Erp Project Cost And Effort Estimation

Published Date: 02 Nov 2017

Edwin Yee Hsien Loong

Universiti Putra Malaysia

Author Note

This paper was prepared for SIM4354 Enterprise System Development, taught by Dr. Koh.

Abstract

In recent years, the interest in Enterprise Resource Planning (ERP) solutions had been continuously expressed by organisations. Significance of ERP systems greatly helped in promoting the growth of the industry. Regardless of its rapid growth, the problems with estimating implementation cost of ERP systems linger to trouble all parties involved. Despite all traditional estimation models we have, it is still hard to account all possibilities reside with a complex system. In this review report, analysis on three journal papers will be performed to find out glitches with existing estimation methods like COCOMO. Possible method proposed by researchers to overcome these problems will also be discussed further in this document in order to identify the best estimation model for future development purpose.

ERP Project Cost and Effort Estimation

The world nowadays in growing as fast as it could and that stimulates the needs of ERP Solutions because of its important functionalities. The fast track in the industry has urged more rapid development of new ERP systems, thus, precise and accurate cost and effort estimation became more significant than ever. In order for development process to be smooth and fast, accurate estimation must be done. For this to be done, three journals which are Balancing uncertainty of context in ERP project estimation: an approach and a case study by Daneva M. (2010), Effort estimation for enterprise resource planning implementation projects using social choice – a comparative study by Koch S. and Mitlöhner J. (2010), and COTS integration and estimation for ERP by Rosa W., Packard T., Krupanand A., Bilbro J. W., and Hodal M. M. (further known as Rosa W. et al. in this document) (2012) had been studied in detail.

Enterprise Resource Planning (ERP) software is business management software that allows an organisation to benefit a system of integrated applications to govern the business. ERP software integrates all core functions needed for any operation, including sales, manufacturing, development and marketing. Due to increasing demand for ERP systems, development process is expected to be faster. Continuous efforts had been done in the past in hope to tailor cost or effort estimation suitable for ERP systems. Daneva M. (2010) had brought our attention to six different challenges when developing ERP systems which are ERP solution ownerships, architecture boundary, configuration, project duration, complexity, and size. While on the other hand, methods that are readily available nowadays left large areas to be defined by user and while most users do not have the data upon which to base detailed estimates, the ability to obtain accurate cost estimation remains difficult. (Rosa W. et al., 2012) Furthermore, approaches based on function point metric or approaches designed for new software development like COCOMO, COCOMO II are problematic. (Koch S., Mitlöhner J., 2010)

One of the most difficult aspects of estimating ERP effort is that of evaluating size and complexity of the ERP implementation. (W. Rosa et al., 2012) Equal hassle has been described in another two journals mainly because of the size (Daneva, 2010) and customising the ERP system (Koch S., Mitlöhner J., 2010). Despite all these hassle, why we still estimate the cost and effort needed by ERP development or integrating process? Effective estimation is undoubted factors for any successful development projects. A well-defined project can also contain all uncertainties that may occur in the middle of development. Project with proper and effective estimation keeps the whole costs under the budget, helps with delivering system in promised deadline, and most importantly, satisfied customers. Contrary, disastrous estimation results in failed development projects. In simple, effective estimation determines the success or failure of the ERP system. To solve the difficulties with current ERP project estimation methods, we will look deeply into three journals mentioned for the best solution.

Daneva M. (2010) showed her solution by complementing multiple techniques which are COCOMO II reference model, the Monte Carlo simulation, and the effort-and-deadline-probability-based portfolio management concept. This method was previously used by Jiamthubthugsin W. and Sutivong D. but Daneva M. (2010) applied this combination in new setting, namely ERP. (Daneva M., 2010) The method consider project quote that consist of three components, estimated cost, turnover, and contingency. However in the journal, this method does not cover much about profit and contingency. The approach by Daneva M. (2010) posits a trade-off of probability of success for schedule and other resources when producing estimation. Precondition for this approach to be applied are the ERP adopter’s organisation has reasonably characterised the context, has established some operational variance for the size of the ERP and the complexity of the subprojects included in it, has some understanding of the ranges of difficulty in obtaining the right configurations for the system, and has some calibrated way of processing this information. Diagram below is a diagram cited from the same journal – Balancing uncertainty of context in ERP project estimation showing simple flowchart of the proposed methodology.

