ArticlePDF Available

Evaluating the Impact of a New Product on the Sales of other Products

Authors:

Abstract and Figures

The purpose of this article is to evaluate the impact of a new product on the sales of other products. Launching a new product may lead to increase or decrease of sales in the products of the same group. Customers may continue buying standard products or they may be oriented to new products. New products may cause internal competition. Statistical methods are applied. Time series analysis is used. Transactional database of a confectionary factory is the main source of data. The time series analysis is carried in two datasets – quantities of sales of three products on a daily basis and on a monthly basis. The main methods used are time series analysis and regression analysis. The three time series (corresponding to the three products) are separated into two parts – before and after launching the new product. It is proved that the new product does not affect the sales of the two other products. The new product is well accepted and its sales increase together with the sales of the two other products. Introduction Sales analysis consists of a lot of techniques. Many methodologies have been developed to forecast sales. Building a model
Content may be subject to copyright.
Journal of Economics and Business Research,
ISSN: 2068 - 3537, E ISSN (online) 2069 9476, ISSN L = 2068 3537
Year XX, No. 2, 2014, pp. 7-20
Evaluating the Impact of a New Product on the Sales of
other Products
J. A. Vasilev
Julian Andreev Vasilev
Faculty of Informatics
Varna University of Economics, Bulgaria
Abstract
The purpose of this article is to evaluate the impact of a
new product on the sales of other products. Launching a
new product may lead to increase or decrease of sales in
the products of the same group. Customers may continue
buying standard products or they may be oriented to new
products. New products may cause internal competition.
Statistical methods are applied. Time series analysis is
used. Transactional database of a confectionary factory is
the main source of data. The time series analysis is
carried in two datasets quantities of sales of three
products on a daily basis and on a monthly basis. The
main methods used are time series analysis and
regression analysis. The three time series (corresponding
to the three products) are separated into two parts
before and after launching the new product. It is proved
that the new product does not affect the sales of the two
other products. The new product is well accepted and its
sales increase together with the sales of the two other
products.
Keywords: launching a new product, sales analysis,
forecasting, time series analysis, SPSS, transactional
sales database, data transformations, SQL
Introduction
Sales analysis consists of a lot of techniques. Many
methodologies have been developed to forecast sales. Building a model
J. A. Vasilev
8
with several independent variables and one independent variable
(usually the quantity of future sales) gives a vague idea of what happens
in real life. Usually regression analysis is applied to find a trend the
dependency of sales on time. It should be marked that all models have
an error. The error shows that other factors influence sales, but they are
not included in the model.
Launching a new product is a creative technique. New products
are invented because the life of people changes. Some people die, other
people are born, other people start working. The change of generations
leads to an inevitable change in the portfolio of products. A new product
may be well accepted by the market or it may be rejected. Sometimes
the logistics curve (or logistics function) is used by marketing
specialists to describe the lifecycle of a new product. Since the market is
very sensitive to all changes there may be a slight difference between
textbooks and business.
Forecasting of sales is widely discussed and described. The
launch of a new product starts generating transactional data. Since the
product is new, the transactional database does not contain enough
historical data to predict sales. Moreover the real trend may not be
noticed graphically. Seasonal decomposition could not be done.
Literature review
Conventional retail markets are known to be quite different from
electronic markets. Ahluwalia et al. (2013) examine the effect on
pricing in relation to proximity to a culturally and socially significant
peak shopping day [7]. They study the effects of consumer’s product
rating, product popularity, and featured product website rankings. This
study is based on a real case study. Data are extracted from a B2C
retailer. Some articles (Brent at al., 2014, Singh and Das, 2013) focus
on salesperson turnover [2], [8]. Their study uses database collected
from ten different firms. Discriminant analysis is used.
Sales analysis is often connected with analysis of seasonal
variations. Andy et al. (2013) try to identify seasonal variations in store-
level visitor grocery demand [1]. Store trading information is used.
Spatial analysis is applied. The authors prove a significant degree of
seasonality in terms of their revenue.
New product development is widely studied. Lofsten (2014)
studies the product innovation processes. He finds determinants of
Evaluating the impact of a new product on the sales of other products
9
product innovation processes [4]. The author tests the relationship
between innovation performance and business performance (sales and
profitability).
Some authors (Rojas-Mendez and Rod, 2013) use face-to-face
survey questionnaires to analyze sales of wine producers [5]. They
compare two different instruments for assessing market orientation.
