (REPORTS). Alpha
causes people to start using the particular instant messaging program without being
influenced by direct contact with other users. Innovators are necessary to start the
diffusion process (Milling and Maier 1996).
repeat AOL EXP FRAC FRAC REPDIS
CONTACT koh wall
oe ge
AOL aan
EMAIL AOL REL ADOP FRAC
Ty re
util per ae
SPER: vo Ny dedontion frac
— AOL AOL
bea AOL PATIENCE AOL
alpha K
, Installed Base AOL mera Tumpover
dedoption AOL dee AGL
[-—_ ers
ma
ua — SWITCH FRAC
switch AOL-MSN
a AOL util perus
\OL>
Potential
Adopters <SWITCE
switch AOL-A!
Figure 3: Stock/flow diagram for one instant messaging provider (AOL)
Coefficient beta determines the number of imitators. They are responsible for the
ultimate success of an innovation because only when imitators are attracted to a product
big number of users can be achieved. In contrast to the innovators imitators only start
applying the IM program after they had contact with other current users. As a specialty,
they can be invited by users via direct e-mail invitation (E-MAIL). In the model we
assume that potential adopters receiving such an invitation always start using instant
messaging. The most important part of the model, however, is the word-of-mouth
feedback loop (resulting in variable wom). Starting from the installed base the utility for
each user is calculated (util per user), which is simply assumed to be the number of
other users in the installed base (number of possible communication contacts) in
comparison to all potential users. This (actual) utility is compared to an expected utility
(exp utility). The higher actual utility is compared to expected utility the stronger is the
word-of-mouth effect, i.e. the more potential users can be convinced to start using the
IM client. A contact rate determines the number of contacts between users and potential
adopters. Expected utility is dependent on a relevant adopter fraction (REL ADOP
FRAC) that symbolizes that usually every user only wants to communicate with a small
fraction of other people in the installed base. The relevant fraction of all potential
adopters of IM (REL ADOP FRAC) is multiplied with an aspiration level (EXP FRAC)
that represents the fact that people are satisfied when they can reach a certain amount of
interesting communication partners via instant messaging.
If the relation between utility and expected utility is not satisfying users start to
quit using this instant messaging program (dedoption). Two possibilities exist how they
can proceed in this case. Firstly, they can completely discard using instant messaging at
all and for all times. Secondly, the can stop using instant messaging now but might start
using it again in the future, i.e. the become potential adopters again (repeat). The
relation between the two cases is set through a constant variable (FRAC REPDIS). In
the simulations presented in this paper, repeat purchases are switched off.
Furthermore, users switch to other providers of instant messaging. To mimic this
case, a certain fraction of users of provider 1 switches to provider 2 if provider 2 offers
more utility for its users. The switching function is implemented between all three
competitors in the model., i.e. between AOL and MSN, MSN and other providers, AOL
and other providers.
The values of the model parameters in the base run can be found in the model
listing in the appendix. Most of the variables are accompanied by a comment clarifying
their meaning. Constant values were established using three methods: (1) historical
values, for instance, the market entry times of the competitors, (2) derived from the
survey, for example, the switching fraction, or (3) estimated, if possible according to the
literature, like the contact rate.
Policy analyses
The primary goal of the simulation analysis was to test different policies and to derive
recommendations for successful diffusion management of instant messaging. In the base
tun of the simulation model, the three market players were initialized using the same
values for all parameters except time of market entry (ENTRY xxx). With this parameter
setting AOL always ends up in a monopoly position when it is taken into account that it
was first to the market, twelve month earlier than their competitors (Figure 4). As in all
following graphs of simulation results, the time scale varies from 0 to 120, meaning it
starts in the beginning of 1996 and runs for ten years. The y-axis symbolizes 200 Mio.
people at maximum, which is the estimation for the number of potential adopters of
instant messaging and, thus, its initial value in all simulations. However, scaling for
AOL and Microsoft/alternative providers often vary and differ in the following
diagrams. Furthermore, because we do not differentiate between policies for Microsoft
and the alternative providers in this paper, simulation results for these two players are
basically identical in all cases depicted here.