C:\Users\Edwin Yee\Desktop\Figure1.jpg

Figure 1: High level view of the solution approach (Daneva M., 2010)

The proposed solution by Daneva M. (2010) allows estimation to take consideration of ERP adopter’s specific cost drivers (COCOMO II model), to approach cost drivers’ degrees of uncertainty (Monte Carlo simulation), and to compute the probability for success with planned mutually dependent deadlines for a group of related ERP projects (effort-and-deadline-probability-based portfolio management concept).

The second methodology is proposed by Koch S. and Mitlöhner J. (2010) and the methodology provided is by using social choice in effort estimation. This social choice method is altered by replacing voters with project attributes in order to estimate the effort. Each attributes considered provides a ranking of all projects available in database. It is easy to obtain weak ranking from numeric project attributes because indifferences are allowed. Experts need only to for past projects in terms of the numerous attributes, which simplifies the collection of input data. Each ranking therefore gives the impression of ‘complexity’ or ‘size’ of the projects according to one attribute. The aggregated ranking of all attributes will then represent an overall picture of the projects’ ‘complexity’, and thus a ranking of the efforts required to implement them.

There is no quantification when using this method (Koch S., Mitlöhner J., 2010) the project is just put into respective ranking according to each attributes. Later, the project is ranked again in aggregate ranking with others projects with effort known. The neighbours’ effort value can be used to calculate the mean.

C:\Users\Edwin Yee\Desktop\Figure2.jpg

Figure 2: simple example cited from Koch S., Mitlöhner J. (2010). Effort estimation for enterprise resource planning implementation projects using social choice – a comparative study.

The approach proposed by Koch S. and Mitlöhner J. (2010) is able to cope with multi-dimensional characteristics of a project, which is essential for fully capturing of this type of project. The approach also provides simple and intuitive way to define the project by ranking it among existing projects.

Last but not least, the solution proposed by Rosa W. et al. (2012) in COTS integration and estimation for ERP. The tactic used is built on causal relationships established in earlier studies, given that a set of Schedule Estimating Relationships (SERs) and Cost Estimating Relationships (CERs) developed using regression analysis and based on readily existing parameters- reports, interfaces, conversions, and extensions (RICE) objects. (Rosa W. et al., 2012)

C:\Users\Edwin Yee\Desktop\Figure3.jpg

Figure 3: Correlation coefficient matrix cited from Rosa W. et al (2012)

Rosa W. et al. chose pairwise correlation analysis as they only have 20 observations while 200 observations are needed for structural equation modelling (SEM). There are also some variables not measured in the analysis as they showed poor correlation to effort: users, migrated legacy systems, legacy system interfaces, and number of sites. (Rosa W. et al., 2012) RICE object can explain up to 89% of the variation in ERP software engineering effort but there are four limitations of this approach: larger dataset (>20) is preferred to increase validity and accuracy, non-random sample will also be favoured, the population responded to study is only 20, and the study doesn’t check for response from non-respondents.

Studies performed in three journals are extremely useful for further work of estimating effort for ERP system as they certainly posed the different level of method for different users groups. Koch S. and Mitlöhner J. (2010) provide an easy to understand comparison method for entry level, while Daneva M. (2010) and Rosa W. et al. (2012) on the other hand proposed deeper level of calculations that have higher learning curve for high level users. It’s certain that every method have different approaches to calculate the effort needed during ERP system development. The differences between these approaches can only differentiated in real development projects, some of the data provided are might not as accurate as the data might be incomplete (Rosa W. et al., 2012) or fraught (Daneva M., 2010). Apart from what we had shown in this document, there are other approaches (Koch S., Mitlöhner J., 2010). Further studies needed to be done in order to find an all-rounder approach in estimating effort in ERP projects.