Food supply chains and small producers in confectionary factories are
studied by Ogelthorpe and Heron (2013). They try to identify the
observable operational and supply chain barriers and constraints that
occur in local food supply chains [3].
New products usually affect sales of other products. Mokhtar
(2013) investigates the relationship between customer focus and new
product performance [9]. Data are collected using mail questionnaire
survey approach. The results revealed that customer focus has a
statistically significant association with new product performance.
Graner and Missler-Behr (2013) try to find key determinants of the
successful adoption of new product development methods [6]. Their
article adopts a structural equation modeling approach and analyzes the
subject based on a large empirical sample of 410 product development
projects.
Data preparation
For the sake of the analysis of the impact of a new product in a
confectionary factory, transactional data for sales are used. The period is
January 2011 until May 2013. The new product is croissant with
chocolate, launched on 30 March 2012. The two other products are
croissant with Turkish delight and croissant with marmalade (launched
on 14 January 2011). We do not use questionnaires. We use
transactional data to monitor sales. The transactional database consists
of 332 000 records. The extraction of data starts with Query 1. The
result of its execution is given in Table 1.
Table no.1. Extracting the first date of launching products
Code_item
Name_item
First Ofdate
189
Croissant with Turkish delight
14 January 2011
190
Croissant with chocolate
30 March 2012
191
Croissant with marmalade
14 January 2011
J. A. Vasilev
10
The SQL code for getting the initial information is the
following:
SELECT Sales_transactions.code_item, Items.name_item,
First(Sales_transactions.date_doc) AS FirstOfdate_doc
FROM Items INNER JOIN Sales_transactions ON Items.code_item =
Sales_transactions.code_item
GROUP BY Sales_transactions.code_item, Items.name_item
HAVING (((Sales_transactions.code_item)=189)) OR
(((Sales_transactions.code_item)=190)) OR
(((Sales_transactions.code_item)=191))
ORDER BY Sales_transactions.code_item,
First(Sales_transactions.date_doc);
Now we have three time series for the sales of three products.
We have to monitor sales of items 189 and 191 before and after 30
March 2012. The new product is 190. Since the data in the database
consists of transactional data, we have to extract data by using select
queries. We use only sold quantities. We do not use values of sales.
Table no.2. A part of the result dataset of “Total sales by dates for
article number
Total quantity
19
27
29
20
32
18
The SQL code is the following:
SELECT Sales_transactions.[date_doc],
Sum(Sales_transactions.[quantity]) AS SumOfquantity
FROM Sales_transactions
WHERE (((Sales_transactions.[code_item])=189))
GROUP BY Sales_transactions.[date_doc];
Evaluating the impact of a new product on the sales of other products
11
As it is obvious, the last SQL query gives a two dimensional
dataset for sales. For those dates with zero sales no rows are added in
the resulting dataset. So, some transformations in MS Excel may be
made to fix this problem. By changing the string “189 to “190” and
“191” we make two more select queries to extract time series for sales
for items 190 and 191. But the problem with zero sales on some days
exists. The result of the three queries is exported to MS Excel on a
single worksheet. In a separate column all dates from 14 Jan 2011 until
5 May 2013 are filled automatically (844 days). By using the
VLOOKUP function three new time series are created. They continue
zero and non-zero values for sold quantities, for each date
We use three named ranges (SKU_189, SKU_190 and
SKU_191). The first and the third data range consist of 819 rows and
the second one, 396 rows. The formula for extracting zero and non-zero
quantities is the following:
=IF(ISERROR(VLOOKUP(A3;SKU_189;2;FALSE));0;VLOOK
UP(A3;SKU_189;2;FALSE))
Now the three new data series consist of 844 values one value
for each day. Each data series represents one stock keeping unit (SKU)
Table no. 3. A part of the three new data series
dates
SKU
189
SKU
190
SKU
191
14.1.2011
19
0
14
15.1.2011
0
0
0
16.1.2011
27
0
14
17.1.2011
29
0
25
18.1.2011
20
0
19
19.1.2011
32
0
40
20.1.2011
0
0
0
21.1.2011
18
0
28
J. A. Vasilev
12
Analyzing daily datasets in SPSS
The three data series contain numeric values. They may be
exported from MS Excel and imported into SPSS [10]. Three variables
are defined in SPSS SKU_189, SKU_190 and SKU_191. All of them
are on an interval scale. A string variable with dates is added in SPSS. It
is recommended to use the natural logarithm function (ln) and seasonal
decomposition before analyzing data series with time series analysis and
building regression models. Our assumption is that a new product
affects the sales of common products. The effect may be positive, the
sales of two other products may increase or decrease or there may be no
influence.
