See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/242636096

Growth Models for the Forecasting of New Product Market Adoption

Article · January 2008

CITATIONS READS

19 1,917

1 author:

Mladen Sokele
Zagreb university of applied sciences
27 PUBLICATIONS 118 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Determination of time criteria for assessment in Learning Management Systems View project

Research on the importance of blended learning components for student exam success View project

All content following this page was uploaded by Mladen Sokele on 30 October 2015.

The user has requested enhancement of the downloaded file.

Growth Models for the Forecasting of
New Product Market Adoption
MLADEN SOKELE

Appropriate forecasting of product market adoption enables optimal planning of resources, investments,
revenue, marketing and sales. Quantitative forecasting methods for the new product adoption rely on
the S-shaped (sigmoidal) growth model such as the logistic, the Richards and Bass growth models. This
paper presents adaptations of these models by introducing explanatory and marketing variables which
are suitable for the forecasting prior to product launch or in the early phases of the product life cycle.

Mladen Sokele
is Assistant
1 Introduction N(ti) – f (ti; α1, α2, ..., αk) = 0, i = 1, ..., k (1)
Director in
Croatian Tele- Growth models are widely used in quantitative
com Inc, Zagreb research in order to understand the forces that influ- System (1) is usually a nonlinear system, so iterative
ence growth in the sense of its dynamics, market numerical methods for its solution are needed to be
capacities as well as forecasting of growth in the performed (eg. Newton method).
future. Particularly, diffusion of innovation and new
technology, market adoption of consumer durables, In cases when k or more data points are available, the
number of users of subscription services (for exam- weighted least squares method can be used for
ple: telecommunications services) and allocation of parameter estimation to adjust the parameters of a
restricted resources at the beginning of a product life model so as to best fit a data set. The objective is to
cycle (PLC) have sigmoidal growth. Such products/ minimize the sum of squared difference between data
services that do not include repeat sales are denoted points and model evaluated points:
products, and the cumulative volume of adopted n
 2
products by users, number of users. S= wi · [N (ti ) − f (ti ; α1 , α2 , ..., αk )] (2)
i=1

Growth models based on the logistic, Richards and where wi are weights. When weights are equal to 1
Bass models involving explanatory and marketing (wi = 1), the method is called ordinary least squares
variables will be presented and examined for the fore- method (OLS).
casting application in cases of new product market
adoption. Minimisation of (2) can be done by software tools
such as Microsoft Excel solver. Analytically, values
of parameters are resulting from solution of the sys-
2 Growth Forecasting tem of equations (3):
Growth forecasting relies on the basic principle: The
∂S
growth model will be valid in the perceivable future, = 0, j = 1, ..., k (3)
∂αj
and the forecasting result could be obtainable by
extrapolation of observed values sequentially through By use of least squares method, obtained values for
time and supplementary information. In general, this parameters are statistically smoothed; ie. the influ-
principle is valid only for stable markets where inter- ence on parameter values due to particular measure-
nal forces remain the same (eg. same market segment ment errors (such as unanticipated seasonal variation,
boundaries, competition, cause-and-effect among uncertain measure, etc) is reduced.
products, etc) and without change of external influ-
ences (eg. technology, macroeconomics, purchasing For forecasting purposes, parameter estimation
power, regulatory, etc changes). This type of fore- is usually focused on the time interval near the last
casting belongs to the quantitative time series meth- observed data point. Thus, weights in equation (2)
ods [1, 11]. can be set to higher value for the most recent data
points than for data points in far history. For example,
For time series growth model f (ti; α1, α2, ..., αk) the geometric series for weights
based on a set of k parameters {αi}, at least k known
1
data points (ti; N(ti)) are needed for the complete wi = , q>1 (4)
q n−i
parameter estimation. In cases when exactly k data
points are available, parameters αi are solutions of
a system of equations (1):

