- Email: [email protected]
Why do quality and reliability feedback loops not always work in practice: a case study
Reliability Engineering and System Safety 75 (2002) 295±302
www.elsevier.com/locate/ress
Why do quality and reliability feedback loops not always work in practice: a case study P.A. Molenaar a, A.J.M. Huijben b, D. Bouwhuis c, A.C. Brombacher a,* a
Section Product and Process Quality, Faculty of Technology Management, Eindhoven University of Technology, Building Pav., P.O. Box 513, 5600 MB Eindhoven, The Netherlands b Philips Medical Systems, P.O. Box 90050, 5621 JG Eindhoven, The Netherlands c Faculty Technology Management, Building TEMA., P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Abstract The increasing competition in the market of consumer electronics forces industry to simultaneously improve the functionability, reliability and costs of their products. Due to the strong dynamics of this ®eld an important measure in improving product quality is the feedback of information on actual ®eld behaviour of a product which is extremely relevant for product development. This information, usually gathered by the service department, can help development to learn from the failures of earlier products and to improve the next product generations. This feedback process, however, does not always bring the expected bene®ts. This paper presents a model, based on an actual case study, of the structure of the feedback process in terms of information ¯ows. The information ¯ow model describes the chain or network of activities in the feedback process as a function of (different) drivers such as cost, time or quality. Using this information ¯ow model it is possible to explain how quality and reliability related information propagates through a business process and where and why communication problems lead to disruptions in the feedback of this information. With this model, it is easier to understand the function of different parties in the service or after sales process, so that management of the information in this process becomes more transparent and can be improved. q 2002 Elsevier Science Ltd. All rights reserved. Keywords: Product quality; Product reliability; Customer feedback; Service; After sales process; Information ¯ow; Maturity index on reliability
1. Introduction The market for consumer electronics shows an increasing complexity. Several trends that make the development of new products more dif®cult can be distinguished. Products get more complexity all the time, while quality and reliability demands an increase in products. Changes in warranty legislation result in the fact that manufacturers have a broader responsibility for the behaviour of their products during a longer period. These contradictory terms of increasing product complexity and increasing product liability are accompanied by the need to shorten the development times. This is due to the heavy competition in the market of consumer goods where `time to market' is currently one of the dominant business drivers [1]. Ways to improve quality and reliability of products without increasing the development times have been the subject of numerous research projects over the last decades. In the past, the main focus on product reliability was on the reliability of components * Corresponding author. Tel.: 131-40-247-2390; fax: 131-40-246-7497. E-mail address: [email protected] (A.C. Brombacher).
[2]. Both due to the improvements in component reliability and due to a number of other aspects that will be addressed in this paper component failures no longer dominate product reliability currently [3]. Earlier papers stressed the importance of the reviewing quality and reliability not only of the products but also of the underlying business process [4]. Especially in fast product creation processes (PCPs) the majority of the problems is often caused by lack of information deployment (problems are known in the PCP but not at the right moment at the right places) and not by problems that were generally unknown (or accepted) in the PCP [5]. This has led to models that use the analysis of information ¯ows in the PCP for analysis and improvement of the quality and reliability of the product. A schematic overview of the business processes considered in this paper is shown in Fig. 1. In those theoretical cases where product , customer interaction is fully known throughout the process it could be possible to design a PCP without the necessity for a feedback system. In branches such as high-volume consumer products, due to the strong degree of innovation and the strongly dynamic business processes, this will hardly ever
0951-8320/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0951-832 0(01)00117-X
296
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
Fig. 1. Information ¯ow in a product creation process [6].
be the case [7]. Due to combinations of new technology, new customers and new structures used in the development process [8] certain aspects of product quality and reliability will only be found at the customer. In many companies the after sales (or service) channel is one of the two interfaces with a customer. In the marketing department the attitude of the customer towards a future product is anticipated and projected into future products. In the service channel, however, the experiences of real customers with real products are available, especially where dissatisfaction is involved. Especially with strongly innovative products this information may have considerable differences with the information anticipated by the marketing department. Information on the root causes of these failures, therefore, must be deployed to all relevant parties involved in the creation of the product. The structure of this feedback process can to a large extend determine, as this paper will show, the company's (in)ability to prevent problems in the future. This paper focuses on an information ¯ow model of the after sales process (ASP) and shows why in many cases the feedback process is not able to generate the required information, relevant for other parts of the development process. To model this structure the so-called `Maturity Index on Reliability' (MIR) method was used. Section 2 will ®rst explain the MIR model and will then apply the MIR model to the service or ASP of an actual industrial case study. 2. Analysing the information ¯ow for the service- or after sales process In order to model the quality of feedback structures in a business process, it will be necessary to distinguish the separate ¯ows in the process. In a business process, such as the development and realisation of high-volume consumer
products two main ¯ows and a set of control factors can be distinguished: ² Physical ¯ow: The transformation process of ideas and (raw) materials into a working product on the market (feedforward) and of malfunctioning products back to the supplier (feedback). ² Information ¯ow: Information on the above products with respect to function, cost, time and quality. ² Control factors: Factors that either control (enable/ disable) the physical ¯ow or the information ¯ow. Traditionally the main emphasis in quality control has been on the control of the quality of goods: the physical ¯ow. This paper will take a different viewpoint and will look especially into the information ¯ow and will develop a model to explain product quality in terms of the quality of these information ¯ows. It will focus especially on disruptions in information ¯ows in order to explain (lack of) product quality. 2.1. Analysing information ¯ows in business processes The availability of information in one part of a business process by no means guarantees that this information reaches the right people at the right moment with the right level of detail [9]. The transfer of information will require certain activities, each of which are enabled or facilitated by certain enabling conditions or business drivers (Fig. 2). A major problem in analysing information ¯ows is the socalled `quality of information' [4]. Although the statement `the product does not work' may seem as a valid statement from a customer's perspective this information is, in most of the cases, useless for engineers involved in the development of future products. In order to determine whether the right information reaches the right person at the right moment in time a so-called
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
297
Fig. 2. Information ¯ow model of service process.
