How To Establish A It Outsourcing Contract

Published Date: 02 Nov 2017

According to Hodosi (2010) a previous Computer and Systems Sciences lecturer at Stockholm University, Information technology outsourcing is not without problems despite the steep growth trend. The outsourcing of IT has not always been prosperous in terms of economical savings or IT-services stipulated. A SLA acts as a supplement to the main contract defining the agreed services, opening hours and response times between two parties in an outsourcing agreement. According to Wang et al (2008) many firms are steadily increasing their IT outsourcing operations in order to increase value for customers and profit for the company. Furthermore Hodosi (2010) claims that all IT-outsourcing (ITO) has not been successful and that service quality and relationship quality are main reasons for ending contracts and partnerships. Cong & Chau (2010) states similar opinions regarding the outsourcing deals that are a failure and emphasize the lack of guidance when outsourcing IT-services.

A later study conducted by Cong & Chau (2012) claims that relationship and contracting management plays an important role in order to obtain successful factors when outsourcing IT. A study conducted by Goo, Huang & Hart (2008), the foundation, governance and change SLA characteristics were aligned with intended outsourcing objectives which were gathered from select informants using a survey. The informants answered the survey through an organizational or group setting, instead of expressing their own opinion and answering through their personal perception. The study is based on the fact that mechanisms used to create SLAs are poorly understood which results in insufficient or complex contracts which is proven ineffective. The study concludes that a well-structured SLA is important to reach the desired outcome of outsourcing. Furthermore the authors state that barely any SLAs have all possible outcomes defined in detail as it would be too complex or result in increased costs.

In a similar study (Goo & Huang, 2008), the authors state that outsourcing contracts are often constructed through mutual trust rather than written terms. The authors claim that it is due to difficulty in foreseeing all possible outcomes of a service or the possibility of distrust between the parties. This study further confirms the importance of a well-structured SLA and its positive effects on relationships. Goo, Kishore, Rao & Nam (2009), confirms that SLAs have a positive effect on the relationship when they are well structured. The authors claim that the service provider´s performance can be measured using well-structured SLAs. The authors claim that this would develop partnership-style relationships with high levels of trust and commitment. The end result would be the contribution to effective management of outsourcing agreements, the authors claim.

According to Beaumont (2006) comprehensive SLAs contribute to beneficial relationships between business partners when subcontracting information technology. Beaumont claims that a contract containing all of the details concerning changes in a complex IT application would end up being too long. Beaumont points out that the client may have to trust the vendor with their professionalism and their will to uphold their reputation on the market.

In a study by Rouse (2005), a survey was sent out to organizations on the Australian market where detailed SLAs displayed improved cost and service outcomes. The study also showed that outsourcing success was achieved through frequent renegotiation of service levels between the client and the vendor. Rouse (2005) further suggests detailed service descriptions and service levels or standards expected by the client. The author’s pilot study suggests that defining all of the details surrounding an IT service would lead to complex, costly and time consuming process. The process would probably require external assistance and would lead to an excessive amount of information which might affect important aspects of the service, the author states.

Rouse (2005) came to the conclusion that there were no statistical differences between organizations who already had a formal way of renegotiating their service levels and those who did not. The author suggests that this might be due to the level of satisfaction for the client as the formal renegotiation process might not suit their needs. The author claims that organizations which choose the formal way of renegotiating service levels would keep doing so since the benefits results in outsourcing success. Other organizations would use the formal renegotiation process to regain control, the author states. The author however states that further research should be done to support the formal renegotiation theory. The statistical difference in the study by Rouse (2005) was visible between the organizations who met frequently to renegotiate service levels and the ones choosing not to. Organizations choosing not to meet frequently with their clients did not experience the same positive outcomes.

Previous studies discusses the level of detail and complexity in SLAs. A comparison between two types of service level agreements with different level of detail would provide valuable insights to ITO decision makers. We define these types as standardized and detailed SLAs. We want to contribute with significant information to previous work made within the research area, IT-management. Previous studies suggest well-structured SLAs without it becoming detailed while Rouse advocates a detailed approach. This thesis intends to examine whether the level of detail can be achieved from a well-structured SLA while maintaining a healthy relationship. This will be done by suggesting the appropriate type of SLA to choose for this purpose.

The Vendor is one of the leading IT-consulting firms in Sweden specialized in IT-security, infrastructure and applications. Providing services to banks, governments, telecom operators, insurance, and IT companies. The main service areas are implementation, maintenance, support and security. As requested by the Vendor, detailed information will not be provided as the Vendors participation relies strictly on confidentially.

The Client is one of the leading telecom operators in Sweden. They are providing services covering several important IT areas. As requested by the Client, detailed information will not be provided since the Clients participation relies strictly on confidentially.

2. 1 Establishing the contract

2.1.1 Vendor approach

The process of establishing a contract between the Vendor and their customers can be divided into two stages. Firstly, the account manager holding future or present responsibility over the customer initiates the relationship with a product display and presents possible solutions to the customer’s needs. The customer chooses a solution whereupon the account manager introduces several additional support packages and their respective service levels available in relation to the solution selected by the customer. The account manager then creates a quotation of the selected solution with its belonging packages and sends it to the customer.

This initiates the second stage where the customer reviews the quotation and brings forth any parts they would like to renegotiate with the account manager. A new quotation, based on whether renegotiations took place, is then sent from the account manager to the customer for review where both parties come to an agreement. The quotation and the support and services agreement contract is then passed to the respective organizational managing director who signs the contracts. There are different types of contracts. The quotation is a contract containing the products and licenses offered through the solution.