144 ISSN 0085-7130 ©Telenor ASA 2009 Telektronikk 3/4.2008

leads to the following weights: 1 for (the last known results of judgmental forecasting in cases when little
point) tn , 1/q for tn-1 (the penultimate known point), or no historical data points are available. In addition,
1/q2 for tn-2, etc. the model can have auxiliary parameters which adjust
the model to practical requirements. Auxiliary param-
In some forecasting cases, model f (ti; α1, α2, ..., αk) eters do not need additional data points for their
is modified to include the fixed value of the last data determination.
point (tn, N(tn )). Therefore, the model has one param-
eter less, because αk is obtained from the equation: In general, grouping of forecast results for specific
market segments (eg. separate for residential seg-
N (tn ) − f (tn ; α1 , α2 , ..., αk ) = 0 (5) ment, for business segment and/or for segments
related to specific life-styles, etc) yields better
The above mentioned simplification is used only forecasting accuracy than aggregate forecasting
when it is certain that the last data point is obtained performed for the whole market.
with negligible measurement error.
Due to the measurement errors of input data, associ-
Furthermore, relationships between model parameters ated uncertainties of estimated model parameters
and explanatory and marketing variables can be used can be represented by a confidence interval. Conse-
for forecasting purposes, aiming at additional reduc- quently, forecasting results can be represented by a
tion of the number of unknown parameters in the prediction interval between pessimistic and optimistic
growth model f (ti; α1, α2, ..., αk), eg. including values. The range depends on determined confidence
information of exact time when product introduction level, which is typically 95 %. Besides that, a sensitiv-
starts, time and value of anticipated sales maximum, ity analysis of model parameters and/or explanatory
market capacity, product price, advertising expendi- variables should be deployed to examine what effect
tures, etc. This approach is also suitable for applying their variations have upon the forecasting result.

100 100

% %
a=A a=A
75 a = -A 75 a = 0.5•A

50 50

25 25

0 0
B-15

B-10

B-5

B+5

B+10

B+15

B-15

B-10

B-5

B+5

B+10

B+15

100 100
% %
b=B M= MC
75 b = B-5 75 M = 0.7•MC

50 50

25 25

0 0
B-15

B-10

B-5

B+5

B+10

B+15

B-15

B-10

B-5

B+5

B+10

B+15

Figure 1 Effect of logistic model parameter change on the form of S-curve

Cases (from top left to bottom right): positive and negative growth rate parameter; 50 % decrease of growth
rate parameter; decrease of time shift parameter for 5 time units (eg. years); and 30 % decrease of market
capacity parameter

Telektronikk 3/4.2008 ISSN 0085-7130 ©Telenor ASA 2009 145

 M Sales(t1 , t2 ) = L(t2 ) − L(t1 )
L(t)  
t2 + t 1 (9)
L’(t) ≈ (t2 − t1 ) · L
GR 2
The maximum of L'(t), as well as the time point tI
t
eaΔ -1 when L(t) has inflexion are obtained from the solu-
I tion of equation L''(t) = 0, where L''(t) is the second
M/2
derivative of L(t):

t d2 L(t)
(eaΔ -1)/2 L (t) = =
dt2
aM/4  
−a2 M · e−a(t−b) · 1 − e−a(t−b)
=  3 (10)
0 b
Time 1 + e−a(t−b)
From (10) follows that L(t) has inflexion for tI = b,
Figure 2 Characteristic values and points of the which is for a > 0 the maximum of L'(t), too (see
logistic growth model Figure 2):

aM
max L (t) = ⇔ a > 0; t = b (11)
4
3 Logistic Growth Model
The logistic model L(t) describes growth in the num- The value of the logistic model at point of inflexion is
ber of users observed over time in a closed market, (see Figure 2) L(b) = M / 2.
without the impact of any other product. The model
is defined by three parameters: M – market capacity, Accordingly, maximum of sales occurs at t = b = tI
a – growth rate parameter and b – time shift parame- when penetration is 50 %. For t1 and t2 near b from
ter. To emphasise the model’s dependence on its (9) follows that sales in time interval [t1, t2] can be
parameters, it is convenient to indicate the model as approximated by:
L(t; M, a, b) [11]:
aM
max Sales(t1 , t2 ) ≈ (t2 − t1 ) · (12)
M 4
L(t; M, a, b) = L(t) = (6)
1 + e−a(t−b) Asymptotes of logistic growth for positive and nega-
Figure 1 shows the effects of change of parameters a, tive parameter a can be summarized in (13):

b and M on the form of S-curve. L(t) = 0 ⇔ a > 0
lim
t→−∞ L(t) = M ⇔ a < 0
3.1 Characteristics of the Logistic Growth 
L(t) = M ⇔ a < 0
Model lim (13)
t→+∞ L(t) = 0 ⇔ a > 0
The logistic model is a widely used growth model
with many useful properties for technological and The growth rate GR for time interval Δ t is:
market development forecasting. Model (6) is the
L(t) − L(t − Δt)
solution of differential equation (7) which consists of GRΔt = =
L(t − Δt)
exponential growth term and negative feedback term
[5]. In the beginning, the growth of the logistic model 1 + e−a(t−Δt−b) (14)
= −1
is identical to exponential growth, but later negative 1 + e−a(t−b)
feedback slows the gradient of growth as L(t) app- For positive a, the growth rate is always positive and
roaches the market capacity limit M: the maximum of growth rate is when t –> – ∞ (see
Figure 2, grey columns):
 