maturity index on reliability (MIR) model was developed. This MIR model uses a scale of four qualitative levels to describe the quality of information in an information ¯ow model: ² MIR level 1 (How much): Quantitative information on failures is available to the relevant persons ² MIR level 2 (Where): The relevant actors, with the ability to in¯uence/control the mentioned failures can be identi®ed and have access to this information ² MIR level 3: (Why): The technical and/or organisational root-cause of the failure can be determined and controlled by the mentioned actors. ² MIR level 4: (What to do): Adequate measures to prevent re-occurrence of the failure can be identi®ed and implemented. An example of a strongly simpli®ed MIR model is given in Fig. 3. It presents a strongly cost driven repair shop. In order to operate repair activities ef®ciently, information on causes and remedies (MIR3) is stored for internal use but no structural feedback to the manufacturer takes place on the causes of failures and the possibilities to prevent them in future products. The information exchange to the supplier concentrates only on ®nancial aspects (number of repairs to
be reimbursed; MIR 1). Therefore failures most likely will re-appear in future products. It can also be argued whether the repair strategies adopted by the repair shop (with limited detailed knowledge of the product) will result in optimal repair strategies, especially when products with a high degree of innovation are involved. A full MIR information ¯ow model is not based on the analysis of a single activity but on the analysis of a network of interrelated quality and reliability activities Ð and the information exchange between these activities Ð in a given business process. The main activities and the information ¯ows between the activities are determined both from interviews and from the analysis of actual documents (digital or on paper) generated during the activities. The interviews are also used to establish information chains. Per activity it is determined who the suppliers of the required information are and for what purpose and by whom the information will be used. Also per activity the main business drivers are determined. In order to validate the thus established information ¯ows people are, per activity, interviewed both on operational and on management level. In those cases where information is not used by any further activity it is identi®ed as an off-process activity. In those cases where information for a certain activity is structurally lacking, this lack is
Fig. 3. Simpli®ed MIR model.
298
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
identi®ed as an information disruption. (For a formal description of MIR models see Ref. [10].) This paper describes the development of such a MIR model for an actual, industrial case [11] and uses the MIR concept to explain (losses in) the transfer of information. 2.2. Case study: setting-up an information ¯ow model for the service- or after sales process The way the service- or after sales process is organised in different companies is rather uniform. The ASP described in this paper is typical for `transportable' consumer electronics. It is a characteristic for this class of products that failing products are being taken to a repair centre. The tasks performed in the ASP are often carried out by separate parties. These are a call centre, one or more repair centres, and sometimes a retailer or a ®rst line support group in the ®rst contact with the customer. The execution of these tasks is sometimes outsourced. The interaction of the parties is shown in Fig. 4. The main question for analysing this structure is to learn how fast and with what level of detail product quality and reliability information from the customer reaches the manufacturer. For that purpose Sections 2.2.1±2.2.5 will explain in more detail the tasks of the various parties as well as the ¯ows between the parties. 2.2.1. Customer The customer is not a part of the ASP. The customer is, however, (currently) the initiator of service-actions. The customer starts the ASP in the case that the performance of the product does not meet the expectations of the performance. The in¯uence of the customer on the results of the ASP cannot be denied. The input of the customer is a critical factor in the performance of the ASP. This input consists of the failed product itself and the possible background information about the failure mode of the product. The quality of the information depends on various factors, such as the skills or background knowledge of the customer, the expectations the customer has of the product, and the
Fig. 4. Organisation structure of the ASP.
idea that the customer has about the operation of the product. 2.2.2. Third party (retailer or internal ®rst line support group) The structure of the ASP with a call centre as a central player leaves no separate role for a retailer or a ®rst line support group. A call centre performs optimally when there is a direct contact with the customer. Nevertheless, the intervention of a third party between the customer and the call centre is not unusual. The third party, either the retailer or ®rst line support group, does the intake of the product. Sometimes the product is also pre-tested before the product is passed on to the repair centre. The failure data is passed on to the call centre. Although these parties have a marginal task in the ASP, their in¯uence is large. This will be explained later in this paper. 2.2.3. Call centre The main task of the call centre is to ®lter the soft failures (failures that can be resolved remotely such as problems with the operating instructions) from the stream of calls and to solve those failures directly. The products of customers who have a product that apparently has a hard failure (failures that (may) require an intervention in the hardware) are dispatched to the repair centre. The call centre agent is likely to be the ®rst person of the after sales organisation that has contact with the customer. Gathering of failure information is therefore a task that should be done by the call centre, particularly where it concerns failures due to customer expectations or behaviour. Apart from the failure information provided by the customers and the failure information sent to the repair centres, there is another ¯ow of information: the product information from the manufacturer that enables the call centre agents to help their customers as good as possible. 2.2.4. Repair centre The task of the repair centres is to repair the failed products that the helpdesk judged to have a hard failure. Repair centres are commonly responsible for the swap of the product at the customer or the third party. The repair centres have to meet the targets on turn around times of repairs so that the promises done in the warranty can be kept. The processes at the repair centres are therefore primary logistical processes. The repair centres are the prime source of quality and reliability feedback where it concerns information about the technical behaviour of the product (not necessarily under user conditions). They receive the failure information from the call centre and product and repair information from the manufacturer. Their output consists of repaired products and of information to the manufacturer (so-called failure information).