We have enough data to make the analysis. The period from
14.1.2011 until 6.5.2013 may be divided into two sub periods before
30.3.2012 (the lunch of the new product) and after this date. We may
use also case number (case numbers 1-441 are before launching the new
product) to filter data series. Since we have a time series dataset, we
have to define dates (Data/Define dates).
Three line charts may be made (before using the logarithmic
function and seasonal decomposition) to see whether there is a trend in
each time series.
Simple charts and values of individual cases are presented below.
Fig. no. 1. Line chart of SKU_189
Evaluating the impact of a new product on the sales of other products
13
It is obvious that there is a trend and sales of SKU 189 increase
after case number 441.
Fig. no. 2. Line chart of SKU_191
It is obvious that there is a trend in SKU 191. Again the sales
start to increase after case number 441. The graphs of SKU 181 and
SKU 191 are almost the same.
Fig. no. 3. Line chart of SKU_190 (the new product)
The left part of the graph represents the period when the product
was not launched. There is a trend for increasing of sales. It is obvious.
J. A. Vasilev
14
As a new product its sales are similar to the logistic function
representing the launch of a new product.
The three time series seem to be correlated. It has to be checked
for correlation. The check should be made for the two sub periods. First,
we start with the first sub period (Data/Select Cases). We use selection
of cases based on a satisfied condition (DAY_<=441). Then the
existence of correlation is checked (Analyze/Correlate/Bivariate). Since
the values in the three time series are in an interval scale, the Pearson
correlation coefficient may be used. Two-tailed test of significance is
made. The correlation between SKU 189 and SKU 191 is 0.915. The
correlation is significant at 0.01 level (2-tailed).
Curve estimation may be made to check if the quantity of sales
(of SKU 189 and SKU 191) depends on time
(Analyze/Regression/Curve estimation). ANOVA tables are displayed.
The constant is included in the equation. SPSS says that 17 cells have
zero values so the compound, power, S, growth, exponential and logistic
models cannot be calculated. All calculated models have R-square value
less than 0.2. It means that it is almost impossible to create an equation
which best represents the quantity of sales based on time for the first
period. The check for correlation between the three time series has to be
made for the second sub period (cases > 441). Again a check for
correlation is made. The correlation between SKU 189 and SKU 191 is
0.965 even a higher value than the first sub period. The correlation
between SKU 190 and SKU 189 is 0.949. The correlation between SKU
190 and SKU 191 is 0.974. The correlation coefficients are significant
at 0.01 level (2-tailed). It means that when sold quantities of one of the
three products increase, there is a great possibility the sales of the two
other products two increase. Again curve estimation may be made for
the three time series for the second sub period. ANOVA tables are
displayed. The constant is included in the equation. The highest R-
square value is for the cubic function (0.331) but it is not very high to
make a meaningful equation. By excluding the constant in equation the
highest value of R-square has the cubic function (0.812).
Regression analysis is a common technique for analyzing
economic phenomena. Before applying regression analysis the trend has
to be removed and seasonal decomposition has to be made. All cases
have been selected. A natural logarithm of the three data series is
calculated (Transform/Compute variables). Three new data series are
Evaluating the impact of a new product on the sales of other products
15
automatically calculated lnSKU_189, lnSKU_190 and lnSKU_191.
Seasonal decomposition is made (Analyze/Forecasting/Seasonal
Decomposition). Since we have very detailed data (for daily sales)
SPSS says “Seasonal decomposition requires at least one periodic data
component to be defined”. If we aggregate data on a monthly basis we
will not get this error but we will lose a lot of detailed information. The
dispersion in the three initial series is stabilized by calculating the
natural logarithm.
A check for autocorrelation has to be made for each time series
(Analyze/Forecasting/Autocorrelations). For the three time series
(lnSKU_189, lnSKU_190 and lnSKU_191) the partial autocorrelation
coefficients (ACF) are outside the upper and the lower confidence limit.
It means that there is autocorrelation in the three time series. We have to
create a time series by using the first order of the difference function for
the three time series lnSKU_189, lnSKU_190 and lnSKU_191
(Transform/Create time series).