dL(t) L(t)
= aL(t) · 1 −
dt M (7) max GRΔt = eaΔt − 1 ⇔ a > 0; t → −∞
←→ ←→
Exponential Negative
growth feedback When t = b = tI , a sales peak occurs. The growth rate
First derivative of L(t) is given in (8): at this point in time is half of its maximum value:

dL(t) a · M · e−a(t−b) eaΔt − 1

L (t) = =  2 (8) GRΔt = ⇔t=b (15)
dt 1 + e−a(t−b) 2
Contrary to S-shaped cumulative adoption L(t), adop- Above described characteristics of the logistic growth
tion per period (sales) is a bell-shaped curve (see Fig- model with its explanatory attributes can be used as
ure 2), and it is proportional to the first derivative helpful input for estimation or assessment of model
L'(t) of cumulative adoption: parameters for forecasting purposes.

146 ISSN 0085-7130 ©Telenor ASA 2009 Telektronikk 3/4.2008

3.2 Logistic Model through two Fixed Points M
The modification of model (6), which has embedded v·M
L(t)
values of two data points (ts , u . M) and (te , v . M) is
shown in Figure 3. For this case, it is suitable to
define new parameters ts and Δ t instead of a and b;
as well as to introduce two auxiliary parameters u and Δt

v. New parameters have explanatory meaning: time

ts when the product perceivably starts with penetra-
tion level u, Δ t – period needed until penetration
grows to level v, eg. characteristic duration from
product start to product maturity [8, 12]. u·M
ts t e Time
Parameters a and b in (6) should be substituted with
expressions (16) and (17), which are dependent on Figure 3 The logistic growth model defined via
input parameters u, v and Δ t: parameters M, ts , Δt, u and v
    
1 1 1
a= ln − 1 − ln −1 (16)
Δt u v
 
ln u1 − 1 A condition that must be satisfied for model (20) is
b = ts + Δt  1   
ln u − 1 − ln v1 − 1 (17) 0 < u < 1. Used simplification gives a framework for
the forecasting of new product adoption when little or
Conditions that must be satisfied for equations (16) no data are available. In Table 1 are presented result-
and (17) are 0 < u < v < 1. This modified model L(t; ing models for typical values of characteristic dura-
M, ts, Δt, u, v) needs three parameters: M, ts and Δ t to tion Δt for products, product families and basic tech-
be determined. Values of auxiliary parameters u and nologies according to model (20), but uniformed on
v are not needed to be determined, they just allow the the same natural logarithm base e.
forecasting practitioner to choose the levels of starting
and ending the penetration he/she wants to deal with. 3.3 Logistic Model Through one Fixed Point
Modification which has embedded value of one data
In the case of symmetrical u and v, ie. u = 1 – v, point (tp, N(tp)) in model (6) is called local logistic
equations become simpler: model LL(t):
 
2 1 LL(t; M, a, tp , N (tp )) =
a= ln −1 (18)
Δt u
Δt M · N (tp )
b = ts + (19) = (21)
2 N (tp ) + [M − N (tp )] · e−a(t−tp )

Therefore model L(t; M, ts, Δ t, u) has only one auxil- Local logistic model is useful for forecasting from the
iary parameter u: last observed point t > tp. The idea is that it is better
to start forecasting from a known base rather than to
M
L(t; M, ts , Δt, u) = 1 1−2(t−ts )/Δt (20) rely on an anticipated but un-modelled reversion to a
1+ u −1 historical trend. [4]

u = 5 %, v = 95 % u = 10 %, v = 90 %

M M
Δt = 2 years N (t) = N (t) =
1+ e−2.944(t−ts −1) 1+ e−2.197(t−ts −1)
M M
Δt = 5 years N (t) = N (t) =
1+ e−1.178(t−ts −2.5) 1+ e−0.879(t−ts −2.5)
M M
Δt = 10 years N (t) = N (t) =
1 + e−0.589(t−ts −5) 1 + e−0.439(t−ts −5)
M M
Δt = 15 years N (t) = N (t) =
1 + e−0.393(t−ts −7.5) 1+ e−0.293(t−ts −7.5)
Note: Characteristic duration Δt according to [6] can be assumed as follows: products consist of units sold that have a typical life cycle
of 6 to 10 quarters; product families consist of related products that have a typical business cycle of 5 years, and basic technologies
consist of a set of related product families that have a typical cycle of 10 to 15 years.