There’s a separate support and services agreement contract (appendix 2) containing information regarding:

The purpose of the agreement

Scope of delivery

The customer’s responsibility towards the Vendor

A list of appendices

The duration of the agreement

Notice of termination and cancellation

Financial terms

Additional notes

Signature fields used by both parties.

The appendices referred to in the support and services agreement contract consists of terms and conditions, the SLA (appendix 3) and the customers operating environment (appendix 4).

The quotation and the appendix, operating environment referred to in the support and services agreement contract, both consist of products and licenses included. The difference is that the quotation contains product and license prices with signature fields for both parties. The product specification referred to post-contracting would therefore be the operating environment appendix and not the original quotation as the operating environment appendix serves as a reference to what the support and services agreement covers. The terms and conditions appendix is a standardized set of terms and conditions applied when establishing contracts about IT-related services. The SLA appendix contains a document description, the agreed service levels and definitions of working days, public holidays and major public holidays.

2.1.2 Contracting

Many organizations do not invest the time and resources, nor have the specific skills, to conduct a detailed due diligence and objective organization assessment before engaging in a relationship controlled by a contract between the two parties. This usually results in an incomplete contract and product requirements which causes conflicts between the client and the vendor resulting in a bad relationship. An incomplete contract with insufficient product requirements would end in expectations not being met, excessive change management and organizational disputes. (Power et al., 2006). According to Järveläinen (2012) contracting is used to maintain a balance between power relationships, in order to better achieve common goals were both parties profit from the business relationship. Goo (2008) states that the increase in IT-outsourcing reflects focus on the contracting stages aswell as the SLA, making both parts very important in order to achieve high benchmarks in relational assessments.

To create a successful outsourcing relationship, it is important to understand where you stand today and where you want or need to be in the future. While keeping a contract specific, it has to be comprehensive and cohesive and offer some flexibility which can be a challenge. It is therefore better to include more information in a contract rather than less while still knowing when it’s gone too far. (Power et al., 2006). According to Lacity & Willcocks (2009), two types of experts are recommended during negotiations - a technical expert and a legal expert. A technical expert for measuring baseline services and a legal expert for ensuring that the customer’s requests are documented properly in the contract.

2.1.3 How to establish an IT-outsourcing contract?

Power et al.(2006), describes outsourcing with two words, "out and sourcing". Where sourcing relates to the ability to transfer work, responsibility and decision rights to someone else. The main reasons for outsourcing work are because it’s cheaper, faster and better. According to Power et al.(2006), there are two main interrelated acts in contracting - negotiations and contract management. Negotiations refer to the discussion that occurs between the two parties before signing an agreement. Negotiations most commonly use a timeframe in which both parties have come to an agreement or disagreement. If agreement is not possible both parties will try to find another source which will meet their demands and expectations.

If agreement is made, the next phase is the contracting management. In this stage, all discussions regarding material, services, support and responsibility is written down into a legal document that connects both the vendor and client. Power et al.(2006), claims that if the agreement is not put into writing there is a great chance of confusion between the vendor and the client. The contract represents the most crucial part of the outsourcing relationship. This is due to that if problems occur, the binding legal agreement would be the first document both parties audits in order to clarify misunderstandings and solve the problems that serves as a starting point for a dispute.

Both stages are of great importance and closely interrelated to one another. If negotiation skills are poor, it will likely result in an agreement that is not of economic advantage. However if failure during the contracting stages arise, the contract will lack the core essence of the negotiation. Furthermore Power et al.(2006), states that relationship management is of great importance when contracting IT-outsourcing. Failure to acknowledge the different aspects in relationship management might have devastating effects. This will simply lead to a deal that is of disadvantage for both parties.

2.2 The contract

A contract can either be legal agreement or normative agreements were their main purpose is to define the matter of exchange in adequate detail (Mooi & Ghosh, 2007). According to Kendrick (2009) the contract represents the legal agreement between two parties in a business relationship. The contract is a legal framework defining the management, operation, services and responsibilities for the two parties. According to Lee (1996), good contracts are the main key for successful relationships when outsourcing IT.

Kendrick (2009) states that the contract has the following functions:

It works as the written agreement of negotiations that has been agreed upon

It works as the potential risk management tool to resolve disputes and avoiding problems.

It works as the framework that has been accepted and concluded by two parties in a business relationship.

2.2.1 Contract structure

Brown (2005) claims that the contract contains details of both parties related to the business relationship and general details regarding services to be supplied.

Furthermore the contract contains Vendors costs and additional charges and the time frame of the contract. Legal, liability and jurisdictional terms is also provided in the contract according to Brown (2005).

2.2.2 Contract duration

The duration of the contract may be for a fixated or continuous period of time. It may also be renewable or ending depending on the agreed duration and agreed terms. In case of violation of the contract it may be terminated by mutual consent of the two parties involved in the business relationship. Kendrick (2009), states that the contract duration should be carefully considered by both parties. The general contract varies from 1-5 years depending on the service provided but may also be longer or shorter. Kendrick (2009), claims that longer duration of the contract most commonly results in more secure and healthy relationship between the parties involved since all are committed but may also result in disputes and limit the parties. Reasons for termination of a contract may be of various make-ups but mainly because the client is not satisfied with the contract.

2.2.3 Service level agreement

According to Brown (2005), the SLA can be defined as the agreement between the two parties regarding the scope of the project, the levels of service and the supplier’s service delivery charges. The SLA is an appendix to the main contract. Kendrick (2009), states that there are three main areas, service levels, scoping and pricing. Practitioners state that the key to successful IT-outsourcing relationships is through formal and comprehensive SLAs. These SLAs specifies different facets of the service to be provided at certain levels to meet requirements (Goo et al., 2009).