200 user
2 user
100 user
1 user
0 user
0 user
0 12 24 36 48 60 72 84 96 108 120
Time (Month)
Installed Base AOL : Base run: user
Installed Base MSN : Base run: user
Installed Base Alt : Base run user
Figure 4: Simulation results for base run of the model
In the base run AOL achieves a monopoly position, absorbing nearly the complete
base of potential adopters (some quit using the product and do not start using it again,
therefore AOL’s installed base is slightly smaller than 200 Mio. people). The diffusion
of the product is very slowly in the beginning (1996 until 2000) depicting the so-called
“penguin effect” (Farrell and Saloner 1986). However, in the three years after that (2001
until 2003) a rapid diffusion takes place, which is also called “bandwagon effect”
(Leibenstein 1950). Microsoft and the alternative providers do not reach a critical mass
of users and, thus, diffusion does not take off.
A sensitivity analysis showed that the models behavior is highly sensitive
regarding market entry time (Figure 5). In this analysis market entry time for AOL is
varied from 0 to 12. One can see that only in a small amount of cases AOL’s installed
base just reaches a third of the total market, i.e. each of the three competitors got a
market share of 66 Mio. customers. This only happens when the market entry time of
AOL approaches the entry time of the other two players, i.e. all enter the market around
the same point of time (month 12, meaning the beginning of 1997). In the majority of
cases, however, even a small lead conceming entry time leads to a significantly bigger
market share compared to its competitors. This result is in accordance with the
literature, which suggests that— everything else being equal—a pioneer will always
succeed when diffusion follows the Bass model. This argument is (particularly) valid
for markets with network extemalities.
Base run
50% 75% (NNN 95% 100%
Installed Base AOL
200
150
100
50
0 30 60 90 120
Time (Month)
Figure 5: Sensitivity analysis of market entry time (showing installed base of AOL)
Because the earlier market entry of AOL is a historical fact Microsoft and the
altemative providers need to take on other measures in order to overrule AOL.
Basically, their strategic options offer four possible ways:
1 Increasing the number of innovators (coefficient alpha), i.e. increasing
the effect of reports in the media or of advertisement measures.
2. Directly increasing the number of imitators (coefficient beta) which can
be achieved by a number of possibilities, e.g. diminishing the gap
between actual and expected utility (by lowering the aspiration level or
the relevant adopter fraction) or increasing the contact rate between
adopters and potential adopters.
3. Indirectly increasing the number of imitators (coefficient beta) by
enlarging the installed base, for instance through compatibility to other
providers. This, in tum, would increase the actual utility of all users in
the joint installed base.
4. Lowering the number of dedopters, for instance through increasing the
patience of the members of the installed bases.
In the rest of this section, some of these alternatives are tested and simulation
results from the model are presented. If not otherwise stated the development of the
installed bases of the three market players is depicted.
In Figure 6 the result for an increase in coefficient alpha for both, Microsoft and
the other providers is presented. This case is to a certain degree hypothetical because the
main effect of a higher number of innovators occurs in the beginning of a product's
diffusion; in the case of instant messaging in the past (around 1997/98). In other words,
this simulation experiment does not offer any strategic possibilities which can be
employed today. However, one can observe that even if the values for innovators are
doubled for Microsoft and the other providers, AOL still reaches a by far better
position.
200 user
5 user
100 user
2.5 user
0 user
0 user
0 12 24 36 48 60 72 84 96 108 120
Time (Month)
Installed Base AOL : Alpha changed: user
Installed Base MSN : Alpha changed- user
Installed Base Alt : Alpha changed user
Figure 6: Simulation results for a higher number of innovators for Microsoft and alternative
providers
The E-MAIL constant for Microsoft and the alternative providers was used in the
next simulation experiment in order to strengthen the word-of-mouth effect for these
players. It was assumed that people directly invited via e-mail to use a specific instant
messaging program account for an autonomous increase of 3 % of imitator adoption.
Results of this experiment are depicted in Figure 7. It can be observed that with this
setting AOL’s success can be hindered; nevertheless, the other two players neither do
succeed.