Three new time series are created: lnSKU__189, lnSKU__190
and lnSKU__191. Again a check for autocorrelation is made. The
partial ACF are again outside the lower and upper confidence limit. So
the conclusion is that a mathematical formula for calculating future
sales cannot be made. Moreover linear regression for SKU_189 and
SKU_191 cannot be done neither for the first, nor for the second sub
period. Thus a calculation for the influence of the new product 190
cannot be calculated. Even though sales of the three products are
increasing during the second sub period it may not be proved that the
increase of sales is caused by launching a new product.
It may be argued that when sales of SKU 189 decrease, we may
expect a decrease of sales in SKU 190 (the new product). Since there is
a great correlation between these two products the sold quantities may
depend on other factors such as season or location. This dependency
may be showed graphically.
J. A. Vasilev
16
Fig. no. 4. Dependency between lnSKU_189 and lnSKU_190
By using curve estimation the linear model may be checked. The
independent variable is lnSKU_189 and the dependent variable is
lnSKU_190. The R-square value is 0.814. The ANOVA test shows that
the model is adequate. The unstandardized coefficient B is 0.903. It is
statistically significant. The constant in the equation is 0.274. It is
statistically not significant. So the conclusion is that we may do the
curve estimation by excluding the constant in equation. Now the R-
square value is 0.993. The ANOVA test shows that the model is
adequate. The unstandardized coefficient B is 0.957. It is statistically
significant. The equation is the following:
Ln( Y ) = 0.957 * Ln( X )
Y is the predicted quantity of sales of SKU 190 and X is the
quantity of sales of SKU 189. This formula shows that a small change in
the sales of SKU 189 affects SKU 190. The coefficient 0.957 is near 1
but less than one. So the items sold by SKU 189 are almost the same as
the items sold by SKU 190. The coefficient is below 1 because there is a
lag effect or other factors affect sales of the new product.
Evaluating the impact of a new product on the sales of other products
17
Table no. 4. Predicted quantities for SKU 190 (variable Y) on the basis
of sold quantities for SKU 189 (variable X)
X
ln X
ln Y
Y
Y/X
2
0.693147
0.663342
2
1.0000
4
1.386294
1.326684
4
1.0000
6
1.791759
1.714714
6
1.0000
8
2.079442
1.990026
7
0.8750
10
2.302585
2.203574
9
0.9000
12
2.484907
2.378056
11
0.9167
14
2.639057
2.525578
12
0.8571
16
2.772589
2.653367
14
0.8750
18
2.890372
2.766086
16
0.8889
20
2.995732
2.866916
18
0.9000
The top left corner of the table is situated in C2. The formula for
calculating ln X is: “= ln (C3)”. The formula for ln Y is “=D3*0.957”.
The formula in Y column is “=ROUND(EXP(1)^E3;0)”. The last
column “Y/X” is “=F3/C3”. The last column calculates the slope of the
line.
Analyzing monthly datasets in SPSS
Using daily datasets we proved that we could not make
regression analysis. By grouping the initial three datasets on a monthly
basis, similar calculations may be made. Our prediction is the same. The
trend line of sales of SKU 189 changes when launching the new product
190. We will skip the dataset for SKU 191, because we calculated a
great correlation between SKU 189 and SKU 191.
Aggregating time series may be done in MS Excel or directly
with a SQL query. We will use the first approach. Firstly, we copy the
first three columns from table 3 on a separate worksheet. By using the
built-in functions “month” and “year”, the month and year of each date
is extracted. Data are grouped by a Pivot table. “Year” and “month” are
used for row labels. “Sum of SKU 189” and “Sum of SKU 190” are
used as summary values.
Now the table in MS Excel consists of 29 rows starting from
January 2011 and ending in May 2013 one row for each month. Now
J. A. Vasilev
18
we use the total sold quantity of each product for each month. We lost a
lot of data but we may make seasonal decomposition and try to make
regression analysis for the two periods for SKU 189.
In SPSS we define four variables on an interval scale year,
month, SKU_189 and SKU_190. Dates are defined. Cases are years and
months starting from year: 2011 and month: 1. Natural logarithm of
both time series is calculated. Seasonal decomposition has to be done.
The additive model is used. Seasonal decomposition cannot be done
because of missing data for SKU 190 it is a new product. Seasonal
decomposition cannot be done for SKU 189 because at least four full
seasons of data must exist.
Now three linear models for SKU 189 have to be tested for the
whole period, before and after launching the new product. The models
may be only linear, because we used the logarithmic function.