Service levels

Kendrick (2009), claims that the main intention of the SLA is to determine the commitments concerning the performance of the main contract. He states that the SLA is a legal document that specifies details on the agreed service levels. According to Kendrick (2009), Andrew Giverin suggests three service levels, continuous, event-based, and sampling measurement in order to accurately measure the SLA. Furthermore the SLA is suggested to be all-inclusive; this means that if the SLA is violated in any way there will be penalties in relation to the violations.

Scope

Kendrick (2009), states that the scope is the quality of service that is provided. Furthermore he claims that it is of great importance that the scope is clear and precise in order to avoid misunderstandings during the outsourcing project.

Pricing

The pricing must be accurate and certain or else it will defeat the purpose of the SLA. The pricing structure must also be reviewed regularly due to changes that might occur in market and future. Kendrick (2009), states that the pricing level must be directly interrelated to the performance.

Contract exit

According to Kendrick (2009) the ending of an outsourcing contract demands numeral factors to be taken into consideration. Should the contract be extended, terminated or renegotiated.

Extended

If the relation between the vendor and the customer are good the contract will most likely be extended. In case that both parties are satisfied with the business relationship no further negotiation is needed Kendrick (2009).

Terminated

If agreement is not reached or the customer selects another partner the contract will be terminated (Kendrick 2009). The contract will most likely be terminated if the relationship between the vendor and customer is bad, damaged or if the customer has found another vendor that is more suitable for their needs.

Renegotiated

In case the contract is in need for modification or adjustments, the renegotiation of the original contract might arise. This might happen during the contract period or after the contract period has expired (Kendrick 2009)

2.2.4 SLA framework

In regards to the service, there might be one or more SLAs. The SLA should be simple and easy to understand. According to Kendrick (2009), the SLA in its simplicity should be able to be read and understood by several of people including managers and user groups. Complicated terms should therefore be avoided in the SLA framework. Since SLAs are different depending on the outsourcing projects, there should not be a standard framework for how the SLA is designed in its whole.

Kendrick (2009) claims that the SLA contains the following sections:

Introduction

This section covers the name of the involved parties, a short explanation of the service and the length of the SLA.

Service

This section provides information of the service, responsibilities and the duration of the contract.

Security

This section provides information regarding the measures on securing electronic data and agreement on current industry standards.

Performance

This section covers agreed performances containing performance reviews, performance integration and resolution procedures.

Costs and charges

This section covers the process of payment and charges, terms of agreement regarding costs and payment, additional brief information of charges and penalties if violations are made.

Dispute resolution

This section covers thorough resolution in case of dispute and penalties in case the supplier does not deliver agreed terms.

2.2.5 SLA type definitions

There is currently no standard regarding the structure and contents of a SLA. There is however common practices used to create them. We list our definitions of standardized and detailed SLA’s below to clarify the meaning of the two concepts.

2.2.5.1 Standardized service Level agreement

A standardized SLA (Appendix 3) is something we categorize as a corporation wide template used for services the corporation provides for their clients. The template is used as a foundation when the SLA is negotiated between the corporation and their clients. The template can be modified according to the client’s needs or requests during negotiations. It can also be modified to better suit the type of service the client is buying.

2.2.5.2 Detailed service level agreement

We categorize the detailed service level agreement in different ways. The detailed SLA could be a standardized type of SLA with tailored sections containing sample errors and problems for a specific type of product or service, additional service level descriptions, extended requirements and information about financial penalties. The detailed SLA could also be completely tailored for one client. Everything included in the SLA is thus adapted to a certain environment or service bought by a client. We have included an extract of Verizon’s SLA for IP VPN services (appendix 8) to demonstrate how definitions of terms can differentiate a standardized SLA from a detailed SLA.

3.1 Research Strategy

According to Bryman (2011) there are two contrasting research strategies, deductive and inductive theory. A deductive approach means that the researcher based the study in theory, create a hypothesis on its origin, tested the hypothesis against the current situation in order to confirm or reject the hypothesis. However an inductive approach means that the researcher based the study on the results of their observations or findings and only then develops a theory. Case studies are most commonly but not necessarily associated with a qualitative approach since these reflect an intensive and detailed examination of a specific case (Bryman 2011). For this study a deductive approach would mean that a theory was generated based on previous research.

The lack of previous research and theories in the area of comparing detailed or standardized SLAs when outsourcing IT-services to maintain healthy relationships, an inductive approach would be more suitable for the study conducted. Furthermore Bryman (2011) describes two ontological orientations connected to quantitative and qualitative research methods, objectivism and constructionism. Objectivism is defined as an ontological position that implies social phenomena and their meanings have an existence independent of social actors. While constructionism is an ontological position that implies that social phenomena and their meaning is something that social actors continuously may position. This means that social phenomena and categories are created through social interactions but also are in a state of constant change.

Objectivism is related to a quantitative research approach which is theory testing while constructivism is related to a qualitative research approach and is theory generating (Bryman, 2011). As mentioned previously, the lack of theories relating to the conducted study, constructivism would therefor provide the proper ontological position for this examination.

3.2 Choice of Method

Bryman (2011) claims that there are two main types of research methods, quantitative and qualitative. Bryman (2011) states that the main difference between the two is that quantitative methods are based on data that can be quantified, meaning data which can be calculated and compared statistically or scientifically. Qualitative methods however are based on the emphasis on words and how individuals interpret its social surroundings. The benefit of using a quantitative method for this study would be faster and wider response covering a larger area. Simply by increasing the number of respondents, the results would thereby have greater statistical validity. Another advantage of quantitative method is that the results can be presented in tables and graphs, furthermore compared and repeated in similar or future studies. The disadvantage is that the method is not good at measuring deeper relation.

