Packaged information systems and Enterprise Resource Planning (ERP) systems
integrate information and processes of organizations. The acquisition of information
system and software can be motivated by the need for a new information technology
or changing the existing one, both may be aiming to reduce costs, support current
way of doing business (Hecht, 1997), or to facilitate
change in the ways of doing business and thus, to improve effectiveness or to
gain strategic advantage (Silk, 1990; Fitzgerald,
1998). Large investments of money, time and expertise are required to configure
and customize these generic software packages to the specific needs of organizations,
sectors or countries (Davenport, 1998; Klaus
et al., 2000). Another important problem of customization is the
incompatibility of the customized version with the organizations functions
and systems (Butler, 1999; Kremers
and Dissel, 2000; Sumner, 2000; Light
et al., 2001). Some organizations choose to adapt the generic software
without customization and change their processes accordingly (Klaus
et al., 2000; Kumar and Hillegersberg, 2000;
Bernroider and Koch, 2001; Everdingen
et al., 2000), while others try to select the best fit software with
minimum cost and minimum amount of customization (Light
et al., 2001; Davison, 2002; Soh
et al., 2000).
The main intend of software selection and/or evaluation is to identify the
best alternative for the organizations requirements. Most enterprises
have critical processes that uniquely define the power that makes an organization
successful. These critical processes should become critical requirements during
a software selection process. If the software vendor fails to meet these critical
requirements, they become fatal flaws that will adversely affect the implementation
(O Connor, 2008).
Traditional software selection relies on a long list of functional requirements
(sometimes in the range of thousands) that identify the vendors with the best
functional fit (Verville and Halingten, 2003). Most
enterprise applications are purchased after a process of investigation, piloting
and comparison using Request For Information (RFI) and/or Request for Proposal
(RFP) documents together with vendor demonstrations to evaluate the software
technically and financially (Tatsiopoulos and Mekras, 1999;
Ziaee et al., 2006).
These evaluation techniques are becoming increasingly inadequate for complex
integrated systems. Enterprises can instead use a hybrid approach called scripted
scenarios, as an alternative method for vendor selection (Peterson,
2003). A scripted scenario describes a unique problem that an enterprise
wants to be resolved and give project teams an opportunity to express their
vision for the business environment with the use of new software (Peterson,
2003). Scripted scenarios are valuable tools but they are not the only ones
in the overall software evaluation process (Lin et al.,
2006; Wu et al., 2007). It is the best to
use them together with RFIs and RFPs and take into account the results of site
visits, other reference checks and performance benchmarks.
Application vendors continue to support the rule of thumb that a 75 to 80%
fit between business requirements and available functionality is good enough.
|| Member functions for software acceptability
In order to increase this ratio and make a better selection new approaches are
required such as fuzzy logic or neural networks.
Real world is vague and assigning rigid values to linguistic variables means
that some of the meaning and semantic value is invariably lost. Fuzzy logic
that is emerged from the development of fuzzy sets theory by Zadeh
(1965), operates on a concept of membership such as the statement Software-X
is acceptable and can be translated as Software-X is a member of the set of
acceptable software and can be written symbolically as m (SOFTWARE ACCEPTABILITY),
where m is the membership function that can return a value between 0 and 1 depending
on the degree of membership.
Figure 1 shows, the membership function for the objective term (SOFTWARE ACCEPTABILITY) which has fuzzy sets Anot acceptable, Aindecisive and Acceptable. In this case when the score is 25 or below the software is not acceptable, when it is 75 or above it is acceptable. For values between 25 and 50 it is either Aindecisive but close to not acceptable where for values between 50 and 75 it is Aindecisive but close to acceptable.
The fuzzy set theory attempts to follow more closely the vagueness that is
inherent in most natural language and in decision-making processes. Therefore,
fuzzy logic has found many real-world applications that involve imitating or
modeling human behavior for decision-making in the real world.