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
2.2.5. The manufacturer The manufacturer of the products can be described as the owner of the ASP. This owner has several tasks. Dependent on the extent to which the different tasks are outsourced, the tasks of the manufacturer are: providing the parties involved with the needed technical product information and hardware payment for services performed to the other parties deployment of information within the own organisation trouble shooting when unexpected failures occur. 3. Developing a MIR model The above paragraphs have outlined a service (or after sales) structure where many parties interact for different purposes. The main question, addressed in this paper, is whether this complex structure of interrelated activities will be able to generate adequate reliability information for the people involved in the development of (future) products. In order to answer this question Sections 3.1 and 3.2 will develop a MIR information ¯ow model where the structure of the service process, the resulting information ¯ows and the control factors steering this information are used to better understand information exchange in the ASP. 3.1. Information ¯ow(s) and control factors The information ¯ow in the ASP is not just one ¯ow of information that starts at the customer and ends at the manufacturer. Information is (ex)changed at every contact between the different parties. What information is exchanged with what speed and what level of detail will, to a large extend, be determined by a number of so-called `control factors'. The purpose of many companies is to achieve and increase pro®tability. The after sales process can contribute to pro®tability in three different manners. ² The reduction of repair costs: One way to improve pro®tability using the after sales process is to reduce the direct `out of pocket' expenses required to operate the ASP. In those cases where this does not affect customer satisfaction or the quality of the products it is the fastest way to improve pro®tability. ² The reduction of the unavailability of the product to the customer: Customers will, in most cases, see service to a product as an annoyance. By a very smooth (little disturbance to the customer) and fast-operated ASP the customer's dissatisfaction will be reduced and, via mechanisms such as brand loyalty, the pro®tability of the business process will increase. Because the mechanism operates via slow-working principles such as brandloyalty the improvement will be slower than by direct cost reductions. The service process, however, should neither cause excessive extra costs (see previous
299
mechanism) nor reduce the knowledge on product± customer interaction (see next mechanism) ² The development of products that better ®t the customer needs: Especially with products with a high degree of innovation it can be very well the case that a product is being developed that not fully meets with the customer's expectations. In those cases the ASP also serves as a mechanism for market feedback in order to improve the speci®cation and development process of future products. Via an improved customer satisfaction and, from there, an increasing market share, this mechanism can also in¯uence pro®tability. The mechanism is, time wise, even slower than the previous mechanism because it involves both the mechanism of `new product development' and `brand loyalty' (Fig. 5). As mentioned earlier the three sub-processes (or optimisation loops) are strongly inter-dependent. Optimising one process to the extreme can have adverse effect to the other sub-processes. As the three processes have different time constants it can be very well the case that a short-term optimal feedback process (fast repairs against minimum costs and therefore minimum generation of information) has adverse effects on long term processes. Table 1 gives an example of extreme optimisation of the different sub-processes. In this industrial case study the different parties involved in the after sales process have chosen to optimise against different, often locally determined, business drivers. This leads to con¯icts in the optimisation of the complete after sales process with, as end result, the fact that the contents of the Q & R feedback does not comply with the requirements. Sections 3.1.1±3.1.4 describe in detail the attitude of the different parties towards the three business drivers. 3.1.1. Third party retailers or internal ®rst line support group When a third party is involved in the after sales process, the main driver of this third party turned out to be in this case product availability from a customer perspective. A third party can either be a retailer or a ®rst line support group. The retailer is keen to help the customer as quickly as possible since the availability of the product can in¯uence the loyalty of this customer. The main driver of ®rst line support groups is availability as a result of agreed maximum downtimes. Providing reliable failure information to the manufacturer is of less importance for these third parties. This can result in extra measures to guarantee a quick settlement of the complaint, like extra in-house product-buffers and swap of the product without proper diagnosis of the failure mechanism. This can cause a considerable loss of information. 3.1.2. Call centre Two types of call centre exist: Call centres whose primary task is to do the intake of the products to repair (thus time/cost driven).
300
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
Fig. 5. Different optimisation strategies in the ASP.