With a qualitative approach, it would be harder to reach as many respondents but on the other hand, make it possible to receive in depth insight of which is most significant detailed or standardized SLAs when outsourcing IT-services to provide healthy relationships, which would have been lost in a quantitative study. To achieve success in a quantitative research, the researchers are required to know in advance what questions to use in order to obtain relevant answers. In the study conducted the lack of sufficient knowledge therefor excludes the quantitative method. However in a qualitative study, the respondents themselves have more opportunities to share how they perceive a situation, personal thoughts and expertise related to the specific case (Bryman, 2011).

For this particular examination, a case study is selected as the choice of method, using qualitative semi-structured interviews to obtain data and grounded theory as the analysis method. Examining not only relational status but also comparing SLA types in connection to the outcome of a relationship. It therefor demands both parties’ personal expertise and experiences related to the specific case. Other methods that might have been used to answer the research question is experiments and surveys. The reason for not using experiments is that this would be highly time-consuming demanding a controlled and closed working environment for the results to be succesful. The reason for not using survey as a method is that these are often used in a quantitative manner reflecting a great number of respondents, this would be difficult to achieve since many documents and customers are classified (Bryman, 2011).

3.3 Analytical approach

According to Bryman (2011), grounded theory is the most common approach to analyze data from qualitative research. Initially, each interview is analyzed to find answers relevant to our research question. Secondly, the relevant answers are compared with each other in order to locate a connection regarding the two SLA types. The common approach when using grounded theory based analysis is to begin with the data collection and then form a hypothesis. After the data collection, important factors are extracted from the text and grouped into concepts to make it easy to take in the whole picture. Categories are then formed and used as a base for the theory. SKRIV OM

3.4 Case Study

Qualitative case studies can be described as an intensive, holistic description and analysis of our selected firm (the Vendor). Case study involves examining one or few objects in a variety of ways. Bryman (2011) states that case studies are common in small-scale studies. Bryman (2011) furthermore claims that the spotlight on a case study should be pointed at individual units rather than a broad spectrum. A case study is an intensive study and the duration of time might be long or short. In our study the investigation is for a short limited time. Our goal with the case study is to conduct a detailed study of one particular case referred to as the "Firewall" case. From this we analyze the results to answer our research question. This particular case is specific for the firms we examined, but may also be used to study wider cases relating to SLAs within the same research area.

Brymans (2011) description of a representative case is that it can represent a wider and more general category while being applicable to similar cases. Our main intention is that the Firewall case with the participants involved may be a representative case for other firms in similar situations when evaluating which SLA is appropriate given the situation and circumstances. By providing indications of a problem that occurs when outsourcing IT-services with the wrong SLA. Furthermore Bryman (2011) states that there are several types of cases. This case is presented as a representative case. The results might point out that although a single case study approach might be insufficient in order to offer generalizability, it can represent a significant contribution to knowledge by permitting theory building (McIvor, 2010). The results of this study could be used as a reference for other corporations in similar situations.

3.5 Data collection

Data for the study conducted was acquired by using qualitative semi-structured interviews. Interviews are a common qualitative method used to determine cutting conditions, opinions, beliefs. In a semi-structured interview, the interviewer according to Bryman (2011) uses a list, called interview guide, with themes or topics to be addressed during the interview. However, the limits of the interview are flexible and the interviewer has the freedom to customize the questions from the respondent’s answers. The questions do not necessarily come in the same order as the interview guide. It is possible for the interviewer to change or ask new questions during the interview. With regard to the survey question, the semi-structured interviews offer several advantages.

It allows a flexible interview processes, the respondents themselves can explain their thinking process and it becomes possible to capture the essence. The disadvantage of this type of interview is that it is time consuming and places high demands on the ability to process and structure data. Because each interview is unique, it may be difficult to compare the number of interview responses with each other. Despite the drawbacks mentioned, the semi-structured interviews will still be the best method to answer our research question. The method enables the compromise between flexibility and controllability through follow-up questions as necessary to get respondents to convey the sometimes unconscious information that the study requested.

3.5.1 Other methods for data collection

3.5.1.1 Participation observations

Participant observation is a qualitative technique in which researchers, for a relatively long time, are involved in a particular social environment. That is in order to observe the behavior of individuals (Bryman 2011). It is up to the observer to assess to what extent they will participate. The disadvantage of using participant observation in this study would be time-and space-consuming which would make it difficult to find time to reach a sufficient number of respondents in a representative sample within reasonable time limits.

3.5.1.2 Survey open-response

A survey is another method which is advantageously used to reach a larger number of respondents in a shorter period of time. Surveys can be used with both closed and open response. For this study, a survey with response options would be most relevant since the answers that are given need to be analyzed. The disadvantage of questionnaires with open answer is that they do not allow the researcher to clear up misunderstandings or supplementary questions. Just the opportunity to ask supplementary questions is essential to answer our research question, since there are no known studies to base questions on. For this reason it is not an appropriate method for the study we are conducting.

3.5.1.3 Unstructured interviews

According to Bryman (2011), an unstructured interview is less controlled than a semi-structured. The interviewer has at most a PM with themes such as memory support. No complete questions are worded and the respondent usually gets the opportunity to associate freely. The advantage of this in terms of the current investigation is that the interview would be more informal, comparable to a regular conversation. This might make the interviewees less nervous and more likely to make further statements. The disadvantage compared to a semi-structured interview is that it requires considerable experience and skill of the interviewer. Using unstructured interviews would create difficulty in comparing the respondent’s answers to the questions asked, as they would differ widely between interviews.