Gungor Sen and Baracli (2006), specified in their study that decision making
strategies such as fuzzy methods are becoming to be more and more important
and will be better tools for appropriate ERP software selection. Xu
et al. (2007) has applied fuzzy methodology successfully for ERP
software selection of small and medium enterprises.
Shipping industry is one of the sectors that have limited number of related software producers with software already customized for their own customers. Traditional software evaluation techniques are not adequate for this industry. Most of the shipping companies use in-house developed software which is ineffectual. The aim of this study is to show that fuzzy logic can be used effectively for ERP software selection of shipping companies.
MATERIALS AND METHODS
The company that is considered as the real case application, is a medium-sized liner shipping company in Istanbul, Turkey, which is experiencing difficulties and delays in reviewing, approving, controlling and reporting operations between Head Office and Agencies due to processes that were overburdened with manual tasks, e-mail based workflow and lack of process information. Therefore, the company is in need of an information system to increase its effectiveness.
The company was established in 1996 within another company as a distinct, autonomous entity in order to accomplish the first national-flagged container transportation in the country. Companys philosophy is to provide the best service with a workforce that is expert in their field. It provides services with 18 of its own ships and with 62 years of experience in transportation.
Company has a total of 5 departments: container movement control department, operations department, trade department, line management department and foreign accounts department. These departments will be used for determination of critical processes.
Mainly, AS 400 system is used at Head Office and agencies. Some patch programs are prepared to solve problems like special freight rate approval or free time extension for ports but these were not effective because of AS 400 database. Eighty percent of work is done manually. Agencies send fax or e-mail to inform the Head Office about operations they perform. Head Office users enter all data into current system as much as possible. Delays and errors occur frequently in stored data. There are a lot of paperwork and double entry to different software. Reports are prepared manually after a time consuming effort. Variety and correctness of the reports are very poor. Controlling agency operations are sometimes impossible. Company profitability calculations, line management, container control, ship husbandry, disbursement accounting can not be made efficiently with the current system.
A program that supports operations between Head Office and agencies as well as between agencies and customers is desired. Agencies should contribute to the system through internet, connect to the main server and perform authorized operations. In this way, the data will be entered only once. Container movements should be monitored on a daily basis. Line profitability, costs and revenues per voyage, per container and per port should be correctly calculated. Disbursement accounting between agencies and Head Office should be established. Customer relationship management should be supported. Several types of reports should be generated.
Goals of the new system can be summarized as:
||Real time information and provision of estimates
||Less transactions and less errors with automated processes
||Control on agents, equipments and vessels
||Managing disbursement accounting
||Allowing managers to concentrate on strategic tasks and future
There are only 3 major information system software vendor companies in shipping
software business. The products of these companies have many differences in
details and workflows. For this reason it is hard to evaluate the performances
of these companies with traditional approaches. This study will combine RFIs
and RFPs together with scripted scenarios. Subsequently, fuzzy logic will be
used as decision aid in order to find the optimum software package.
The ERP software selection project started in March 2005 and ended in December 2006. The project is performed in 5 steps:
||Project teams preparation
||Preliminary RFI preparation
||Scripted scenario and RFI development
||Scoring of RFIs
Project team is formed by including members from all departments as studies
by Verville and Halingten (2003) and Wu
et al. (2007) had suggested. In order to speed up the documentation
process the team is divided into two: (1) Head Office Information Technology
(IT) team and (2) Office team including members from. Daily meetings are performed
for the preparation of RFI/RFP, scenario and scenario evaluation documents.
These meetings proved that communications of departments were very poor. Every
single Office team member prepared a requirements list for their own department.
The requirements definition consisted of functional requirements, managerial
requirements and technical requirements (Nikolaos et
al., 2005; Van Staaden and Lubbe, 2006). Head
Office IT team combined the requirements into 16 operational groups and prepared
an RFI document for distribution to shipping software vendors. Questions in
each operational group can furthermore be grouped as follows:
Functional: These questions referred to the business needs of the company based on specific operations of shipping industry. For vessel operations questions as ADoes the system allow multiple vessel details to be captured or Does the system hold port profiles of all seaports around the world? were added.