Table 1 Logistic model framework for forecasting of new product adoption prior to launch

Telektronikk 3/4.2008 ISSN 0085-7130 ©Telenor ASA 2009 147

 100 100
c= 0.2 c= 0.2
% %
c = 10 c = 10
75 Δt = D 75 Δt = D/2

50 50

25 25

0 0
ts ts+Δt ts ts+Δt

Figure 4 Richards model for different parameters c and Δt

3.4 Limitations of the Logistic Model R(t; M, ts , Δt, c, u) =

Although the logistic model is widely used for fore-   √ − 1−ts −c
√  1/ c u − 1 Δt

casting purposes, it is not suitable for modelling the = M 1 + 1/ c u − 1 · √ (25)

1/ c 1 − u − 1
product adoption when the number of users grows
fast instantly after the product is introduced. The
reason is in the shape of logistic growth that ‘hardly Like the logistic model, the Richards model cannot
starts to grow up’. This problem is visible from the model the time point when the product is introduced,
condition for equation (16), ie. it is not possible to ie. when N(t) = 0 because only for t –> – ∞ R(t)
model the time point when the product is introduced, approaches 0.
and its penetration is 0 (u = 0), because equations will
give infinity value for parameter a. This deficiency is The Richards model through one fixed point (tp,
solved with the Bass model. N(tp )) has the following form:

The second main deficiency is fixed inflexion point LR(t; M, a, c, tp , N (tp )) =

I (b, M /2), which is not crucial for most forecasting M
=   c
purposes, but it is solved with the Richards growth M (26)
1+ c
N (tp ) − 1 · e−a(t−tp )
model, which is sometimes called the four-parameter
logistic model [7]: and could be called the local Richards model due to
the similarity with the local logistic model. The
M
R(t; M, a, b, c) =  c (22) model is useful for forecasting from the last known
1 + e−a(t−b)
data point t > tp .
with parameters M – market capacity, a – growth rate
parameter, b – time shift parameter, and c – shape
parameter which determines position of the inflexion 4 The Bass model
point. R(t) has inflexion for t = tI: The best known model for a full description of the
genesis and extensions of new product market adop-
ln c
tI = b + ⇔ R (tI ) = 0 (23) tion (when interaction with other products can be
a
neglected) is the Bass model. In distinction from
Minimal value of R(tI) / M arises for c –> ∞ and can- logistic and Richards model, Bass model B(t) intro-
not be smaller than e-1 ≈ 0.368 (minimal vertical duces the effect of innovators via coefficient of inno-
position of an inflexion point). For c = 1, the vation p in differential equation of growth (27),
Richards model is identical to the logistic model and which makes it suitable for modelling market adop-
R(tI) / M = 0.5. Maximal value is without restriction, tion immediately after product is introduced. Model
ie. R(tI) / M –> 1 for c –> 0: considers a population of M adopters who are both
 c innovators (with a constant propensity to purchase)
R(tI ) c
e−1 < = <1 (24) and imitators (whose propensity to purchase is influ-
M 1+c
enced by the amount of previous purchasing) [2, 3,
Based on reparameterization shown in Section 3.2, 11].
the Richards model through two fixed points (ts, u . M)  
dB(t) B(t)
and (ts + Δt, (1 – u) . M); with condition that 0 < u < 1, = qB(t) 1 − + p (M − B(t))
dt M
has the form (see Figure 4): ← → ← →
Effect of imitators Effect of
(Logistic growth) innovators

(27)

148 ISSN 0085-7130 ©Telenor ASA 2009 Telektronikk 3/4.2008

Solution of the differential equation (27) gives the Bass Sales(t1 , t2 ) =  
t2 + t 1
diffusion model (28) defined by four parameters: M – = B(t2 ) − B(t1 ) ≈ (t2 − t1 ) · B  (30)
market capacity, p – coefficient of innovation, p > 0, 2
q – coefficient of imitation, q ≥ 0 and ts – time when Maximum of B'(t), as well as the time point when
the product is introduced, B(ts) = 0. To emphasise the B(t) has inflexion, is obtained from the solution of
model’s dependence on its parameters, it is conve- equation B''(t) = 0, where B''(t) is the second deriva-
nient to indicate the model as B(t; M, p, q, ts), t ≥ ts: tive of B(t):

1 − e−(p+q)(t−ts ) d2 B(t) (p + q)3

B(t; M, p, q, ts ) = M (28) B  (t) = = M
1 + pq e−(p+q)(t−ts ) dt2 p
 
q −(p+q)(t−ts )
The Bass model has the shape of an S-curve, identical pe − 1 · e−(p+q)(t−ts )
to the simple logistic growth model, but shifted down  3 (31)
1 + pq e−(p+q)(t−ts )
on the y-axis. Figure 5 shows the effects of different
values of parameters p and q on the form of S-curve, From (31) follows that B(t) has inflexion for t = tI:
with fixed values for M and ts .  
1 q
tI = t s + ln ⇔ B  (tI ) = 0 (32)
p+q p
4.1 Characteristics of the Bass Model
The Bass model has many common characteristics and maximum of B'(t) also occurs for t = tI, when it
with the logistic growth model. First derivative of has the value:
B(t) is given in (29):
(p + q)2
dB(t) max B  (t) = M ⇔ t = tI (33)
B  (t) = = 4q
dt
(p + q)2 e−(p+q)(t−ts ) The value of the Bass model at point of inflexion is
=M  2 (29) (see Figure 6):
p
1 + pq e−(p+q)(t−ts )
(q − p)
Adoption per period (sales) is a bell-shaped curve B(tI ) = M (34)
2q
(see Figure 6), and it is proportional to the first
derivative B'(t) of cumulative adoption:

100 100

% %

75 75

50 50

25 25 p=0.053, q=0.473
p=0.026, q=0.236
p=0.139, q=0.015 p=0.279, q=0.031
0 0
ts-5

ts+5

ts+10

ts+15
ts

ts+20

ts+25
ts-5

ts+5

ts+10

ts+15
ts

ts+20

ts+25

100 100

% %
75 75

50 50

25 25 p=0.008, q=0.747
p=0.004, q=0.374
p=0.298, q=0.003
p=0.149, q=0.002
0 0
ts-5

ts+5

ts+10

ts+15

ts+20

ts+25

ts+5

ts+10

ts+15

ts+20

ts+25
ts

ts-5

ts

Figure 5 Effects of different values of parameters p and q. Chosen values are explained in Section 4.2

Telektronikk 3/4.2008 ISSN 0085-7130 ©Telenor ASA 2009 149

 M M
Mp
(p+q) 2
M 4q
s ≥ 0.5 s < 0.5
(q-p)
M 2q I
0 0

ts+10
tI

ts

ts+20

ts+10
tI

ts+20
ts
(q-p)
M 2q
I
q≤p q>p

B(t) B(t)
dB(t)/dt dB(t)/dt

Figure 6 Characteristic values and points of the Bass growth model

In cases when q < p, the inflexion point and maxi- dependent while shape Bass model S-curve. Namely,
mum of B'(t) occurs before the product starts (tI < ts), the value of a characteristic duration of a product is
and the value of the Bass model at that point is nega- provided only indirectly through the values of p and
tive according to (34), therefore interior maximum of q parameters (see Figure 5).
B'(t) occurs at t = ts . Similarly, in cases when q = p,
inflexion point and maximum of B'(t) occurs when The idea is to replace p and q with two independent
the product starts (tI = ts). For q > p, the sales peak explanatory parameters: a parameter that describes
occurs in conventional sense of a PLC (tI > ts). The the vertical shape of the S-curve s and characteristic
above mentioned is summarized in (35): duration (time to reach certain penetration level mea-
 2 sured from ts) Δt. Expected penetration level at time
 M (p+q) ⇔ q > p, t = tI
max B (t) = 4q (35) point ts + Δ t is v (see explanation for Δt and v in Sec-
M p ⇔ q ≤ p, t = ts
tion 3.2).
Accordingly, a maximum of sales occurs when pene-
tration is (q – p) / 2q in cases when q > p (at t = tI), The shape parameter s is chosen in order to encom-
and in cases when q≤ p, maximum of sales occurs at pass the relation between the amplitude of the posi-
t = ts when penetration is 0, which is summarized in tive S-curve part and the amplitude of the negative
(36): S-curve part. Asymptotes of the Bass model are:

max Sales(t1 , t2 ) ≈ p
 lim B(t) = − M lim B(t) = M
2 t→−∞ q t→+∞
(t2 − t1 ) · M (p+q) ⇔ q > p; t1 and t2 are near tI
≈ 4q
(t2 − t1 ) · M p ⇔ q ≤ p, t1 and t2 are near ts (36) The ratio between negative asymptote and the dis-
tance of these asymptotes lays in range (0,1] which is
The growth rate GR for time interval Δt is always convenient to choose as the shape parameter s, and
positive: which can be measured in percent. In fact, according
to the value of s, the S-curve is stretched in the verti-
B(t) − B(t − Δt)
GRΔt = cal direction (on the y-axis) preserving the total mar-
B(t − Δt)
ket capacity M. [9]
Due to the fact that the Bass model starts from ts,
B(ts) = 0, the growth rate for t –> ts goes to infinity. The distance between these asymptotes is M . (1 + p /
q), so the shape parameter s is:
Above described characteristics of the Bass model
pM/q p
with its explanatory attributes can be used as helpful s= = , p > 0, q ≥ 0 (37)
M + pM/q q+p
input for estimation or assessment of model parame-
ters for forecasting purposes. Characteristic values of s are:

4.2 The Bass Model with Explanatory s –> 0 negative asymptote –> 0, imitation prevails,
Parameters curve is similar to a simple logistic growth
Parameters M and ts are descriptive and can easily be model, (q >> p > 0),
linked with market conditions. Although p as a coef-
ficient of innovation and q as a coefficient of imita- s = 0.5 sales peak occurs at time when product starts
tion have explanatory features, they are mutually (q = p > 0)

150 ISSN 0085-7130 ©Telenor ASA 2009 Telektronikk 3/4.2008

s=1 negative asymptote –> ∞ , innovation pre- Graph in Shape Shape Characteristic
vails; curve is similar to an exponential satu- Figure 5 parameter parameter duration to 95 %
ration growth model (q = 0, p > 0). s1____ s2 ____ penetration Δt

Top-left 10 % 90 % 20 years
From (37) follows:
Top-right 10 % 90 % 10 years

p = (p + q) . s; q = (p + q) . (1 – s) (38) Bottom-left 1% 99 % 20 years

Bottom-right 1% 99 % 10 years
Putting information about penetration level B(ts + Δ t)
= vM in (28) together with (38) give (39) and (40) Table 2
 
1 v
p+q = ln 1 + (39)
Δt s(1 − v)
B(t; M, ts , Δt, s → 0, v) →
B(t; M, ts , Δt, s, v) = M
→  − t−t − sM ≈ L(t; M, a, b)
 − t−ts
1 v Δt
s

v
1 − 1 + s(1−v)
Δt
1+ · s s(1−v)
=M  − t−ts (40) where parameters of logistic growth model a and b
Δt
v
1 + (1/s − 1) · 1 + s(1−v) are:
 
Expression (40) is the reparameterized Bass model 1 v ln s
a= ln , b = ts −
with explanatory parameters (instead of p and q) Δt s(1 − v) a
where: M – market capacity; ts – time when product • For s = 0.5 the Bass model gets a form:
is introduced, B(ts) = 0, ts ≤ t, Δ t – characteristic
 − t−ts
duration of product, Δ t > 0, s – shape parameter, 1− 1+v Δt

1−v
0 < s ≤ 1; and v – penetration at time point ts + Δ t, B(t; M, ts , Δt, s = 0.5, v) = M  − t−ts
=
Δt
1+v
0 ≤ v < 1. The model from (40), B(t; M, ts , Δ t, s, v), 1+ 1−v
needs four parameters; M, ts , Δ t and s to be deter- 2M
=  − t−ts
− M = L(t; 2M, a, b) − M
mined. The value of the auxiliary parameter v does 1+v Δt
1+ 1−v
not need to be determined, it just allows the forecast-
ing practitioner to choose which level of penetration This curve has a shape of logistic model with dou-
he/she wants to deal with (ie. 90 %, 95 %, etc). ble market capacity M but vertically shifted down
by M. Parameters a and b of this ‘halved’ logistic
Special cases of (40): model are:
 
• For v = 0, the value of model B(t) is zero: 1 1+v
a= ln , b = ts
Δt 1−v
B(t; M, ts , Δ t, s, v = 0) = 0 • For s = 1 the Bass model degrades into an exponen-
tial saturation growth model:
• For s –> 0, the Bass model degrades into a simple  
t−ts
logistic model: B(t; M, ts , Δt, s = 1, v) = M 1 − (1 − v) Δt

s = 20 % s = 50 % s = 80 %

1 − 9.80−(t−ts ) 1 − 6.24−(t−ts ) 1 − 4.97−(t−ts )

Δt = 2 years M M M
1 + 4 · 9.80−(t−ts ) 1 + 6.24−(t−ts ) 1 − 0.25 · 4.97−(t−ts )

1 − 2.49−(t−ts ) 1 − 2.08−(t−ts ) 1 − 1.90−(t−ts )

Δt = 5 years M M M
1 + 4 · 2.49−(t−ts ) 1 + 2.08−(t−ts ) 1 − 0.25 · 1.90−(t−ts )
1 − 1.58−(t−ts ) 1 − 1.44−(t−ts ) 1 − 1.38−(t−ts )
Δt = 10 years M M M
1 + 4 · 1.58−(t−ts ) 1 + 1.44−(t−ts ) 1 − 0.25 · 1.38−(t−ts )
1 − 1.36−(t−ts ) 1 − 1.28−(t−ts ) 1 − 1.24−(t−ts )
Δt = 15 years M M M
1 + 4 · 1.36−(t−ts ) 1 + 1.28−(t−ts ) 1 − 0.25 · 1.24−(t−ts )

Table 3 Reparameterized Bass model framework for forecasting new products adoption prior to launch.
s – shape parameter, Δ t – characteristic duration (time to reach penetration level vM measured from ts ),
value for v is chosen for 95 % penetration (v = 95 %)