The first model (for the whole period) including the constant in
equation is checked. The R-square value is 0.703. The ANOVA test
shows that the model is adequate. The unstandardized coefficient B
value is 0.069, the constant is 6.775. Both coefficients are statistically
significant. By excluding the constant in equation the R-square value is
0.821. The ANOVA test shows that the model is adequate. The
unstandardized coefficient B value is 0.414. It is statistically significant.
For the whole period we have increasing sales. The slope of the line is
0.414. The second model (before launching the new product) including
the constant in equation is checked. The R-square value is 0.542. It is
comparatively low. The model and the calculated coefficients are
significant at Alfa 0.05. Not including the constant in equation the R-
square value is 0.813. The model is adequate. The slope of the line is
0.771. The third model (after launching the product) with constant in
equation has a very low value of R-square 0.126. Excluding the constant
in equation the R-square value is 0.787. The unstandardized coefficient
B (the slope of the line showing the natural logarithm of sales) is 0.813.
It is statistically significant.
The local conclusion is the following. Before launching the new
product the slope of the line of sales (using the natural logarithm of sold
quantities) is 0.771. After launching the product, the slope is 0.813.
There is a change in the slope, but we are not sure whether the change is
caused by the launching of the new product.
Evaluating the impact of a new product on the sales of other products
19
Conclusions
Confectionary factories have a lot of data stored in a
transactional database. Transactional databases are mainly used to
record sales and organize the logistics activity. In rear cases they may
be used for data mining. Launching a new product may cause serious
fluctuations in the sales of other products from the same group. In this
study three products are monitored. Two of them are sold 844 days. The
third one goes out to market on the 441 day. The initial prediction is that
the new product affects the sales of the two other products. It is
calculated that there is a strong correlation between the three products.
As a whole (during the whole period) the sales of the three products
increase. The period of 844 days is divided into two parts before
launching the new product and after launching the product. The time
series of one of the products is analyzed, because there is a strong
correlation between the sales of the three products. Three linear models
are estimated for the whole period, before launching the new product
and after launching it. It is proved that the new product does not affect
the sales of the two other products. Time series analysis is used.
Regression analysis is not applied at its final stage, because of a strong
autocorrelation in the three time series. Autocorrelation cannot be
removed by standard statistical methods. Future research may focus on
extended models where other independent variables are included.
References
[1] Newing, A., Clarke, G., Clarke, M. (2013). Identifying seasonal
variations in store-level visitor grocery demand, International
Journal of Retail & Distribution Management, Vol. 41 Issue: 6,
p. 477 492.
[2] Wren, B. M., Berkowitz, D., Grant, E. S. (2014). Attitudinal,
personal and job-related predictors of salesperson turnover,
Marketing Intelligence & Planning, Vol. 32 Issue: 1, p.107
123.
[3] Oglethorpe, D., Heron, G. (2013). Testing the theory of constraints
in UK local food supply chains, International Journal of
Operations & Production Management, Vol. 33 Issue: 10,
p.1346 1367.
J. A. Vasilev
20
[4] Löfsten, H. (2014). Product innovation processes and the trade-off
between product innovation performance and business
performance, European Journal of Innovation Management,
Vol. 17 Issue: 1, p. 61 84.
[5] Rojas-Méndez, J. I., Rod, M. (2013). Chilean wine producer
market orientation: comparing MKTOR versus MARKOR,
International Journal of Wine Business Research, Vol. 25
Issue: 1, p. 27 49
[6] Graner, M., Mißler-Behr, M. (2013). Key determinants of the
successful adoption of new product development methods,
European Journal of Innovation Management, Vol. 16 Issue:
3, p. 301 316
[7] Punit, A., Hughes, J., Midha, V. (2013). Drivers of e-retailer peak
sales period price behavior: an empirical analysis, International
Journal of Accounting and Information Management, Vol. 21
Issue: 1, p. 72 90.
[8] Singh, R., Das, G. (2013). The impact of job satisfaction, adaptive
selling behaviors and customer orientation on salesperson's
performance: exploring the moderating role of selling
experience, Journal of Business & Industrial Marketing, Vol. 28
Issue: 7, p. 554 564.
[9] Mohd Mokhtar, S. S. (2013). The effects of customer focus on new
product performance, Business Strategy Series, Vol. 14 Issue:
2/3, p. 67 71.
[10] Hadzhiev, V. et.al. (2009). Statistical and econometric software.