Call centres with the task to gather information (thus information driven). In this case (common in high volume consumer products), the main driver turned out to be time/cost. The ¯ow of complaints about failing products can be pretty large when it concerns mass-produced products and must thus be handled in a cost-ef®cient way to keep the overall costs of service at a reasonable level. A call centre of the ®rst type is a good option to perform this task. This con¯icts, however, with the wish to retrieve useful quality and reliability information from the ®rst contact with the customer. The gathering of the desired information could, theoretically speaking, also be separated from the main ¯ow of handling complaints. This can be done by establishing a call centre of the second type, for a small proportion of the calls. By correlating the information gathered in the
two information ¯ows it could become possible to optimise the two processes simultaneously. Establishing these correlations, however, could be a considerable challenge (one system should be extremely ¯exible in order to handle especially unexpected complaints; the other system should have a strong degree of standardisation in order to operate cost ef®cient) and for that reason the company had decided not to implement it in this case. 3.1.3. Repair centre The repair of products turned out to be mainly a logistical process. Repair centres certainly have to meet their targets with respect to the availability of the repaired products or replacement products, but they can be more ¯exible in this as the third parties described before. Availability is no ground for competition for repair centres as it is for retailers. Repair centres are mainly judged on their time/cost-performance
Table 1 Control factors in¯uencing the service process Control factor
Key performance indicator
Optimisation
Extreme situation
Costs
Overall costs of ASP
Minimise cost of ASP
Logistics
Turnover time per repair
Information on product quality
Number of unknown ®eld problems
Minimise unavailability of product to the customer Minimise number of unknown ®eld problems
Subcontract repair on cheapest price/repair Swap products at customer Detailed analysis of failure cause at the customer
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
(cost per repair). This again con¯icts with the provision of useful feedback of information on the quality and reliability of the product. In order to operate the service process ef®ciently the repair centre maintained a considerable knowledge based on failures in products but this database was created independently without input from or output to the supplier of products. 3.1.4. The owner of the service-process The product manufacturer has in fact an internal con¯ict between cost and availability on one side versus information on the other side. Contracts with the external ASP parties were made by the purchasing department and are strongly driven by price. Requests for additional information from development had to be included in these same contracts via the same purchasing department. The different drivers used by different departments (accounting/purchasing and development) explain that the need for more information in this model automatically lead to an increase in the costs of service or to an increase in the time needed to repair the products. Since the repair time is ®xed by the promises in the warranty, more information implies higher costs. To create an after sales process that is both cost-ef®cient and information-ef®cient, tasks could be split up. The after sales process can be divided in a large organisational force that handles the stream of ®eld returns in a time-ef®cient way and a small force that investigates a small part of the returned products more severely to create the wished Q & R feedback. A different mindset requires different processes and different organisations. An important issue in the organisation of the after sales process is the measurement of the performance of the involved parties. Attention must be given to the creation of a system of performance measurement re¯ecting which of the business drivers is pursued. 3.2. The MIR model Based upon the above information a MIR model was developed for the after sales process. A simpli®ed version of the model is given in Fig. 6. The model includes the main activities, information ¯ows (customer±supplier relations) and the main control factors for the different activities. In the detailed model also dependencies between the control factors were taken into account but these have not been included in Fig. 6. From the MIR model, a number of conclusions could be drawn: ² The information ¯ow from the end customer is taking place via a large number of interrelated activities ² In spite of the large number of activities only limited information reaches the supplier of the product ² Information disruptions (loss off information, strong delay in information) especially takes place at either
²
²
²
²
301
organisational interfaces or when abrupt changes in control factors occur. In some locations in the model high-quality information is available but mostly used for local purposes only. (The example given earlier on the cost driven repair shop can almost literally be found back in the model.) The information, ®nally reaching the supplier, contains little detail (MIR , 1) and takes a long time to reach the supplier (.1.5 year three product development cycles). Given the existing structure this is not likely to improve without a structural change to the process. Although this ASP operates in a strongly innovative market the structure of the ASP is not able to handle innovations very well; most of the activities are locally optimised on short term pro®tability. It can be argued whether structures like this (outsourcing the entire ASP to different third parties) are able to provide, especially in a strongly innovative market, in any manner relevant information in a reasonable amount of time.
Although ®eld information is, as addressed earlier, one of the two vital channels for market information this channel appears not to work very well in the presented case. In order to explain this inadequate performance the MIR model was extended with so-called control factors. Per activity it was determined what the main business drivers are. In the model only three different business drivers are used: cost, time and quality. The business drivers are directly derived from the performance indicators, used in the business process. Fig. 6 gives the simpli®ed MIR model including the main control factors. 4. Conclusions In a situation where product functionality, quality, costs and time to market are under strong pressure it is unlikely that a business process can do without fast and adequate feedback from the market. One of the more common structures for this purpose is via the service- or after sales process. It is questionable whether, without special adaptations, a service process is able to generate the required information in the required time. This paper therefore analyses a typical, industrial, case of an after sales process in order to see whether this process is structurally capable of meeting the stated requirements. For this purpose a socalled MIR model was developed. Using the MIR model it is not only possible to visualise the information ¯ows in a given business process but also to monitor the quality of the information and the propagation delays for this business process. Using the MIR model it was not only possible to determine that the ASP was structurally inadequate for the required fast and high-quality feedback but also that the process was subject to various
302
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
Fig. 6. The MIR model of the after sales process.
forms of sub-optimisation. Especially where parts of the sub-process where subcontracted to third parties, the internal priorities of this third party often caused serious loss of information. The result of operating an ASP like this may be that existing quality and reliability problems remain for a long time undiscovered by the supplier of products. This will, again most likely, be the case to an even larger extend for products with a high degree of innovation.
[4] [5]
[6] [7] [8]
References [9] [1] Brombacher AC, Sander PC. Product reliability and quality of business processes; requirements for developing reliable products. Reliab Engng Syst Safety 2000, Special issue on ESREL'99 conference. [2] Military handbook reliability prediction of electronic equipment (MIL-HDBK-217E), United States Department of Defence, 1987. [3] Geudens W. Philips CFT internal report (the exact title of this report
[10]
[11]
cannot be disclosed due to con®dentiality reasons. Full details, however, are known to the authors), 1996. Brombacher AC. MIR: covering non-technical aspects of IEC 61508 reliability certi®cation. Reliab Engng Syst 1999. Lu Y. Quality and reliability in the PCP of derivative products (working title). PhD Thesis. Eindhoven University of Technology, expected Jan 2002. Molenaar PA. MSc Thesis at Eindhoven University of Technology, 1999. Petkova VT, Sander PC, Brombacher AC. The role of the service centre in the improvement process. Quality Reliab Engng Int 1999. Brombacher AC, De Graef M, Den Ouden E, Minderhoud S, Lu Y. Bedrijfszekerheid van technische systemen bij veranderende bedrijfsprocessen (in Dutch). J Bedrijfskunde 2001. Lu Y, Loh HT, Ibrahim Y, Sander PC, Brombacher AC. Reliability in a time driven product development process. Quality Reliab Engng Int 1999. Berden T, Brombacher AC, Sander PC. The building bricks of product quality: an overview of some basic concepts and principles. Int J Prod Econ 2001. Rijks TJM. MSc Thesis at Eindhoven University of Technology, 1999.