Telektronikk 3/4.2008 ISSN 0085-7130 ©Telenor ASA 2009 151

 Table 2 gives the explanation of the chosen values els require finding four parameters instead of the
of parameters p and q presented in Figure 5, that are three needed for the logistic model.
selected according to shape parameter and character-
istic duration. There are several different circumstances when and
how to use the logistic and the Bass model, but in the
Similarly to the framework for forecasting of new main there are three cases:
products adoption prior to launch presented in Sec-
tion 3.2, model (40) can be used in cases when little Little or no data is available
or no data is available by comparison with other simi- In cases of product market adoption forecasting prior
lar products histories. to product launch, an insufficient set of historical data
is available. Market capacity M should be estimated
4.3 Bass Model through one Fixed Point by market research and/or market segmentation
Similarly to the concept of the local logistic model techniques. Comparison with other similar product
described in Section 3.3, the Bass model with histories (ie. forecasting by analogy) or judgmental
explanatory parameters which have embedded value assumption is needed for the following parameters:
of one data point (tp , N(tp )) has the following form:
• ts, Δ t in model: L(t; M, ts, Δ t, u); auxiliary para-
LB(t; M, ts , s, tp , N (tp )) = meter u is usually set at 5 % or 10 %
 − tt−ts
N (tp )
1− 1+ s(M −N (tp ))
p−t s
• ts, Δ t, c in model: R(t; M, ts, Δ t, c, u); auxiliary
=M  − tt−ts (41) parameter u is usually set at 5 % or 10 %
N (tp ) −t
p s
1 + (1/s − 1) · 1 + s(M −N (tp ))
• ts, Δ t, s in model: B(t; M, ts, Δ t, s, v); auxiliary
and could be called the local Bass model. By default, parameter v is usually set at 90 % or 95 %.
the local Bass model as well as the Bass model, have
an embedded value of starting point (ts, 0). The local For the forecasting of new product adoption prior to
Bass model is useful for forecasting from the last launch (when no historical data are available), the
known data point t > tp . practical framework for the logistic and the Bass
model are given in Table 1 and 3, respectively.
4.4 Limitations of the Bass Model
The Bass model is the most convenient model for In addition, it is possible to utilize information
market adoption forecasting of new product in sense about peak of sales and/or growth rate as additional
of flexibility vs. number of free parameters needed to explanatory variables (see Section 3.1 for the logistic
be estimated. Estimation of parameter values when model and Section 4.1 for the Bass model). Values
limited data is available can be improved by introduc- of obtained parameters are uncertain (ie. their confi-
ing Bass model with explanatory parameters. Although dence cannot be tested), therefore sensitivity analysis
several generalizations of Bass model expand model of forecasted market adoption depending on change
usage for later phases of PLC, numerous supplemen- of their values is strongly suggested. Usually, values
tary parameters demand a large set of known data for these parameters are assumed in interval: opti-
points, which limits their application for the forecast- mistic – pessimistic.
ing purposes.
Limited data available
A minimal set of historical data is three known data
5 Using the Logistic, Richards points for the logistic model and four known data
and Bass Models for Forecasting points for the Richards and Bass models.
Purposes
The logistic, Richards and Bass models are com- When input data have a high level of uncertainty (due
monly used for forecasting of new product market to errors in measurement, uncorrected seasonal devia-
adoption when interaction with other products can be tion, etc) and/or when observations only at the begin-
neglected. In general, the logistic and Richards mod- ning of PLC are available, sensitivity analysis should
els are not suitable for modelling market adoption be performed. In such cases, to reduce uncertainty of
immediately after a product is introduced due to the obtained results, it is better to estimate market capac-
fact that only for t –> – ∞ , L(t) and R(t) approach 0. ity M by market research and/or market segmentation
In such cases the Bass model is used. The Richards techniques and treat M as a fixed value in models
model is more flexible than the logistic and the Bass [10]. The minimal set of historical data with assumed
model in cases of fitting data with asymmetrical posi- M is two data points for the logistic model and three
tion of inflexion point. The Bass and Richards mod- data points for the Richards and Bass models.

152 ISSN 0085-7130 ©Telenor ASA 2009 Telektronikk 3/4.2008

 Model/equation in text Assumed Number of Fixed point Method / Parameters
known data in model (tp, N(tp)) needed to be estimated by it

Local logistic / (21)* – ≥4 The latest known OLS / M, a

Local logistic / (21)* M ≥3 The latest known OLS / a
Logistic / (6)* – ≥4 – Weighted LS / M, a, b
Logistic / (6)* M ≥3 – Weighted LS / a, b

Local Richards / (26)* – ≥5 The latest known OLS / M, a, c

Local Richards / (26)* M ≥4 The latest known OLS / a, c
Richards / (22)* – ≥5 – Weighted LS / M, a, b, c
Richards / (22)* M ≥4 – Weighted LS / a, b, c

Local Bass / (41)* – ≥5 The latest known OLS / M, ts, s

Local Bass / (41)* M ≥4 The latest known OLS / ts, s
Bass / (28)* – ≥5 – Weighted LS / M, p, q, ts
Bass / (28)* M ≥4 – Weighted LS / p, q, ts

* The numbers in brackets represent references to the equations in the paper.