3.6 Study process flow

4.1 Case description

In October of 2011, the Client experienced major problems with a firewall they had bought from the Vendor in 2010. This problem caused the Client’s IT environment to run on lower performance than usual. This had a significant impact on the performance of the Clients entire IT environment since the firewall holds a central role in keeping the IT environment in operation. Since the product was covered by a support & services agreement between the Vendor and the Client, the Client had the right to send in a support case to the TAC [1] of the Vendor who would handle the case accordingly. The TAC would then try to identify the problem with the description received by the Client. They would solve it using in-house expertise or by forwarding the support case to the manufacturer of the product and have one of their engineers assigned to the case.

Since the TAC could not solve the problem, the Firewall case was forwarded to the manufacturer TAC who assigned one of their own engineers to work with the support case. The cause of the problem was believed to have been determined by the manufacturer engineer whereupon the Vendor assisted the Client in trying to resolve the problem. The proposed solution from the manufacturer did however not solve the case. As the firewall is important to the health of the IT environment, minor problems occurred as a result of the performance decrease. This made the Client send in more support cases to the Vendor TAC.

The case the Client sent in remained unsolved with both the Vendor TAC and the manufacturer TAC. Numerous potential solutions to the problems were tested but were all unsuccessful. The case along with the minor issues related to the firewall problem continued as unsolved for the remainder of the year. During that time, the relationship between the Vendor and the Client was heavily damaged due to the Client’s behavior towards the Vendor and the Vendor not being able to meet the Clients expectations. The Client did not respect the Vendors working hours specified in the support & services agreement and the Clients lack of patience resulted in the Vendor questioning their moral behavior.

The Vendor failed to meet the Clients expectations. The Client wanted the Vendor to use all methods available to have the case solved quicker which the Vendor did not. Since those services were not included in the support & services agreement, both parties had signed approximately a year earlier. This caused a dispute between the Client and the Vendor where the Client believed that the Vendor was not doing everything they could. The Vendor was acting on the basis of the support contract, whereas the Client thought they were as well.

The Client’s firewall can be split up into two parts. One part is active and one is passive. The active part handles the information passed through the IT environment and the passive part acts as backup, if the active part would fail, the previous passive part would take over and be active, thus handle the information passed through the IT environment. In the Client’s case, both parts were in the active state and the reason for this was unknown. The clients firewall was also working near, and sometimes over its maximum, performance wise. The firewall parts are set to run as one active and one passive part. The firewall tried to follow these settings which would randomly result in a restart of one of the parts. This disrupted the IT environment from operating properly. This would cause all of the information previously handled by both parts simultaneously to be handled by one part only. Since both parts together, already worked near its maximum, one part handling all the information by itself would only cause problems.

Because of the negative impact this issue had on the IT environment, the Client was losing income. The information sent from the Client’s customers through the Client’s firewall would not always pass through, as the firewall could not handle the load at all times. The Clients support contract includes on-site support if the Client had a business critical SLA. The business critical SLA is defined as "all of, or major part of the Customer is affected and/or there is a great risk that the Customer will suffer from substantial financial damages and there is not an alternative solution." The Client was assuming that the negative performance impact caused by the firewall problem, was enough to classify the problem as business critical as they were losing income and could not operate as usual. The Vendor’s and the Client’s definition of business critical is there for different even though they signed the same contract. The cause of the firewall problems could not be determined which forced the Vendor to place a consultant on-site to identify why they were having problems.

The consultant came to the conclusion that the hardware the Client was using to host the firewall did not possess sufficient performance to handle the information passed in and out of the IT environment. The solution from the Vendor to the Client was to invest in new hardware to host the firewall.

4.2 Case study discussion

Several contracts established in the past between The Vendor and their customers have created problems resulting in bad relationships. Both the Vendor and Client confirm that a majority of the problems lies within the interpretation of the service level agreements that were negotiated. Our thesis intends to examine and determine which is most significant, detailed or standardized SLAs when IT outsourcing network and security services to maintain healthy relationships. Our study intends to examine a single case referred to as the "Firewall" case which caused bad relationships between the Vendor and Client.

4.3 Literature research

In order to obtain the relevant literature in relation to our research question we used the Stockholm University literature research engine. This search engine provides results from published articles, journals and books from EBSCO, Business Source Premier and many other high ranking and trustworthy databases. Our supervisor also instructed us to contact professors in IT-outsourcing on Stockholm University to receive valuable input and advice regarding our conducted study. Keywords used in our search were, Outsourcing, IT-outsourcing, SLA, relationships, contracting, successful factors, ITO, and IT-decision making. These keywords were mixed and matched together in order to find the most relevant data for our study.

4.4 Pilot studies

Pilot study 1

A pilot study (appendix 1) was conducted in order to find the source of the dispute in the Firewall case between the Vendor and the Client. This was done in order to identify the insufficient part of the contract itself. The Vendor believes that they would be capable of gradually working together with their clients to prevent future disputes if the source of the problem could be identified.

The pilot study was conducted using a qualitative data collection method. Semi-structured interviews were held with 5 respondents from the Vendor and 3 respondents from the Client. The interview guide contained 4 themes. A workplace theme, where the main goal was to focus on what they work with and which type of contracting they were involved in. A personal reflections theme where each respondent was able to freely express their opinions regarding contracting. A relationship theme where the respondents were asked about any previous or current relational problems related to contracts. A contracting theme was also included to ask the respondents about the contents of the contract used between the Vendor and the Client. Only respondents directly involved in the Firewall case or affected by its outcome were interviewed.