www.elsevier.com/locate/ress
Why do quality and reliability feedback loops not always work in practice: a case study P.A. Molenaar a, A.J.M. Huijben b, D. Bouwhuis c, A.C. Brombacher a,* a
Section Product and Process Quality, Faculty of Technology Management, Eindhoven University of Technology, Building Pav., P.O. Box 513, 5600 MB Eindhoven, The Netherlands b Philips Medical Systems, P.O. Box 90050, 5621 JG Eindhoven, The Netherlands c Faculty Technology Management, Building TEMA., P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Abstract The increasing competition in the market of consumer electronics forces industry to simultaneously improve the functionability, reliability and costs of their products. Due to the strong dynamics of this ®eld an important measure in improving product quality is the feedback of information on actual ®eld behaviour of a product which is extremely relevant for product development. This information, usually gathered by the service department, can help development to learn from the failures of earlier products and to improve the next product generations. This feedback process, however, does not always bring the expected bene®ts. This paper presents a model, based on an actual case study, of the structure of the feedback process in terms of information ¯ows. The information ¯ow model describes the chain or network of activities in the feedback process as a function of (different) drivers such as cost, time or quality. Using this information ¯ow model it is possible to explain how quality and reliability related information propagates through a business process and where and why communication problems lead to disruptions in the feedback of this information. With this model, it is easier to understand the function of different parties in the service or after sales process, so that management of the information in this process becomes more transparent and can be improved. q 2002 Elsevier Science Ltd. All rights reserved. Keywords: Product quality; Product reliability; Customer feedback; Service; After sales process; Information ¯ow; Maturity index on reliability
1. Introduction The market for consumer electronics shows an increasing complexity. Several trends that make the development of new products more dif®cult can be distinguished. Products get more complexity all the time, while quality and reliability demands an increase in products. Changes in warranty legislation result in the fact that manufacturers have a broader responsibility for the behaviour of their products during a longer period. These contradictory terms of increasing product complexity and increasing product liability are accompanied by the need to shorten the development times. This is due to the heavy competition in the market of consumer goods where `time to market' is currently one of the dominant business drivers [1]. Ways to improve quality and reliability of products without increasing the development times have been the subject of numerous research projects over the last decades. In the past, the main focus on product reliability was on the reliability of components * Corresponding author. Tel.: 131-40-247-2390; fax: 131-40-246-7497. E-mail address: [email protected] (A.C. Brombacher).
[2]. Both due to the improvements in component reliability and due to a number of other aspects that will be addressed in this paper component failures no longer dominate product reliability currently [3]. Earlier papers stressed the importance of the reviewing quality and reliability not only of the products but also of the underlying business process [4]. Especially in fast product creation processes (PCPs) the majority of the problems is often caused by lack of information deployment (problems are known in the PCP but not at the right moment at the right places) and not by problems that were generally unknown (or accepted) in the PCP [5]. This has led to models that use the analysis of information ¯ows in the PCP for analysis and improvement of the quality and reliability of the product. A schematic overview of the business processes considered in this paper is shown in Fig. 1. In those theoretical cases where product , customer interaction is fully known throughout the process it could be possible to design a PCP without the necessity for a feedback system. In branches such as high-volume consumer products, due to the strong degree of innovation and the strongly dynamic business processes, this will hardly ever
0951-8320/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0951-832 0(01)00117-X
296
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
Fig. 1. Information ¯ow in a product creation process [6].
be the case [7]. Due to combinations of new technology, new customers and new structures used in the development process [8] certain aspects of product quality and reliability will only be found at the customer. In many companies the after sales (or service) channel is one of the two interfaces with a customer. In the marketing department the attitude of the customer towards a future product is anticipated and projected into future products. In the service channel, however, the experiences of real customers with real products are available, especially where dissatisfaction is involved. Especially with strongly innovative products this information may have considerable differences with the information anticipated by the marketing department. Information on the root causes of these failures, therefore, must be deployed to all relevant parties involved in the creation of the product. The structure of this feedback process can to a large extend determine, as this paper will show, the company's (in)ability to prevent problems in the future. This paper focuses on an information ¯ow model of the after sales process (ASP) and shows why in many cases the feedback process is not able to generate the required information, relevant for other parts of the development process. To model this structure the so-called `Maturity Index on Reliability' (MIR) method was used. Section 2 will ®rst explain the MIR model and will then apply the MIR model to the service or ASP of an actual industrial case study. 2. Analysing the information ¯ow for the service- or after sales process In order to model the quality of feedback structures in a business process, it will be necessary to distinguish the separate ¯ows in the process. In a business process, such as the development and realisation of high-volume consumer
products two main ¯ows and a set of control factors can be distinguished: ² Physical ¯ow: The transformation process of ideas and (raw) materials into a working product on the market (feedforward) and of malfunctioning products back to the supplier (feedback). ² Information ¯ow: Information on the above products with respect to function, cost, time and quality. ² Control factors: Factors that either control (enable/ disable) the physical ¯ow or the information ¯ow. Traditionally the main emphasis in quality control has been on the control of the quality of goods: the physical ¯ow. This paper will take a different viewpoint and will look especially into the information ¯ow and will develop a model to explain product quality in terms of the quality of these information ¯ows. It will focus especially on disruptions in information ¯ows in order to explain (lack of) product quality. 2.1. Analysing information ¯ows in business processes The availability of information in one part of a business process by no means guarantees that this information reaches the right people at the right moment with the right level of detail [9]. The transfer of information will require certain activities, each of which are enabled or facilitated by certain enabling conditions or business drivers (Fig. 2). A major problem in analysing information ¯ows is the socalled `quality of information' [4]. Although the statement `the product does not work' may seem as a valid statement from a customer's perspective this information is, in most of the cases, useless for engineers involved in the development of future products. In order to determine whether the right information reaches the right person at the right moment in time a so-called
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
297
Fig. 2. Information ¯ow model of service process.