Table 4 Framework for parameter estimation in cases of a larger set of known data than minimal

In cases when only a minimal set of historical data is References

available, values of model parameters are obtained 1 Armstrong, J S (eds). Principles of Forecasting:
from the solution of system (1). A Handbook for Researchers and Practitioners.
Kluwer Academic Publishers, 2001.
For a larger set of known data than minimal, parame-
ter estimation can be done by the OLS method on 2 Bass, F. A new product growth for model con-
models that treat the latest known data as a fixed sumer durables. Management Science, 15 (5),
point (local logistic, local Richards or local Bass 215-227, 1969.
model) or by weighted least squares method on logis-
tic/Richards/Bass model with higher weights for the 3 Bass, F, Gordon, K, Ferguson, T L, Githens, M L.
most recent data points, which is summarized in DIRECTV: Forecasting Diffusion of a New Tech-
Table 4. nology prior to Product Launch. Interfaces, 31
(3), S82-S93, 2001.
Extensive set of input data
In cases of mature products, an extensive set of his- 4 Meade, N. A Modified Logistic Model Applied to
torical market adoption data is known for the whole Human Populations. Journal of the Royal Statisti-
interval of PLC. The fit of the logistic, Richards or cal Society. Series A (Statistics in Society), 151 (3),
Bass models is usually very strong when the product 491-498, 1988.
is sole on the market and can be measured with corre-
lation coefficient R. Due to the fact that an extensive 5 Meade, N, Islam, T. Modelling and forecasting
set of data has to be known already, this case has low the diffusion of innovation – A 25-year review.
usability for practical forecasting purposes. However, International Journal of Forecasting, 22 (3), 519-
it could be useful for accurate estimation of model 545, 2006.
parameters for the certain product – and could later
be used for forecasting by analogy of a subsequent 6 Modis, T. Conquering Uncertainty: Understand-
product or for penetration forecasting of identical ing Corporate Cycles and Positioning Your Com-
products on comparable markets. pany to Survive the Changing Environment. New
York, Business Week Books (McGraw-Hill), 1998.
Parameter estimation can be done by the OLS method
on logistic/Richards/Bass model. 7 Richards, F J. A flexible growth curve for empiri-
cal use. J. Exp. Bot., 10, 290-300, 1959.

Telektronikk 3/4.2008 ISSN 0085-7130 ©Telenor ASA 2009 153

 8 Sokele, M, Hudek, V. Extensions of logistic 11 Sokele, M. Growth models / Logistic Growth
growth model for the forecasting of product life Model / Bass Model. Dictionary of Quantitative
cycle segments. Advances in Doctoral Research Research Methods in Management. London,
in Management, 1, 77-106, 2006. World Scien- SAGE Publications. (Forthcoming)
tific Publishing
12 Stordahl, K. Long-term telecommunication fore-
9 Sokele, M. Incorporating Market and Competitive casting. Norwegian University of Science and
Analysis Insights into your Forecast Models to Technology, Trondheim, 2006. (PhD thesis)
Improve Accuracy. Proc. of the Telecoms Market
Forecasting Conference, IIR, London, 2007.

10 Sokele, M. Uncertainty of forecasted new service

market capacity obtained by logistic model. Proc.
of the 28th International Symposium on Forecast-
ing, Nice, 2008.

Mladen Sokele is Assistant Director of the Operative Planning and Project Management Department in Croatian Telecom Inc.,
Zagreb, Croatia. He graduated in 1983 and received his Master’s degree in 1987 at the Faculty of Electrical Engineering and
Computing of the University of Zagreb. After graduation he joined Ericsson Nikola Tesla Company and from 1989 he was
lecturer at the Faculty of Electrical Engineering in Osijek. He has been with Croatian Telecom Inc. since 1992. Mr. Sokele is
author or co-author of more than fifty published scientific and professional papers in the field of telecommunications and
related indicators on modelling and forecasting. He is also author of the LOgistic Spline Trend (LOST) method and program
tool for the prediction of service penetration over time.

[email protected]

154 ISSN 0085-7130 ©Telenor ASA 2009 Telektronikk 3/4.2008

View publication stats