Designing FSS for the Supply Chain

Designing FSS for the Supply Chain:Through Action-Oriented User Interfaces

Stavros Asimakopoulos1, Robert Fildes1, Alan Dix2,

1 Department of Management Science, Lancaster University Management School, LA1 4YX, UK,

2 Computing Department, Lancaster University, Bailrigg, Lancaster, LA1 4YR, UK

Email: , ,

Stavros Asimakopoulos, Robert Fildes, Alan Dix

Abstract

This paper explores forecasting support systems (FSS) design issues based on interview and observational data gathered from professional designers and users working in supply chain forecasting. We show that combining action-based product knowledge with historical data can be usefully incorporated in an FSS user interface. The emergent theoretical framework depicts various motives behind the use of FSS: the need for balanced visualization of data and product knowledge, and attention to product behavioural characteristics. Moreover, user interfaces should support negotiation and informal communication aspects that are generated through forecast reports and review meetings. The framework emphasizes also the need to merge reasonable forecasts and action-oriented user interfaces. The design implications for existing and new FSS are discussed. Indeed, these novel human-system interactions approach to the design of FSS as actionable and knowledge rich resources that address temporal organizational arrangements.

Keywords: Action-oriented interfaces, FSS design, grounded theory, supply chain forecasting.

1. Introduction

Being able to forecast product sales demand has been widely recognized as an important aspect of business planning and management. Prediction of future sales demand may also help organizations to identify market opportunities, enhance customer relationships, increase customer satisfaction, and reduce inventory investments while improving customer service [1],[2],[3]. Forecasting support systems (FSS) are typically employed to support managers in effective decision-making. Taken from [3], [dictionary] FSS can be defined as:

“A set of procedures (typically computer based) that supports forecasting. It allows the analyst to easily access, organize, and analyze a variety of information. It might also enable the analyst to incorporate judgment and monitor forecast accuracy.”

According to [4], FSS use organizational databases and statistical models to produce forecasts and allow better decisions. The semantic level operations for decision support systems are often organized into four phases: problem definition, data selection, model selection, and execution. The syntactic level is frequently couched in table-related language using rows, columns, and cells (as in spreadsheets). It also includes mathematical functions and algorithms that can be applied to the data in the tables. Sales forecasting systems are used to compute future sales demand for each stock-keeping-unit (SKU) on a daily, weekly, or monthly basis.

The major themes considered in FSS research as used in organizations are the following: (1) the use of statistical methods so as to ensure ‘optimal’ forecasts, and (2) how to support users when they adjust the statistical forecasts during the forecasting process (see for example [5],[6],[7],[8] and [9]). Current research suggests that FSS are not used effectively so as to support users when they wish to adjust the statistical forecasts [9],[10]. [11] In their review of forecasting software conclude with the following three areas for promising research in FSS:

1. Data management
2. User interface design
3. Forecasting process ‘intelligent’ support.

Most studies tend to either focus on a particular task (e.g. how to support user adjustments to statistical forecasts) or a particular system feature or function (e.g. statistical method selection). The main obstacle of the forecasting literature is that it has difficulty in addressing system design issues that are driven by people-system interactions in the supply chain. Existing forecasting research does not also attempt to equip designers to deal with the complexity of the working context, while users’ role in designing FSS is largely left unexplored.

Moreover, there is no research to date that examines the rich supply chain context based on designers’ and users’ accounts in such a way that is useful and informative for FSS interface designs.

Amongst the few studies that deal with FSS design, [12] suggest that forecasting system designers should consider the ways individuals use their systems to produce forecasts. This is due to the fact that a mismatch between the software designer’s model of how a system will be used and the actual use is likely to impair the system’s functionality. [12] Suggest that adaptive forecasting systems could be designed to recognize particular user strategies at an early stage of the forecasting process, enabling the interface to adapt to the particular needs, strengths and weaknesses of these users. For example, the system could highlight information that was not taken into account sufficiently (such as the lack of fit of the chosen forecasting method or its inability to deal with a trend in the series), and also guide the user towards the selection of most appropriate methods.

[13] (in press) also call for research on FSS design. Specifically, they believe that there is scope for developing and testing software facilities that allow advice and information obtained from multiple sources to be used and combined in a structured way. Given also the widespread use of forecasting review meetings there is also scope for the development of group FSS, which allow managers to feed independent estimates of, required adjustments into the system. Such improvements in system design might also help to mitigate the pressures towards bias, both personal and organizational, that are often evident in supply chain companies [14].

Consequently, this paper investigates FSS design adopting a human-computer interaction (HCI) perspective. The context of research is product sales forecasting that organizations use to inform operations that are taking place along the supply chain (e.g. orders, production, planning, and budgeting). To do so, a theoretical framework has been established derived from interviews with professional designers and users and observations of FSS. Section 2 reviews the major aspects of FSS use. Section 3 describes the theoretical framework and its implications for FSS design. Section 4 summarizes the research findings for FSS design and use.