200 user
15 user
100 user
7.5 user
0 user
0 user
0 12 24 36 48 60 72 84 96 108 120
Time (Month)
Installed Base AOL : e-mail on user
Installed Base MSN : e-mail on user
Installed Base Alt : e-mail on: user
Figure 7: Simulation results for e-mail invitations for Microsoft and alternative providers
In the next simulation run, the utility for adopters in the installed bases of
Microsoft and the altemative providers was increased by extending the number of
people in the installed bases. In order to achieve this, compatibility between Microsoft's
and alternative products was assumed, thus creating one virtual installed base. Results
of this simulation are shown in Figure 8. However, because the number of users in their
installed bases are relatively small compared to AOL due to the later market entry time
compatibility alone does not lead to a significantly better position compared to the base
Tun.
0 12 24 36 48 60 72 84 96 108 120
Time (Month)
Installed Base AOL : compatibility user
Installed Base MSN : compatibility user
Installed Base Alt : compatibility user
Figure 8: Simulation results for compatibility between Microsoft and alternative providers
The next figure (Figure 9) depicts the simulation’s results when the patience
factors for customers in Microsoft's and in the other's installed bases are increased.
Through public relations, promotional, or technical measures they must therefore
achieve that the people in their installed bases wait more patiently for the gap between
actual and expected utility to close. In this case the patience time was prolonged from
one to two years. As with the case of e-mail invitations this measure alone does not help
to succeed against AOL, however, it delays its success.
200 user
15 user
100 user
7.5 user
0 user
0 user
0 12 24 36 48 60 72 84 96 108 120
Time (Month)
Installed Base AOL : patience changed user
Installed Base MSN : patience changed- user
Installed Base Alt : patience changed user
Figure 9: Simulation results for a longer patience of customers of Microsoft and alternative
providers
The last experiment described in this paper deals with the combination of all
measures discussed so far. Its results are depicted in Figure 10. With this combination
of measures Microsoft and the alternative providers succeed over AOL, despite its
earlier market entry. Both, Microsoft and the alternative providers get a market share of
roughly 50 %, which means close to 100 Mio. customers in their installed bases.
2 user
100 user
1 user
50 user
0 user
0 user
0 14 28 42 56 70 84 98 112
Time (Month)
Installed Base AOL : combination- user
Installed Base MSN : combination: user
Installed Base Alt : combinatior: user
Figure 10: Simulation results for a combination of measures
Transferability of insights and further research
The system dynamics model presented replicates historical data from the instant
messaging market as far as it is available (with the exception of ICQ and some smaller
competitors real usage figures can only be estimated). Simulation analyses allow to find
leverage points for a successful diffusion of instant messaging. In contrast to many
other diffusion models, our aim was to examine competition between different
programs providing basically the same functionality. Thus, successful policies for one
organization to get the advantage over its competitors were investigated. The model
suggests that in the instant messaging market the pioneer (AOL) has a strong position
due to its being first to the market. The basic finding from AOL’s competitors point of
view is that only coordinated measures can help weaken this position and reverse the
“naturally” occurrence of the “first takes it all” phenomenon caused by the word-of-
mouth reinforcing feedback loop. One promising, but not sufficient measure is
compatibility between the Microsoft and the smaller players programs.
We assume that the basic structure of the model can be adapted to many diffusion
processes where products with network extemalities are involved. The main thing to
change structurally would be the number of competitors. Furthermore, some parameter
would need a reinterpretation or are obsolete, for instance direct invitation of other users
by e-mail.
Of course, many improvements and analyses need to be done within the instant
messaging context of the model. Besides always necessary efforts in validating
parameter and structure some ideas are:
We did not analyze different policies for Microsoft and the smaller providers of
instant messaging. For instance, it could be tested what effects compatibility
working in only one direction has or how different marketing budgets affect
performance. Furthermore, the effects of bundling the Microsoft Messenger
program to the Windows operation system can be investigated.
It should be taken into account that potential users make their decision not only
based on the comparison of actual and expected utility but to a great amount on
expectations of future utility. The same holds true for the discontinuation of
usage. Sterman’s (1987) TREND function could be used for this purpose.
Effects of a higher willingness to switch between providers have not been tested
intensively so far.