Varna: Nauka i ikonomika (in Bulgarian).
... A free research area is Green IT and IoT in supply chains. The increase of sales in e-commerce forces businessman to use new methods of fast analyzing big datasets [25]. Moreover the increase of sales in e-commerce leads to a growing forward material flow in logistics within supply chains. ...
Article
Full-text available
Nowadays ecology aspects are very crucial. They apply to all areas related to human life. The development of ICT (Information Communication Technologies) means that ecological issues are becoming increasingly important. Thus Green Computing or sustainability is not only the operational energy consumption of computing equipment but also it must take the product life cycle into consideration, from production to operation to recycling. In this study the authors are taking into consideration the role of Green Computing and Green Internet Of Things. The major challenge in Green Computing and Green Internet Of Things is the recyclability of the hardware, software, computer items such as cartridges and toners. Moreover, the save of energy is important as well. This study is interdisciplinary. The knowledge about reverse logistics of computer hardware needs specific management. That is why the knowledge management of green computing is a topic that is not widely discussed, but it is very important because it has a strong impact on the existence of Earth.
Article
Full-text available
Purpose The purpose of this paper is to assess the degree of market orientation of a sample of Chilean wine producers; to compare two different instruments for assessing market orientation in this context; and to comment on the possible cultural sensitivities of these two measurement instruments developed from a North American context but applied in culturally dissimilar contexts. Design/methodology/approach In total, 69 CEO and Marketing Managers, representing approximately one quarter of the total number of wineries in Chile, completed a face‐to‐face survey questionnaire that utilized both the Narver and Slater MKTOR and the Kohli and Jaworski MARKOR market orientation scales. SmartPLS was used to carry out the measurement and structural analysis. Findings Results reveal that more than half of surveyed Chilean wine producers are market oriented, with 65 per cent congruence between the two scales. Cluster analysis also reveals three distinct segments and sets of characteristics that distinguish market oriented from non‐market oriented wineries. MKTOR and MARKOR scales show similar level of predictive power when using subjective or perceptual measures of performance as dependent variables. However, the MARKOR scale is found to be better in explaining changes in the dependent variable when the latter is measured by actual sales and gross margins (objective performance). National cultural dimensions (power distance and uncertainty avoidance) have an impact within organizations in the implementation of a market‐oriented strategy in a consistent and coordinated manner. Research limitations/implications The MARKOR scale appears to have superior predictive validity and to be more practical for measuring market orientation since it explains the change in dependent variables to greater degree when performance is measured with objective as opposed to the perceptual measures. Practical implications Chilean winery managers should devote significant attention to market sensing activities and competitive intelligence gathering. The competitive and national cultural environment plays an important role in moderating the relationship between market orientation and a firm's business performance. They may also wish to consider becoming involved in various trade organisations, as well as collaborative partnerships with academic institutions, to enhance their competitive intelligence and technological competences. Originality/value This paper is the first to illustrate the market orientation of Chilean wine producers, and one of only a few to discuss the impact of national cultural values on market orientation.
Article
Full-text available
Purpose In recent years, scholars have devoted increasing attention to the use of methods in new product development. Although their positive impact on product success has been confirmed by a series of studies, little research has so far been conducted into the determinants of the successful method application. The purpose of this paper is to analyze two key determinants for the successful use of methods. Design/methodology/approach The paper adopts a structural equation modeling approach and analyzes the subject based on a large empirical sample of 410 product development projects. Findings The way the new product development process is formalized and the extent of top management support have a significant influence on the successful application of methods. Both factors lead to the use of more methods overall and drive the more thorough and intensive deployment of the specific methods used. Practical implications The paper shows how firms that are seeking a more methodical approach to new product development can consciously create the right determinants for the successful adoption of methods and by doing that can increase the success rate of their new product development activities. Originality/value This research adds to the literature on success factors of new product development in two distinct ways. First, the paper investigates the subject at the level of specific product development projects, based on a large empirical sample and using a structural equation modeling approach, for the first time. Second, the methods investigated were chosen in a systematic, multi step process in which the methods used by several functions involved in new product development were taken into account.