maturity index on reliability (MIR) model was developed. This MIR model uses a scale of four qualitative levels to describe the quality of information in an information ¯ow model: ² MIR level 1 (How much): Quantitative information on failures is available to the relevant persons ² MIR level 2 (Where): The relevant actors, with the ability to in¯uence/control the mentioned failures can be identi®ed and have access to this information ² MIR level 3: (Why): The technical and/or organisational root-cause of the failure can be determined and controlled by the mentioned actors. ² MIR level 4: (What to do): Adequate measures to prevent re-occurrence of the failure can be identi®ed and implemented. An example of a strongly simpli®ed MIR model is given in Fig. 3. It presents a strongly cost driven repair shop. In order to operate repair activities ef®ciently, information on causes and remedies (MIR3) is stored for internal use but no structural feedback to the manufacturer takes place on the causes of failures and the possibilities to prevent them in future products. The information exchange to the supplier concentrates only on ®nancial aspects (number of repairs to
be reimbursed; MIR 1). Therefore failures most likely will re-appear in future products. It can also be argued whether the repair strategies adopted by the repair shop (with limited detailed knowledge of the product) will result in optimal repair strategies, especially when products with a high degree of innovation are involved. A full MIR information ¯ow model is not based on the analysis of a single activity but on the analysis of a network of interrelated quality and reliability activities Ð and the information exchange between these activities Ð in a given business process. The main activities and the information ¯ows between the activities are determined both from interviews and from the analysis of actual documents (digital or on paper) generated during the activities. The interviews are also used to establish information chains. Per activity it is determined who the suppliers of the required information are and for what purpose and by whom the information will be used. Also per activity the main business drivers are determined. In order to validate the thus established information ¯ows people are, per activity, interviewed both on operational and on management level. In those cases where information is not used by any further activity it is identi®ed as an off-process activity. In those cases where information for a certain activity is structurally lacking, this lack is
Fig. 3. Simpli®ed MIR model.
298
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
identi®ed as an information disruption. (For a formal description of MIR models see Ref. [10].) This paper describes the development of such a MIR model for an actual, industrial case [11] and uses the MIR concept to explain (losses in) the transfer of information. 2.2. Case study: setting-up an information ¯ow model for the service- or after sales process The way the service- or after sales process is organised in different companies is rather uniform. The ASP described in this paper is typical for `transportable' consumer electronics. It is a characteristic for this class of products that failing products are being taken to a repair centre. The tasks performed in the ASP are often carried out by separate parties. These are a call centre, one or more repair centres, and sometimes a retailer or a ®rst line support group in the ®rst contact with the customer. The execution of these tasks is sometimes outsourced. The interaction of the parties is shown in Fig. 4. The main question for analysing this structure is to learn how fast and with what level of detail product quality and reliability information from the customer reaches the manufacturer. For that purpose Sections 2.2.1±2.2.5 will explain in more detail the tasks of the various parties as well as the ¯ows between the parties. 2.2.1. Customer The customer is not a part of the ASP. The customer is, however, (currently) the initiator of service-actions. The customer starts the ASP in the case that the performance of the product does not meet the expectations of the performance. The in¯uence of the customer on the results of the ASP cannot be denied. The input of the customer is a critical factor in the performance of the ASP. This input consists of the failed product itself and the possible background information about the failure mode of the product. The quality of the information depends on various factors, such as the skills or background knowledge of the customer, the expectations the customer has of the product, and the
Fig. 4. Organisation structure of the ASP.
idea that the customer has about the operation of the product. 2.2.2. Third party (retailer or internal ®rst line support group) The structure of the ASP with a call centre as a central player leaves no separate role for a retailer or a ®rst line support group. A call centre performs optimally when there is a direct contact with the customer. Nevertheless, the intervention of a third party between the customer and the call centre is not unusual. The third party, either the retailer or ®rst line support group, does the intake of the product. Sometimes the product is also pre-tested before the product is passed on to the repair centre. The failure data is passed on to the call centre. Although these parties have a marginal task in the ASP, their in¯uence is large. This will be explained later in this paper. 2.2.3. Call centre The main task of the call centre is to ®lter the soft failures (failures that can be resolved remotely such as problems with the operating instructions) from the stream of calls and to solve those failures directly. The products of customers who have a product that apparently has a hard failure (failures that (may) require an intervention in the hardware) are dispatched to the repair centre. The call centre agent is likely to be the ®rst person of the after sales organisation that has contact with the customer. Gathering of failure information is therefore a task that should be done by the call centre, particularly where it concerns failures due to customer expectations or behaviour. Apart from the failure information provided by the customers and the failure information sent to the repair centres, there is another ¯ow of information: the product information from the manufacturer that enables the call centre agents to help their customers as good as possible. 2.2.4. Repair centre The task of the repair centres is to repair the failed products that the helpdesk judged to have a hard failure. Repair centres are commonly responsible for the swap of the product at the customer or the third party. The repair centres have to meet the targets on turn around times of repairs so that the promises done in the warranty can be kept. The processes at the repair centres are therefore primary logistical processes. The repair centres are the prime source of quality and reliability feedback where it concerns information about the technical behaviour of the product (not necessarily under user conditions). They receive the failure information from the call centre and product and repair information from the manufacturer. Their output consists of repaired products and of information to the manufacturer (so-called failure information).