Through the interview results and the analysis using grounded theory, we found that the SLA definitions were unclear. This caused misinterpretations between the Vendor and the Client. The Client was not able to interpret the SLA definitions as the Vendor intended. The Firewall case submitted by the Client wasn't resolved according to their desires as a result of interpretation errors. This caused the dispute between both parties as the Vendor considered their actions to be entitled according to the contract they had with the Client.

Pilot study 2

A pilot study (appendix 5) was conducted with one respondent in order to validate the questions from the semi-structured interview guide which was partially inspired by pilot study 1. The questions were developed from a brainstorming session with pilot study 1 in mind involving the criteria’s from Kvale (Bryman, 2011). The interview guide used in the pilot study 2 is divided in two themes in order to obtain appropriate information in order to answer our research question.

4.4 Data selection

Several factors were taken into consideration in the selection of the data collection. Bryman (2011) suggest a methodology which is called target-oriented selection which was used in this case study. This means that the researcher strategically selects the respondents in a targeted manner in order to answer the research question in a appropriate way. The interviews were conducted strictly with respondents involved in the Firewall case.

There were a total of six conducted interviews. All interviews at both the Vendor and Client were taken place at different times during two weeks, interviewing only one respondent at each time. The interviews with the Vendor were conducted at their office in a small type of conference room. The interviews with the Client were conducted at their office in a large conference room. We wrote down the information during the interviews. Using this method, a potential drawback would be the risk of missing out on important information versus using recordings and transcribe the information . Since the interviews were conducted by two interviewers who were able to complement each other, the structure was considered manageable. The interviews started with explaining the purpose of the study.

We asked the respondents if they were comfortable with the area we were in before the interviews started. The respondents were told that they had the opportunity to review the answers before we ended the interviews. The respondents knew that they did not have to answer questions they weren’t comfortable with. The respondents were also told that they had the opportunity to stop the interview at any given time.

Before conducting the interviews we asked the respondents if they were aware of the SLA dispute between the Vendor and the Client. This was to find out if the respondents could provide us with valuable data for the analysis. It would also give us a hint about what opinions they might bring up in the interview and if the case would be one of them. We chose actively not to interview employees who were not involved in the specific case.

Questions

Kvale (2007) describes several important criteria’s when conducting a qualitative interview. The questions can be divided into several units.

Introductory questions - Introductory questions, these questions aim is to receive general information of the respondent.

Follow-up questions - Follow-up questions, they play an important part in letting the respondent elaborate the answer by not simply answering yes.

Probing questions - These are of great importance to receive a higher depth in the answers given by the respondent.

Specifying questions - Very specific questions in order to obtain data that is strictly required, eliminating the risk of receiving redundant information.

Direct questions - These questions are probably best to ask during the interviews end in order to not control the respondents answer.

Indirect questions - Indirect questions are used to obtain the respondents personal opinions regarding the different topics.

Structured questions - These questions are often used in unstructured or semi-structured interviews in order to make the respondent answer in a well-structured manner.

Interpretive questions - These questions might be a little bit vague but lets the respondent give a wider response in order to cover a larger area of the question. Then it is up to the author to interpret the data obtained to find key-parts important to the study conducted.

All of the criteria’s above have been taken into consideration when creating the interview guide, also the comprehensive investigation on why problems occur after the contracting phase has helped create the necessary questions to the interview guide.

Question

Motivation

1: What is your current position?

Knowing who we are talking to and their position is important for us to know. It might explain certain management questions that might arise.

2: Do you work with SLAs?

We want to know how the respondent is affected by the SLA, if they work with it directly or use it in some way in their line of work.

3: How are your SLAs designed today?

This question is used to find out which types of SLAs the respondents are familiar with.

4: What kind of SLAs are requested, what do they look like?

The client and the vendor might have different opinions about the important parts of an SLA which is why we want to know what kind of requests there are. We hope to be able to match these answers with the previous question thus differentiate reality from theory.

5: Could you explain what a standardized SLA looks like?

This question aims to help us understand what a respondent thinks a standardized SLA actually is and if it differs from our own definition.

6: Can a standardized SLA be considered too simple due to lack of information?

Goo, Huang & Hart (2008), Beaumont (2006) and the pilot study done by Rouse(2005) all point towards complexity issues when using detailed SLAs. While a detailed contract would bring complexity issues, we want to know if a standardized SLA could prove to be lacking information. This is further explained by McIvor(2009) who claims that a major reason for the failure of SLAs includes inadequate definitions.

7: Would several standardized SLAs be beneficial to use for different services?

When we conducted a pilot study interview (appendix 5), the respondent told us that his corporation had once used several SLA’s for different services to get a more accurate SLA. This method was beneficial for the clients and his corporation since it made the SLA levels and the severity descriptions much easier to relate to.

8: Do you find it difficult or easy to design standardized SLAs?

Goo, Huang & Hart (2008) states that mechanisms used to create SLAs are not understood correctly which leads to insufficient or complex contracts which proves ineffective. We therefore want to know if there is a problem already in the design phase of a new standardized SLA.

9: What are the pros and cons of standardized SLAs?

To help us solve the problem in this study, the respondents own experiences With SLAs are important as they’re all involved in the Firewall case. Their own observations could bring us valuable information we might not get since we can’t utilize the SLA ourselves to test its effectiveness.

10: Could you explain what a detailed SLA looks like?

This question aims to help us understand what a respondent thinks a detailed SLA is and if it differs from our own definition.

11: Are there any predefined scenarios in a detailed SLA and how are they designed?

In one of our previous pilot studies (appendix 5), the respondent told us that his previous employer wrote an additional appendix for their SLAs. This appendix covered common problems and errors a client might’ve experienced in order to help them decide which service level their problems or errors were related to. According to the respondent, this was appreciated by the clients.