2. Aspects of FSS use in organizations

Recent studies and reviews have identified gaps in our understanding of the relationships between systems and techniques used for forecasting, and the behavioural factors associated with the management of forecasting in organizations [1],[6],[15]. Forecasting researchers argue that accurate forecasts are highly dependent on organizational activities that enable these forecasts to be effective.

2.1Towards effective forecasting process

In particular, [1] developed a theoretical model of the use of a forecasting system drawing on findings from organizational behaviour and diffusion research literature and interpreted into a multi-unit case study. The authors tested their theoretical arguments in a study of a company with 10 separate divisions. Their final sample consisted of interviews with 45 employees with responsibilities in forecasting (26 product managers, 15 demand planners, and 4 market researchers). [1] suggest that:

• Statistical techniques and forecasting systems should be designed to meet user needs.
• Forecasting implementation depends on the database system and supporting organization design.
• Effective forecasts require people’s accountability, adequate time and resources.

[15] explored the factors associated with better sales forecasting practices in 20 organizations. The study consisted of in-depth analyses of company processes and documents, as well as interviews with forecast users and forecast developers. The findings revealed a set of dimensions for effective forecasting management. Specifically, the proposed framework considers functional integration as the role of collaboration, communication, and coordination of forecasting management with the other functional areas (e.g. marketing, sales, finance, production, and logistics) of the organization. The approach dimension considers the role of products and services to be forecasted, the forecasting techniques used, and the relationship between forecasting and planning. The systems dimension address the evaluation and selection of hardware and forecasting systems to support sales forecasting as well as the integration of forecasting systems with other management information systems within the organization. Lastly, the performance dimension relates to the metrics used to measure sales forecasting effectiveness and its impact in organizations.

These exemplar studies focus on the need for accurate forecasts and organizational strategies that aspire to deliver an effective forecasting process. However, these studies have neglected the role of FSS design in influencing overall forecasting effectiveness.

2.2The role of user adjustments in forecasting practices

The practice of sales forecasting is characterized by a persistent preference for user judgments over statistical models [7],[8],[16],[17]. This strand of research explores how users of forecasting systems should carry judgmental interventions combining them with the benefits of statistical forecasting methods [6],[18],[19]. The survey study by [5] focused on issues of familiarity and use of statistical methods, the ease of use and the satisfaction from use of the statistical methods, and the reasons for applying judgmental approaches to forecasts. [5] attributed user judgments to the following: perceived accuracy, ownership and control of forecasts, incorporation of knowledge from special events, and self-serving biases (at the individual user level), and difficulties in to obtaining historical data, support from upper management, and a lack of training (at the organizational level).

Recent observations of forecasting systems in operation [9] confirmed the importance of user adjustments in organizational forecasting. The following issues were also specifically highlighted that: (i) users adjusted either the parameters of the forecasting method or its components (e.g. seasonal factors) in order to improve the method’s forecasts of the underlying time series (ii) users often selected default parameter values or sub-optimal statistical methods. [13] suggest that requiring forecasters to record reasons for their adjustments in a standard format (e.g. by selecting a reason from a list) might serve to reduce the number of relatively small, but damaging adjustments that may be based on misinterpreting noise as signal or reflect gratuitous tweaking of the forecasts [18]. A list of reasons would also allow forecasters to understand why and how market intelligence is so often misinterpreted. In addition, a list of reasons would assist the decomposition of market intelligence into key drivers, thereby lessening the likelihood of double counting. Experimental evidence by [20] also suggests that the incorporation of guidance systems such as those which allow the formal use of analogies (e.g. past promotions and their effects) would improve the quality of judgments based on market intelligence.

The experimental and empirical evidence point to the need for the design of FSS to explicitly acknowledge the role of judgmental adjustments based on substantive knowledge of market drivers.

3. The theoretical framework and its implications for FSS design

In order to understand the design of FSS, semi-structured interviews and associated demonstrations of systems use were gathered from 10 software designers and 10 users. Their average years of experience in supply chain forecasting was 13.3 for users and 20.1 for designers, respectively. The grounded theory method [21], [22] has been employed in order to coordinate the data collection, and analysis process. The interview guide was partially adapted from [15] to address issues of organizational forecasting but this was further enriched with questions specifically applying to designers and users. The interviews lasted between 1 and 1.5 hours and focused on three key areas. Specifically, the process of developing forecasts was clearly an issue that the interviews addressed. Organizational aspects of systems use were also considered as a focus of the interviews with emphasis on the different people involved in the forecasting process forecast accuracy considerations, and social/organizational factors that may influence forecasts. Lastly, system functionality and design generated a set of thoughts and activities that people engage in when producing product forecasts.