Article
Purpose – To contribute to the understanding of how to manage turnover, the purpose of this paper is to determine if sales managers have the ability to predict high levels of propensity to leave (PL) from variables readily available in personnel records, and on commonly used employee surveys. Design/methodology/approach – The data used for the analysis of the study variables were collected from the sales forces of a total of ten firms across a variety of consumer and industrial product categories, resulting in a sample of 604 respondents. Data were analyzed via multiple discriminant analysis. Findings – The analysis and test results demonstrate that discriminant sets of attitudinal variables, personal characteristics, and aspects of the job can be identified and used to establish meaningful classifications of a salesperson's PL. Organizational commitment, satisfaction with pay, family status, job involvement, level of education, and compensation plan were all found to be significant. Analysis fails to support the existence of several attitudinal variables generally thought to be predictors of PL. Originality/value – The overarching implication to be drawn is that any effort to address salesperson turnover must be holistic, rather than limited to a narrow set of variables. These findings hold implications for sales management researchers and human resource/personnel managers.
Article
Purpose The purpose of this paper is to investigate the moderating effects of selling experience on the relationship between job satisfaction and sales performance, customer orientation and sales performance, and adaptive selling behaviors and sales performance, taking the context of B2B insurance selling. Design/methodology/approach Using a sample of 380 business‐to‐business insurance salespersons from an emerging market (India) to validate their model, the authors tested several hypotheses using structural equation modeling (SEM). Findings The results suggest that experience works with customer‐oriented selling in making the more experienced salespersons better performers. It was also found that for less experienced salespersons, the impact of job satisfaction on performance is weaker than for more experienced salespersons. In addition, it was found that more experienced salespersons' performance is better explained using job satisfaction and customer‐oriented selling rather than their adaptive selling behaviors. Research limitations/implications The study contributes by explaining the mechanism for the above relationships. The study also contributes to knowledge by showing that more experience may not be always good for sales performance. Since the sample comes from an emerging market, the paper extends the knowledge from developed markets, and by testing in emerging markets. Practical implications The managerial implications of this study lie in explaining those situations where experience can make salespersons more productive. The current sales literature on B2B selling contexts falls short of explaining this mechanism in salesperson performance. Originality/value This study contributes to knowledge uniquely by extending the body of empirical evidence that suggests that for experience, more is not always better. The study also shows that a more experienced salesperson does not improve his/her performance by adopting adaptive selling strategies. Such adaptive selling strategies are probably more suitable for younger salespersons, given different expectations from them by customers. For experienced salespersons, job satisfaction and customer‐oriented selling are more important than adaptive selling. This study explains the mechanism for the above relationships.
Purpose ‐ The purpose of this paper is to understand the contribution of visitor demand to the seasonal sales variations experienced at grocery retailers in Cornwall, South West England. Design/methodology/approach ‐ Working collaboratively with a major UK retailer provides access to store trading information and customer data from a popular loyalty card scheme. The authors use spatial analysis to identify revenue originating from outside the store catchment, and explore the spatial and temporal nature of the visitor demand recorded in-store. Findings ‐ The paper demonstrates the significant degree of seasonality experienced around stores in terms of their revenue generated from out-of-catchment visitors, and highlights implications for store location planning. Most notably, visitor expenditure tends to demonstrate far more spatial and temporal clustering than residential demand. The authors argue that it is essential for retailers to ensure that their location planning makes full use of all available consumer data to understand the local nature of demand, including the impact of visitor expenditure. Research limitations/implications ‐ The authors aim to use this insight to develop a spatial decision support system (SDSS) for use within site location planning in the retail sector. This would incorporate a spatial interaction model to estimate and account for variation in local demand generated by seasonal tourist visits. Originality/value ‐ Customer level loyalty card data are rarely available for academic investigations and the authors are able to provide a unique insight into customer expenditure in tourist locations. There has been little exploration of seasonal tourist demand in store location planning, and this study addresses an identified academic and commercial need.
Article
Purpose – This paper is concerned with the management and organization of product innovation processes, and how innovation performance relates to business performance. The underlying rationale is that encouraging firms to innovate will lead to a better business performance. Design/methodology/approach – This study leverages a data set of 99 medium-sized technology firms in Sweden. The first part of the analysis in this study aims at finding determinants of product innovation processes, and the second part is the analysis and trade-off between innovation performance and business performance. First, a research framework is developed in which the link between strategic dimensions, process dimensions and organizational dimensions of product innovation activity and product innovation performance is tested. Second, the research framework tests the relationship between innovation performance and business performance (sales and profitability). Findings – Product innovation performance (patent) is affected by seven variables of the 14 variables that represent product innovation processes. Product innovation performance is not affected by firm size, firm age, branch and product life cycles and, in the regression model, all three innovation performance variables (patents, copyrights and licenses) have a positive effect on the firm's sales, but there were no connections to the firm's profitability. Originality/value – The main implication of the study is the idea supporting a multi-aspects approach to the product innovation processes and performance since product innovation process dimensions (variables used in the study) have only partial influence on innovation-/business performance.