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
2.2.5. The manufacturer The manufacturer of the products can be described as the owner of the ASP. This owner has several tasks. Dependent on the extent to which the different tasks are outsourced, the tasks of the manufacturer are: providing the parties involved with the needed technical product information and hardware payment for services performed to the other parties deployment of information within the own organisation trouble shooting when unexpected failures occur. 3. Developing a MIR model The above paragraphs have outlined a service (or after sales) structure where many parties interact for different purposes. The main question, addressed in this paper, is whether this complex structure of interrelated activities will be able to generate adequate reliability information for the people involved in the development of (future) products. In order to answer this question Sections 3.1 and 3.2 will develop a MIR information ¯ow model where the structure of the service process, the resulting information ¯ows and the control factors steering this information are used to better understand information exchange in the ASP. 3.1. Information ¯ow(s) and control factors The information ¯ow in the ASP is not just one ¯ow of information that starts at the customer and ends at the manufacturer. Information is (ex)changed at every contact between the different parties. What information is exchanged with what speed and what level of detail will, to a large extend, be determined by a number of so-called `control factors'. The purpose of many companies is to achieve and increase pro®tability. The after sales process can contribute to pro®tability in three different manners. ² The reduction of repair costs: One way to improve pro®tability using the after sales process is to reduce the direct `out of pocket' expenses required to operate the ASP. In those cases where this does not affect customer satisfaction or the quality of the products it is the fastest way to improve pro®tability. ² The reduction of the unavailability of the product to the customer: Customers will, in most cases, see service to a product as an annoyance. By a very smooth (little disturbance to the customer) and fast-operated ASP the customer's dissatisfaction will be reduced and, via mechanisms such as brand loyalty, the pro®tability of the business process will increase. Because the mechanism operates via slow-working principles such as brandloyalty the improvement will be slower than by direct cost reductions. The service process, however, should neither cause excessive extra costs (see previous
299
mechanism) nor reduce the knowledge on product± customer interaction (see next mechanism) ² The development of products that better ®t the customer needs: Especially with products with a high degree of innovation it can be very well the case that a product is being developed that not fully meets with the customer's expectations. In those cases the ASP also serves as a mechanism for market feedback in order to improve the speci®cation and development process of future products. Via an improved customer satisfaction and, from there, an increasing market share, this mechanism can also in¯uence pro®tability. The mechanism is, time wise, even slower than the previous mechanism because it involves both the mechanism of `new product development' and `brand loyalty' (Fig. 5). As mentioned earlier the three sub-processes (or optimisation loops) are strongly inter-dependent. Optimising one process to the extreme can have adverse effect to the other sub-processes. As the three processes have different time constants it can be very well the case that a short-term optimal feedback process (fast repairs against minimum costs and therefore minimum generation of information) has adverse effects on long term processes. Table 1 gives an example of extreme optimisation of the different sub-processes. In this industrial case study the different parties involved in the after sales process have chosen to optimise against different, often locally determined, business drivers. This leads to con¯icts in the optimisation of the complete after sales process with, as end result, the fact that the contents of the Q & R feedback does not comply with the requirements. Sections 3.1.1±3.1.4 describe in detail the attitude of the different parties towards the three business drivers. 3.1.1. Third party retailers or internal ®rst line support group When a third party is involved in the after sales process, the main driver of this third party turned out to be in this case product availability from a customer perspective. A third party can either be a retailer or a ®rst line support group. The retailer is keen to help the customer as quickly as possible since the availability of the product can in¯uence the loyalty of this customer. The main driver of ®rst line support groups is availability as a result of agreed maximum downtimes. Providing reliable failure information to the manufacturer is of less importance for these third parties. This can result in extra measures to guarantee a quick settlement of the complaint, like extra in-house product-buffers and swap of the product without proper diagnosis of the failure mechanism. This can cause a considerable loss of information. 3.1.2. Call centre Two types of call centre exist: Call centres whose primary task is to do the intake of the products to repair (thus time/cost driven).
300
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
Fig. 5. Different optimisation strategies in the ASP.
Call centres with the task to gather information (thus information driven). In this case (common in high volume consumer products), the main driver turned out to be time/cost. The ¯ow of complaints about failing products can be pretty large when it concerns mass-produced products and must thus be handled in a cost-ef®cient way to keep the overall costs of service at a reasonable level. A call centre of the ®rst type is a good option to perform this task. This con¯icts, however, with the wish to retrieve useful quality and reliability information from the ®rst contact with the customer. The gathering of the desired information could, theoretically speaking, also be separated from the main ¯ow of handling complaints. This can be done by establishing a call centre of the second type, for a small proportion of the calls. By correlating the information gathered in the
two information ¯ows it could become possible to optimise the two processes simultaneously. Establishing these correlations, however, could be a considerable challenge (one system should be extremely ¯exible in order to handle especially unexpected complaints; the other system should have a strong degree of standardisation in order to operate cost ef®cient) and for that reason the company had decided not to implement it in this case. 3.1.3. Repair centre The repair of products turned out to be mainly a logistical process. Repair centres certainly have to meet their targets with respect to the availability of the repaired products or replacement products, but they can be more ¯exible in this as the third parties described before. Availability is no ground for competition for repair centres as it is for retailers. Repair centres are mainly judged on their time/cost-performance
Table 1 Control factors in¯uencing the service process Control factor
Key performance indicator
Optimisation
Extreme situation
Costs
Overall costs of ASP
Minimise cost of ASP
Logistics
Turnover time per repair
Information on product quality
Number of unknown ®eld problems
Minimise unavailability of product to the customer Minimise number of unknown ®eld problems