12: Can a detailed SLA be considered too complex due to excessive amounts information?

Goo, Huang & Hart (2008), Beaumont (2006) and the pilot study done by Rouse(2005) all point towards complexity issues when using detailed SLAs. We want to know if this statement could be considered correct in our firewall case as well by asking the respondents to share their experiences. This question might help us to decide whether or not a detailed SLA is a good solution.

13: Do you find it difficult or easy to design detailed SLAs?

Goo, Huang & Hart (2008) states that mechanisms used to create SLAs are not understood correctly which leads to insufficient or complex contracts which proves ineffective. We therefore want to know if there is a problem already in the design phase of a new detailed SLA.

14: What are the pros and cons of detailed SLAs?

To help us solve the problem in this study, the respondents own experiences With SLAs are important as they’re all involved in the Firewall case. Their own observations could bring us valuable information we might not get since we can’t utilize the SLA ourselves to test its effectiveness.

15: Could you explain what a structured SLA looks like?

Having well-structured SLAs is something authors talk positively about as seen in chapter 1.1. We want to know what structured means to our respondents. We want to find the best SLA solution while keeping it well-structured which makes it important for us to know how our respondents define structured.

16: Would it be possible to combine a standardized and a detailed SLA while keeping it structured and comprehensive?

We would like to avoid the issue where the SLA type you chose is dependent on which type of service it is supposed to regulate. We hope to do this by possibly combining the best of both worlds while using one type of SLA as a base. The respondent’s opinions regarding this matter are important since they might have real world experience of this.

17: Does your company work to improve the SLAs?

We find this question important as the SLA is a document that needs to be up to date as pointed out by the respondent in our pilot study (appendix 5). It might give us an idea of how the SLA is managed and how the respondents corporations handles complaints etc.

18: Do you work to improve the SLAs?

In reference to the previous motivation, we also want to know what the respondents themselves do to improve the SLA. It gives us perspective over the way information is passed within the organization and if the information arrives in the hand of someone who can make a difference.

19: What expectations do you have on your clients/outsourcing partners?

If we know what the respondents expect from their counterpart, we might get perspective over how to improve or maintain a relationship based on client to partner satisfaction.

20: Do you renegotiate your SLAs with your clients/outsourcing partners?

In the study by Rouse (2005), outsourcing success was achieved through frequent renegotiation of service levels between the client and the vendor as explained in chapter 1.1. We therefore want to know if our respondents renegotiate their SLAs and how this might affect their relationships.

Relationships

21: How important is it for you to have a healhty relationship with a client/outsourcing partner?

This question can be considered obvious but what we really want to know is our respondent’s reasons for wanting a healthy relationship. We hope to find out what a healthy relationship brings our respondents in the form of business success.

22: How do you know if a relationship with a client/outsourcing partner is healthy?

We hope to find out how the communication between both parties works. We believe that communication is essential for business relationships to work. If there are any indications of communication problems, we believe that it could create problems related not only to the SLAs.

23: Do you feel that you can affect a relationship with a client/outsourcing partner?

If our respondents experience any relationship issues between themselves and their counterparts, we want to know if they have the ability to solve the issues. If not, many problems might remain unsolved thus affecting the relationship.

24: What actions take place if you feel that a relationship with a client/outsourcing partner is bad?

This question will give us an overview of how this type of information is handled, either by the respondents themselves or someone concerned about the issue. This is, similar to the previous question, a way for us to know if problems remain unsolved by not reaching someone who can handle them.

25: Does the design of the SLAs influence your relationship with your clients/outsourcing partners?

According to previous studies discussed in chapter 1.1, the SLA does affect the relationship which we would like to confirm with this question. It is important for us to know how it affects the relationship to be able to decide which type of SLA to use in order to maintain healthy relationships.

26: What actions do you take when a relationship is affected by the SLA?

The reason for using this question is similar to that in question 24. It is a way for us to know if problems remain unsolved by not reaching someone who can handle them.

27: Do you renegotiate SLAs in case of relationship problems?

This question is related to question 20. The difference being renegotiations due to relationship problems. We want to know how these problems are being taken care of.

28: Do you have follow-up meetings or similar with your clients/outsourcing partners?

Rouse (2005) states that organizations choosing not to meet frequently with their clients, does not experience the same positive outcomes as the ones who does.

29: What is discussed during these meetings and how do they proceed?

We ask this to get an overview of the meetings, if there are any. By finding out what the topics for the meetings are, we hope to identify if there is something that might be missing which could affect the relationship.

30: Does this happen regularly?

This is related to the motivation given for question 28. Not meeting regularly was proven to be less effective.

31: Is the SLA discussed during these meetings?

This question is used in case the respondents don’t mention the SLA in question 29 as a topic or matter being discussed.

32: What actions are taken after the follow-up meetings?

This relates to question 26 and 24. We intend to identify how the information, mainly from the clients, is handled. We hope to identify possible problems with the way the partner handles the feedback they get from their clients.

4.6 Ethical principles

Before conducting the interviews, we took certain ethical aspects into concern. The respondents were informed that the interviews conducted were strictly anonymous. The respondents also had the right at any time to cancel the interview if desired. Respondents were also informed about their ability to read and examine notes taken during the interview and that they would also receive a copy of the final study. According to Bryman (2011) there are four different aspects which should be considered.

Information requirement - All respondents were informed of the study’s main purpose and assured full anonymity. 

Consent requirement - In addition all respondents was informed that the study was voluntary and that they were able to cancel at any time if they felt uncomfortable or for other reasons did not want to answer or participate in the study. After each interview the respondent was given the opportunity to read through the notes in order to ensure that no mistakes had been made during the transition.  If mistakes had been made during the interview, corrections or additions were made at once after the respondent’s approval.