Managerial: This group consisted of questions related managerial needs of the company including financial and sales/marketing issues. Questions generated were associated with the reporting issues of management such as statistical reports, exception reports, agent performances and other performance metrics. Financial questions included invoice related topics as well as budgeting, exchange rates, transfer rates, freight charges, banker guarantees, cash and check receipts, etc. Sales and marketing questions are related to the information kept for customer profile, sales contact management, customer complaint handling, etc.
Technical: This group of questions dealt with type of reports (screen, print or file), internal and external information updates, data verification, information dissemination techniques (e-mail, fax, etc.), security, user rights and logging issues, other technical issues such as use of barcodes on forms, etc.
Scenarios are prepared with the help of Head Office IT team only for the critical
processes of 5 departments as Wu et al. (2007)
had suggested. An example is a quotation scenario for trade department including
customer detail information, shipment type and corridor details, commodity and
container details, route information, charges, detention and demurrage, development
of processes for quotation closing, approval and finalization as well as booking
from different resources (e.g., quotation, guideline or contract). A second
RFI is prepared to evaluate vendor demonstrations of scripted scenarios. This
RFI is prepared with the aid of Head Office IT team. In this phase questions
are prepared to be more detailed and answer classifications are prepared differently
to evaluate the degree of accomplishment. Questions are classified into two
sections: (a) functional requirements and (b) reporting needs.
|| RFI answer weights and meanings
RFIs can be assessed by scoring of expected performance of an option
and its relative importance to the decision (Nikolaos et
al., 2005). The scoring of preliminary RFI is performed by question
weights (defined at Office team meetings showing relative importance) and answer
weights (defined at Head Office IT team meetings showing expected performance).
Question weights used a 3-point scale; 1 being essential, 0.8 being desirable
and 0.5 being nice to have (Axia, 2009). Answer weights
are given in Table 1. At the end of the evaluation, question
weights are multiplied by answer weights and then summed to form the total score
of the vendor. There were total of 597 questions with a maximum score of 5789.
Scenario Demonstrations RFI consisted of two sections; functional and reporting
needs. The question and answer weights are determined at Office team and Head
Office IT team meetings (Axia, 2009). Question weights
of every functional item are specified as 1 and every reporting item as 0.5.
Answer weights of functional items are specified as 0 for Not Available, 3 for
Modification Needed, 8 for Customization Needed and 10 for Available. Answer
weights of reporting items are specified as 0 for Not Available, 1 for Modification
Needed, 4 for Customization Needed and 5 for Available. At the end of the evaluation,
question weights are multiplied by answer weights and then summed to form the
total score of the vendor. There were 133 questions in this section with a maximum
total score of 1215.
Lien and Liang (2005) stated in their study that there
are three main factors in ERP software selection: cost, development time and
performance. They have specified that cost is the most important factor and
development time is the second important factor. For this reason, total RFI
score, which is the indicator of performance, is used as input to fuzzy engine
together with purchase cost and development time. Development time is retrieved
from RFIs. Purchase costs are considered for 100 user licenses.
for Purchase Cost are specified in the range of $20,000 to $45,000 (Fig.
2). This range is specified according to the prices of major ERP software
packages on the market. Values greater than $40,000 indicate that the software
is expensive. Values less than $25,000 indicate that the software is cheap.
Values between $25,000 and $32,500 are classified as normal-cheap where values
between $32,500 and $40,000 are classified as normal-expensive.
|| Membership functions for purchase cost
|| Membership functions for development time
Membership functions for Development Time are specified in the range of 50 to 350 days (Fig. 3). This range is specified according to the prices of major ERP software packages on the market. Values greater than 300 days indicate that the development time is long. Values less than 100 days indicate that the development time is short. Values between 100 and 200 are classified as medium-short where values between 200 and 300 are classified as medium-long.