The resulting theoretical framework (consisting of the main properties and their properties) accounts for FSS use on supply chain forecasting (figure 1). The framework has then been used to structure FSS design requirements.

Figure 1: Concept map depicting theoretical framework of designers and users

Based on the framework, it is possible to outline five broad design support areas for existing and future FSS interfaces. These are the following:

(1)Demonstrate special features of specific products

(2)Support for product knowledge generated from informal communications during the forecasting process

(3)Provide features that enable dynamic interchange of historical data and product knowledge

(4)Provide users with the ability to annotate and negotiate elements of forecasting

(5)Enhance user awareness of organizational knowledge by providing appropriate interface navigational cues.

The theoretical framework provided evidence that knowledge visibility and availability is an essential component of user interaction with forecasting systems and as such it should be incorporated at forecasting user interfaces. Products in the context of supply chain act as focal points for users when forecasting. Interestingly, the specific product characteristics and the knowledge users reported indicate socially constructed aspects of supply chain forecasting.

3.1Implications for FSS research and practice

This study has several implications for FSS research on the supply chain context. First, it indicates that models and studies that are concerned with use, design and evaluation of FSS should take into account specific aspects of the social context. Our study suggests that the concepts of negotiation, informal communication, for example, create the social fabric necessary to facilitate product forecasting through the creation of deeper relationships and increased opportunities for action-oriented user interfaces. The evidence also suggests that simply choosing and fitting statistical models and adjusting forecasts (when appropriate) do not alone create an effective FSS for supply chain applications. Instead, it is the connections gained through product specific characteristics, thinking in terms of actionable forecasts, responding and negotiating based on product specific knowledge, and informally communicates aspects of forecasting process that improves FSS. In a supply chain context, FSS becomes even more important because it affects user effective interaction and produced results.

This study also has multiple implications for practice. First, those designing and implementing FSS should facilitate the creation of action-oriented interfaces as apply to the specific context of use. An excellent way is to provide features that require users to brainstorm when producing forecasts with regard to how they recommend specific actions due to product forecasts. These can be very specific (e.g. advertising) or general (e.g. better communication through regular meetings with marketing and/or sales people). For example, in organizational forecasting some people may have a better understanding of statistics, while others may have a better knowledge of the market. This proposal suggested by designers and expanded in this study not only should point to achieve effective forecasts but consider specific actions that should be taken for richer communication and connectedness.

3.2Limitations

Although the theoretical framework of this study provides insights into action-oriented FSS, a number of limitations must be considered when interpreting the concepts and their properties. First, this study represents the first set of a theoretical framework and should be subjected to further testing with different users and designers, contexts, and specific commercial systems. It should be acknowledged that qualitative data of this nature and grounded theories are focused, substantive and cover the specific context but are also difficult to generalize. Rather, they are substantive to the settings from which they are derived. On the other hand, when reliability is considered, it is very difficult to exactly replicate a grounded theory study because no two situations, contexts or research requirements are alike. It is thus more appropriate to ask whether or not the emergent theoretical framework, if consistently applied to a similar situation, will allow researchers to interpret, understand, and adequately explain newly observed phenomena and social process. To this extent, the grounded theory framework and that emerged from this study can claim reliability. In addition, the method and analysis is prone to researcher bias due to the reliance on one researcher as the primary interpreter of interviews and creator of the essential categories of the theoretical framework. However, every effort has been made to overcome this bias in the data analysis and interpretation process by triangulating and cross validating by using several types of data and documents, and by discussing interpretations with fellow researchers. Thus, the research findings have gained validity.

3.3Future research

Apart from applying some of the findings to similar technologies (e.g. in the design and use of enterprise resource planning systems), the current research suggests four strands of future work that appear particularly promising with regard to FSS design.

Firstly, the investigation of the social and organizational dimensions of demand forecasting systems may be fruitful in order to fully appreciating the ways technological artefacts fit within the particular setting [12] (for an early example). Focus on the adoption and uses of forecasting systems may enable some more focused observations on how (a) negotiation and product knowledge are developed through time in organizational settings (b) how people develop their skills because of the availability of the technology and (c) how the working setting encourages and/ or possibly discourages effective user system interactions. Research of this type would generate an integrative theory of the socio-technical process that takes place around FSS and discuss their implications for design.

Furthermore, the concept of communication/negotiation, its elements and how these may be interpreted through design features and characterizations has not been discussed before in academic research on FSS. Another interesting research question is to investigate how to design FSS to maximize user reactions to the produced forecasts, and intentions to participate in negotiations in organizational initiatives. Research in this area can be informed by longitudinally studying the relationships between user interactions, social context, and how forecasters use the FSS to create a shared understanding and knowledge resource.