Some of the individuals in the customer base could be more active in
communicating the advantages of the product than others. Thus, the customer
base needs differentiation.
The reinforcing effects of complementary goods should be included. In this way,
the focus could also shift to modeling the diffusion of products with indirect
network externalities.
In reality, the number of potential adopters dynamically changes over time. This
effect could be incorporated into the model together with influences of economic
or demographic factors.
In the model so far, it is not discussed how the instant messaging providers try to
benefit from establishing their product as a quasi standard. This is an important
issue because apparently they give away their product for free. Additionally, there
is no structural element of financial and other resources that are needed to manage
the diffusion process, for instance by advertising.
Some of the policies presented are hypothetical insofar as it was assumed that
their effect would be evident from the time of market entry on. Of course, the past
cannot be changed. Therefore, a more detailed investigation of the effects of
changed policies in later phases of the diffusion process needs to be added.
Some of these extensions will be addressed in the final version of this papers or in
subsequent papers on this topic. In the future we want to extend our work in two
directions: firstly, detailing and improving the model of the instant messaging market,
and, secondly, applying the core structure of the model to other diffusion processes of
products with positive demand externalities.
References
Bass, F.M. (1969). A New Product Growth Model for Consumer Durables.
Management Science 15(5), 215-227.
Bass, F.M., Krishnan, T. and Jain, D. (1994). Why the Bass Model Fits Without
Decision Variables. Marketing Science 13(3), 203-223.
Bental, B. and Spiegel, M. (1995). Network competition, product quality, and market
coverage in the presence of network extemalities. Journal of Industrial
Economics 43(2), 197-208.
Evans, P. and Wurster, T.S. (2000). Blown to Bits. How the New Economics of
Information Transforms Strategy, Harvard Business School Press, Boston.
Farrell, J. and Saloner, G. (1986). Installed Base and Compatibility: Innovation, Product
Preannouncements, and Predation. American Economic Review 76(5), 940-955.
Katz, M.L. and Shapiro, C. (1985). Network Extemalities, Competition, and
Compatibility. American Economic Review 75(3), 424-440.
Kelly, K. (1998). New Rules for the New Economy. 10 Radical Strategies for a
Connected World, Penguin, New Y ork.
Leibenstein, H. (1950). Bandwagon, Snob, and Veblen Effects in the Theory of
Consumers’ Demand. Quarterly Journal of Economics 64(2), 183-207.
Mahajan, V. and Peterson, R.A. (1985). Models of Innovation Diffusion, Sage, Beverly
Hills/London/New Delhi.
Mahajan, V., Muller, E. and Bass, F.M. (1990). New Product Diffusion Models in
Marketing: A Review and Directions for Research. Journal of Marketing 54(1),
1-26.
Maier, F.H. (1998). New Product Diffusion Models for Decision Support in Innovation
Management. System Dynamics Review 14(4), 285-308.
Milling, P.M. (1996). Modeling Innovation Processes for Decision Support and
Management Simulation. System Dynamics Review 12(3), 211-234.
Milling, P.M. and Maier, F.H. (1996). Invention, Innovation und Diffusion, Duncker &
Humblot, Berlin.
Shapiro, C. and Varian, H.R. (1999). Information rules, Harvard Business School Press,
Boston, MA.
Sterman, J.D. (1987). Expectation Formation in Behavioral Simulation Models.
Behavioral Science 32, 190-211.
Sterman, J.D. (2000). Business Dynamics, Irwin McGraw-Hill, Boston, MA etal.
Thun, J.-H., Grofler, A. and Milling, P.M. (2000). Simulating network effects of New
Economy goods. Davidsen, P.I., Ford, D.N. and Mashayekhi, A.N. (eds.):
Sustainability in the Third Millennium - Proceedings of the 18" International
Conference of the System Dynamics Society, Bergen.
Notes
1. One can distinguish between direct (e.g., e-mail) and indirect network effects
(e.g., computer platforms). Products showing direct network externalities do not
have any utility per se and are only useful if others also use this product; indirect
network externalities exist when both, an original utility of a product and a utility
from the usage of others can be assumed (Bental and Spiegel 1995).