Article
Purpose The purpose of this paper is to investigate the relationship between customer focus and new product performance. Design/methodology/approach Senior management of Malaysian manufacturing firms was the target respondent of this study. Data were collected using mail questionnaire survey approach. Findings The results revealed that customer focus has a statistically significant association with new product performance; hence, the proposed hypothesis of the study is supported. Practical implications The outcome of this study provides vital information from Malaysian firms' perspective on the practical relationship of customer focus on new product performance. Originality/value This paper offers a better understanding for senior management of manufacturing firms to engage with customers in developing a new product.
Purpose Electronic markets are known to be distinct from, and more efficient than, the conventional retail markets. The purpose of this paper is to examine the effect on pricing in relation to proximity to a culturally and socially significant peak shopping day and the moderating effects of consumers product rating, product popularity, and featured product website rankings. Design/methodology/approach The shopping season during the Thanksgiving holiday in 2010 was used to collect data for this study. This paper uses a case study approach by extracting real store‐level data from the web pages of a B2C e‐retailer. Store level data were downloaded for a total of 19 days, before and after “Black Friday.” The longitudinal data were analyzed using regression analytic procedure to conform the hypotheses. Findings The longitudinal data supported the hypothesized relationship between days to the culturally significant shopping event and e‐retailer selling price. The data also confirmed that featured product ranking is a significant moderator of the above relationship. Research limitations/implications In e‐retailer websites, webpage ranking determines the order of display of products. Literature suggests that buyers choices are influenced by the volume and order of display of information. Therefore, this study includes webpage rankings of featured products, number of consumer reviews, and consumer ratings as independent variables. Another limitation of this study is that it uses data of one large e‐retailer. Future studies may address these limitations. Originality/value This paper examines the pricing behavior of e‐commerce companies during “culturally and socially significant” events. and answers research questions related to the electronic markets: Do e‐commerce companies participate in cultural and social events? How do these companies manipulate pricing during a special shopping season? How are search tools employed to showcase specific products to the buyers? Is there a relationship between proximity to “Black Friday” and product price, product popularity, and product ratings?
Article
Purpose – The purpose of the research is to identify the observable operational and supply chain barriers and constraints that occur in local food supply chains, especially with smaller producers, as they seek to increase market penetration across a wider geographic area. Design/methodology/approach – The research adopts a multiple case study approach using mixed methods of data collection where case study companies are interviewed and complete a questionnaire. This process allows us to create a supply chain map and create a narrative which records the burdens and impacts occurring in local food supply chains and smaller producers looking beyond local markets. This evidence is then set against the theory of constraints (TOC) to provide a theoretical underpinning and to examine consistency of the findings. Findings – Seven broad categories of constraint type are observed: constraints due to the nature of the market; due to scale and the nature of products; constraints related to employment and skills; institutional constraints; constraints in supply chain relationships; certification, policy and regulatory constraints; and constraints around personal beliefs and anthropomorphism. Each is described as to its origin, its limitation to business and where possible, how it might be remedied. The constraints point to some counter-intuitive results as far as common perceptions of local food are concerned but suggestions for improvement are made through collaborative producer efforts, alternative institutional intervention, supply chain re-engineering and logistics innovation. Practical implications – Practical suggestions are made to improve the inclusiveness of distribution networks, to better utilise regional food groups (RFGs), to develop opportunities to set up autonomous supply chain centres, or to broaden the function of farmer co-operatives. The paper also provides an alternative model of the TOC specifically adapted for local food producers, the focus of which plays to their strengths and focuses on building competencies across, up and down the supply chain. Originality/value – The adaptation of the TOC provides an advancement of knowledge in the area of food supply chain analysis and is done in a way that is more practical in use. The paper also provides the opportunity to take a similar approach to examining other niche supply opportunities in the sector, which may be dependent on other geographically defined barriers, such as seasonal or ethnic products.
Statistical and econometric software. Varna: Nauka i ikonomika
  • V Hadzhiev
Hadzhiev, V. et.al. (2009). Statistical and econometric software. Varna: Nauka i ikonomika (in Bulgarian).