Membership functions for RFI Score are specified by the team members in the range of 4500 to 7000 (Fig. 4) where 4500 being the minimum acceptable and 7000 being the maximum score. Values greater than 6200 indicate that the RFI score is high. Values less than 4700 indicate that the RFI score is low. Values between 4700 and 5500 are classified as average-low where values between 5500 and 6200 are classified as average-high.
Fuzzy rules are prepared by IT specialists according to the company needs, project duration constraints and the project budget. An example rule can be given as:
If (RFI score is high) and (development time is medium) then (output is excellent) (0.4).
MATLAB Fuzzy Toolbox is used for evaluations. Membership functions for the output compatibility is specified in the range of 0 to 100 (Fig. 5). Values less than 20 indicate that the compatibility is bad. Values greater than 90 indicate that compatibility is excellent. Values between 20 and 40 are classified as poor-bad, values between 40 and 60 are classified as good-poor, values between 60 and 75 are classified as very good-good, values between 75 and 90 are classified as excellent-very good.
|| Membership functions for RFI score
|| Membership functions for compatibility
RESULTS AND DISCUSSION
In the real case application, software selection process started with RFI scoring. Table 2 shows the preliminary RFI scores. According to these scores best candidates are Company A and Company B.
Table 3 shows the demonstration RFI scores. There is no clear distinction for the scores of all companies which are in the range of 84 to 87%.
According to total RFI scores of the companies, Company A is in the high score range with the value of 6,209. Both Company B and Company C are in average-high range with scores of 5,996 and 5,763, respectively.
Using results of traditional methods organization decided to buy the software from Company A and tried to implement it for a duration of 3 months. In the mean time the study is extended to incorporate fuzzy methods. Table 4 gives the input values of the fuzzy engine together with the results of the fuzzy evaluation.
The development time input pointed out that Company B can be a candidate with short development time and Company C can be the other candidate with low Purchase Cost. With the use of fuzzy methodology it was possible to categorize the companies distinctively. Company B is the best choice with the compatibility range of excellent-very good. Company A was in the range of very good-good where Company C was in the range of good-poor. The implementation is then further reviewed and found out that there was a popular discontent. The organization then decided to stop the first implementation which was not proceeding successfully and decided to switch to the software of Company B. The implementation of this software was successful and the software is running effectively since May 2008.
|| Preliminary RFI scores
|| Demonstration RFI scores
|| Inputs and outputs of the fuzzy engine
This study is aimed at finding a fuzzy based model for ERP software selection of a liner shipping company. For this purpose; two project teams are formed, preliminary RFIs are prepared according to departmental requirements, scripted scenarios are prepared for the critical processes of the organization and results are evaluated by fuzzy methodology.
Based on the results it can be concluded that:
||Traditional software evaluation methods, such as RFI scoring
or scripted scenarios, are not sufficient for ERP software selection. This
result differs from the studies of Lin et al.
(2006) and Wu et al. (2007). This may
be due to the fact that even though the software seems to have appropriate
functionality for the organization, key issues such as development time
and purchase cost can affect the final decision (Lien
and Liang, 2005)
||Fuzzy methodology is a better tool for ERP software selection.
This result is in support of the study of Gungor Sen
and Baracli (2006)
||ERP software selection with fuzzy methodology can be used
effectively for shipping industry. Xu et al.
(2007) have specified that fuzzy methodology can be applied successfully
for ERP software selection of small and medium enterprises
However, the proposed model is not without its limitations. It is limited to the findings of only one sector; therefore it can not be generalized. But the development of a general reference model can be a fruitful area for further study. Vendor site visits, vendorsearlier projects, number of successful projects, evaluation of companies that are currently using vendor software can be other factors that can affect the results and they may need to be further explored.