2. Currently, the Internet Engineering Taskforce (IETF) works on the development
of a standard protocol for the exchange of instant messages (Instant Messaging
and Presence Protocol, IMPP). However, so far exist only draft versions of this
future standard. Furthermore, it is unclear in how far the big players in the instant
messaging market will support the protocol.
Appendix
adoption Alt=
IF THEN ELSE( Time>=ENTRY Alt, (alpha Alt*Potential A dopters)+{beta Alt/
INI*Potential A dopters* Installed Base Alt) ,0 )
Units: user/Month
adoption AOL=
IF THEN ELSE( Time>=ENTRY AOL , (alpha AOL*Potential Adopters) +{(beta AOL/
INI)*Potential Adopters*Installed Base AOL) ,0 )
Units: user/Month.
adoption frac Alt=
1-(exp utility-util per user A It)
Units: Dmnl
adoption frac AOL=
1-(exp utility-util per user AOL)
Units: Dmnl
adoption frac MSN=
1-(exp utility-util per user MSN)
Units: Dmnl
adoption MSN=
IF THEN ELSE( Time>=ENTRY MSN , (alpha MSN*Potential A dopters)+(beta MSN/INI
*Potential A dopters* Installed Base MSN) ,0 )
Units: user/Month.
ADV Alt=
1e-006
Units: Dmnl
Effects of advertisment efforts on coefficient alpha for
altemative providers
ADV AOL=
1e-006
Units: Dmnl
Effects of advertisment efforts on coefficient alpha for AOL
ADV MSN=
1e-006
Units: Dmnl
Effects of advertisment efforts on coefficient alpha for
Microsoft
alpha Alt=
ADV Alt+REPORTS Alt
Units: Dmnl
alpha AOL=
ADV AOL+REPORTS AOL
Units: Dmnl
alpha MSN=
ADV MSN+REPORTS MSN
Units: Dmnl
beta Alt=
wom Alt+'E-MAIL Alt"
Units: Dmnl
beta AOL=
wom AOL+"E-MAIL AOL"
Units: Dmnl
beta MSN=
wom MSN+"E-MAIL MSN"
Units: Dmnl
CONTACT RATE=
0.2
Units: Dmnl
Fraction of contacts between adopters and potential adopters
dedoption Alt=
Installed Base Alt*dedoption frac Alt/PATIENCE Alt
Units: user/Month
dedoption AOL=
(Installed Base AOL*dedoption frac AOL/PATIENCE AOL)
Units: user/Month
dedoption frac Alt=
IF THEN ELSE(util per user Alt>=exp utility, 0, 1-(util per user Alt/exp utility
)
Units: Dmnl
dedoption frac AOL=
IF THEN ELSE(util per user AOL>=exp utility, 0 , 1-(util per user AOL/exp utility
)
Units: Dmnl
dedoption frac MSN=
IF THEN ELSE(util per user MSN>=exp utility, 0 , 1-(util per user MSN/exp utility
)
Units: Dmnl
dedoption MSN=
Installed Base MSN*dedoption frac MSN/PATIENCE MSN
Units: user/Month
DEL REPDIS=
ik
Units: Month
Delay time before finally quit using IM or become potential
adopter again
discard Alt=
Tumover Alt*FRAC REPDIS/DEL REPDIS
Units: user/Month
discard AOL=
Turmover AOL*FRAC REPDIS/DEL REPDIS
Units: user/Month.
discard MSN=
Turmmover MSN*FRAC REPDIS/DEL REPDIS
Units: user/Month.
"E-MAIL Alt"=
0
Units: Dmnl
"E-MAIL AOL"=
0
Units: Dmnl
"E-MAIL MSN"=
0
Units: Dmnl
ENTRY Alt=
12
Units: Month
Time of market entry alternative providers
ENTRY AOL=
0
Units: Month
Time of market entry AOL
ENTRY MSN=
12
Units: Month
Time of market entry Microsoft
EXP FRAC=
0.5
Units: Dmnl
Aspiration level: what fraction of relevant adopters should be
reached
exp utility=
EXP FRAC*REL ADOP FRAC
Units: Dmnl
Expected utility from using instant messaging
FINAL TIME =120
Units: Month
The final time for the simulation.