Subcontract repair on cheapest price/repair Swap products at customer Detailed analysis of failure cause at the customer
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
(cost per repair). This again con¯icts with the provision of useful feedback of information on the quality and reliability of the product. In order to operate the service process ef®ciently the repair centre maintained a considerable knowledge based on failures in products but this database was created independently without input from or output to the supplier of products. 3.1.4. The owner of the service-process The product manufacturer has in fact an internal con¯ict between cost and availability on one side versus information on the other side. Contracts with the external ASP parties were made by the purchasing department and are strongly driven by price. Requests for additional information from development had to be included in these same contracts via the same purchasing department. The different drivers used by different departments (accounting/purchasing and development) explain that the need for more information in this model automatically lead to an increase in the costs of service or to an increase in the time needed to repair the products. Since the repair time is ®xed by the promises in the warranty, more information implies higher costs. To create an after sales process that is both cost-ef®cient and information-ef®cient, tasks could be split up. The after sales process can be divided in a large organisational force that handles the stream of ®eld returns in a time-ef®cient way and a small force that investigates a small part of the returned products more severely to create the wished Q & R feedback. A different mindset requires different processes and different organisations. An important issue in the organisation of the after sales process is the measurement of the performance of the involved parties. Attention must be given to the creation of a system of performance measurement re¯ecting which of the business drivers is pursued. 3.2. The MIR model Based upon the above information a MIR model was developed for the after sales process. A simpli®ed version of the model is given in Fig. 6. The model includes the main activities, information ¯ows (customer±supplier relations) and the main control factors for the different activities. In the detailed model also dependencies between the control factors were taken into account but these have not been included in Fig. 6. From the MIR model, a number of conclusions could be drawn: ² The information ¯ow from the end customer is taking place via a large number of interrelated activities ² In spite of the large number of activities only limited information reaches the supplier of the product ² Information disruptions (loss off information, strong delay in information) especially takes place at either
²
²
²
²
301
organisational interfaces or when abrupt changes in control factors occur. In some locations in the model high-quality information is available but mostly used for local purposes only. (The example given earlier on the cost driven repair shop can almost literally be found back in the model.) The information, ®nally reaching the supplier, contains little detail (MIR , 1) and takes a long time to reach the supplier (.1.5 year three product development cycles). Given the existing structure this is not likely to improve without a structural change to the process. Although this ASP operates in a strongly innovative market the structure of the ASP is not able to handle innovations very well; most of the activities are locally optimised on short term pro®tability. It can be argued whether structures like this (outsourcing the entire ASP to different third parties) are able to provide, especially in a strongly innovative market, in any manner relevant information in a reasonable amount of time.
Although ®eld information is, as addressed earlier, one of the two vital channels for market information this channel appears not to work very well in the presented case. In order to explain this inadequate performance the MIR model was extended with so-called control factors. Per activity it was determined what the main business drivers are. In the model only three different business drivers are used: cost, time and quality. The business drivers are directly derived from the performance indicators, used in the business process. Fig. 6 gives the simpli®ed MIR model including the main control factors. 4. Conclusions In a situation where product functionality, quality, costs and time to market are under strong pressure it is unlikely that a business process can do without fast and adequate feedback from the market. One of the more common structures for this purpose is via the service- or after sales process. It is questionable whether, without special adaptations, a service process is able to generate the required information in the required time. This paper therefore analyses a typical, industrial, case of an after sales process in order to see whether this process is structurally capable of meeting the stated requirements. For this purpose a socalled MIR model was developed. Using the MIR model it is not only possible to visualise the information ¯ows in a given business process but also to monitor the quality of the information and the propagation delays for this business process. Using the MIR model it was not only possible to determine that the ASP was structurally inadequate for the required fast and high-quality feedback but also that the process was subject to various
302
P.A. Molenaar et al. / Reliability Engineering and System Safety 75 (2002) 295±302
Fig. 6. The MIR model of the after sales process.
forms of sub-optimisation. Especially where parts of the sub-process where subcontracted to third parties, the internal priorities of this third party often caused serious loss of information. The result of operating an ASP like this may be that existing quality and reliability problems remain for a long time undiscovered by the supplier of products. This will, again most likely, be the case to an even larger extend for products with a high degree of innovation.
[4] [5]
[6] [7] [8]
References [9] [1] Brombacher AC, Sander PC. Product reliability and quality of business processes; requirements for developing reliable products. Reliab Engng Syst Safety 2000, Special issue on ESREL'99 conference. [2] Military handbook reliability prediction of electronic equipment (MIL-HDBK-217E), United States Department of Defence, 1987. [3] Geudens W. Philips CFT internal report (the exact title of this report
[10]
[11]
cannot be disclosed due to con®dentiality reasons. Full details, however, are known to the authors), 1996. Brombacher AC. MIR: covering non-technical aspects of IEC 61508 reliability certi®cation. Reliab Engng Syst 1999. Lu Y. Quality and reliability in the PCP of derivative products (working title). PhD Thesis. Eindhoven University of Technology, expected Jan 2002. Molenaar PA. MSc Thesis at Eindhoven University of Technology, 1999. Petkova VT, Sander PC, Brombacher AC. The role of the service centre in the improvement process. Quality Reliab Engng Int 1999. Brombacher AC, De Graef M, Den Ouden E, Minderhoud S, Lu Y. Bedrijfszekerheid van technische systemen bij veranderende bedrijfsprocessen (in Dutch). J Bedrijfskunde 2001. Lu Y, Loh HT, Ibrahim Y, Sander PC, Brombacher AC. Reliability in a time driven product development process. Quality Reliab Engng Int 1999. Berden T, Brombacher AC, Sander PC. The building bricks of product quality: an overview of some basic concepts and principles. Int J Prod Econ 2001. Rijks TJM. MSc Thesis at Eindhoven University of Technology, 1999.