FRAC REPDIS=
1
Units: Dmnl
What fraction of dedopters finally quit using instant messaging
INI=
200
Units: user
Initial number of potential users (market size)
INITIAL TIME =0
Units: Month
The initial time for the simulation.
Installed Base Alt=INTEG (
adoption Alt+"switch AOL-Alt'+'switch MSN-Alt"-dedoption Alt,
0)
Units: user
Installed Base AOL=INTEG (
adoption A OL-dedoption AOL-"switch AOL-Alt"-"switch AOL-MSN",
0)
Units: user
Installed Base MSN=INTEG (
adoption MSN+'switch AOL-MSN"-dedoption MSN-"switch MSN-Alt",
0)
Units: user
PATIENCE Alt=
12
Units: Month
How long do customers wait until they stop using alterative
providers
PATIENCE AOL=
12
Units: Month
How long do customers wait until they stop using AOL
PATIENCE MSN=
12
Units: Month
How long do customers wait until they stop using Microsoft
Potential A dopters= INTEG (
-adoption Alt-adoption A OL-adoption MSN +repeat A OL +repeat A lt+repeat MSN,
INI)
Units: user
REL ADOP FRAC=
0.02
Units: Dmnl
What fraction of adopters is relevant, i.e. the user wants to
communicate with?
repeat Alt=
Tumover Alt*(1-FRAC REPDIS)/DEL REPDIS
Units: user/Month.
repeat AOL=
Tumover AOL*(1-FRAC REPDIS)/DEL REPDIS
Units: user/Month.
repeat MSN=
Tumover MSN*(1-FRAC REPDIS)/DEL REPDIS
Units: user/Month
REPORTS Alt=
1e-006
Units: Dmnl
Effects of media reports on coefficient alpha for altemative
providers
REPORTS AOL=
1e-006
Units: Dmnl
Effects of media reports on coefficient alpha for AOL
REPORTS MSN=
1e-006
Units: Dmnl
Effects of media reports on coefficient alpha for Microsoft
SAVEPER =
TIME STEP
Units: Month
The frequency with which output is stored.
"switch AOL-Alt"=
IF THEN ELSE(util per user Alt>util per user AOL, Installed Base AOL*SWITCH FRAC
, IF THEN ELSE( util per user Alt<util per user AOL , -(Installed Base Alt
*SWITCH FRAC), 0))
Units: user/Month.
"switch AOL-MSN"=
IF THEN ELSE(util per user MSN>util per user AOL, Installed Base AOL*SWITCH FRAC
, IF THEN ELSE( util per user MSN<util per user AOL , -(Installed Base MSN
*SWITCH FRAC), 0))
Units: user/Month.
SWITCH FRAC=
0.02
Units: Dmnl
What fraction switches to another provider because it provides
better utility
"switch MSN-Alt'=
IF THEN ELSE(util per user Alt>util per user MSN, Installed Base MSN* SWITCH FRAC
, IF THEN ELSE( util per user Alt<util per user MSN , -(Installed Base Alt
*SWITCH FRAC), 0))
Units: user/Month.
TIME STEP =0.25
Units: Month
The time step for the simulation.
Tumover Alt=INTEG (
dedoption Alt-discard Alt,
0)
Units: user
Tumover AOL=INTEG (
dedoption AOL-discard AOL,
0)
Units: user
Tumover MSN=INTEG (
dedoption MSN-discard MSN,
0)
Units: user
util per user Alt=
MAX (Installed Base Alt/INI, 0 )
Units: Dmnl
util per user AOL=
MAX (Installed Base AOL/INI, 0 )
Units: Dmnl
util per user MSN=
MAX (Installed Base MSN/INI, 0 )
Units: Dmnl
wom Alt=
adoption frac Alt*CONTACT RATE
Units: Dmnl
Word-of-mouth effect for altemative providers
wom AOL=
adoption frac AOL*CONTACT RATE
Units: Dmnl
Word-of-mouth effect for AOL
wom MSN=
adoption frac MSN*CONTACT RATE
Units: Dmnl
Back to the Top
Word-of-mouth effect for Microsoft
CC BY-NC-SA 4.0