Confidentiality requirement - All respondents were informed that the study is strictly confidential and all personal data is to be treated that way.

Use requirement - All respondents were informed that the data which is collected is only to be used in purpose for this study.

4.7 Grounded theory

We began to code the data from the qualitative semi-structured interview results. The data was summarized and turned into concepts by the most frequent answers from our respondents with post it notes that were associated to our research question and problem. The concepts were constantly compared with each other to find key concepts that made categories. Since our examination is based on a case study, the categories were saturated due to the number of respondents and data associated to the specific case. The categories were compared with each other and how they related to our research question and problem. Initially two topics were identified, SLA and Relationship.

These two topics are also found in our interview guide as the main themes. Keywords that were frequent were then extracted from the data obtained from the interviews. Positive and negative effects in regards to the two SLA types were formed in order to see the connection between the relational status and the specific SLA type. The process of the grounded theory analysis is shown in the figure below and examined more carefully in the analysis chapter.

4.8 Research quality

Quantitative research often uses concepts of validity and reliability for evaluating that surveys have quality. According to Bryman (2011), there has also been put forward alternative criteria’s that are better suited for qualitative studies. Bryman (2011) suggests Lincoln & Gubas concepts of reliability and authenticity. Reliability is divided into four sub-criteria, namely credibility, transferability, dependability and ability to demonstrate and confirm. For this reason, all respondents have the opportunity to take part of the interview notes and the final study. Authenticity is a set of terms that consist of ontological, pedagogical, catalytic and tactical authenticity. Ontological authenticity meaning whether the study helps the responders to come to a better understanding of their social situation and the social environment they live in.

Pedagogical authenticity defines whether the study has contributed to better the respondent’s view of individuals in the environment experiences different matters. Catalytic authenticity bears the question of whether the study has made it possible for the respondents to change their situation. Tactical authenticity is the question of whether the study has increased the respondent’s possibilities to take the appropriate actions (Bryman 2011).

To ensure that the interview provides a well-balanced opinion of all views and thoughts expressed by respondents during the interviews, respondents were given the opportunity to read and correct their answers before they were incorporated in the final study. External validity indicates whether we can generalize the results to other populations, or as in our case other organizations (Kvale 1997). In the alternative terminology suggested by Guba and Lincoln (Bryman 2011), the terms of credibility and authenticity as an alternative to reliability and validity. Another important component of their concept is that the researcher must demonstrate that he or she knowingly allowed personal values and biases affect the results or conclusions (Bryman 2011).

Figure SLA theory tree structure made though analyzing the results of the interviews using grounded theory

Figure Relationship theory tree structure created through analyzing the results of the interviews using grounded theory

We asked our respondents if a standardized SLA could be considered too simple due to lack of information.

Figure Question 6 from the interview guide. Can a standardized SLA be considered too simple due to lack of information? (Number of respondents)

The chart above shows that the Vendor doesn’t consider standardized SLA’s to be too simple while the Client does. One of the Vendor respondents however, has an improvement in mind. "Talking to my clients I believe that we could use better descriptions for the SLA levels to make it easier for our customers". The main issue with a standardized SLA according to the Client respondents is the information provided to help them associate their problems with the different SLA levels. If the levels are too hard to relate to, it might prevent them from working properly and affect their environment negatively. A Client respondent says that "it can be simple yet informative which is the best combination" which is somewhat shared by another Client respondent who says that "you need to find some kind of level between simple and complex".

We asked the respondents to give us advantages and disadvantages regarding standardized SLA’s. All of the respondents think positively towards standardized SLA’s. The respondents all agree on the fact that having one SLA makes it easier to understand how it works. A disadvantage with the standardized SLA could be that there isn’t enough information about the SLA levels. A Client respondent says that "When you experience problems and need to register a case with your partner, it might be hard to know the severity level for your problem if the level description isn’t informative enough". One Vendor respondent and all of the Client respondents think that additional information could be necessary to aid the client in determining the severity of their problems. "A disadvantage I could think of would be the fact that the SLA levels themselves can be interpreted in different ways says" a Vendor respondent.

The respondents were asked to explain what a detailed SLA would look like. All but one respondent mentions that a detailed SLA could include sample scenario. The respondent claims that detailed SLA’s only confuses the client." I don’t think a SLA could get detailed except for level descriptions going into depth with more error related descriptions" says a Client respondent. The clients would be able to relate their own problems to the sample scenarios thus helping them chose the correct severity level.

Scenarios are often included in more detailed SLAs which is why we asked the respondents if they knew of their existence. "It could be a positive thing to include " says a Vendor respondent. According to a Client respondent "they’re often designed in cooperation with the partner or already defined by them as examples. They’re based of theoretical scenarios in most cases and sometimes the partner uses real scenarios they have experienced through feedback from other clients." A Client respondent says that it’s a positive approach and would ease their case registrations with their partners. Another Client respondent agrees to this claim and says that it would serve as an advantage to technicians if there were scenarios included.

A Vendor respondent has identified a problem with including scenarios with the SLA. "I’ve seen this once and the idea was scrapped due to the client’s request of including almost every possible scenario. It is impossible" says the Vendor respondent. This relates to what two Client respondents said about standardized SLAs being too simple due to lack of information. This would be done by finding a level between simple and complex while still providing sufficient information.

Could a detailed SLA be considered too complex due to excessive amount of information?

Figure Question 12 from the interview guide. Can a detailed SLA be considered too complex due to excessive amounts information? (Number of respondents)

The respondents all agree on the question above as seen in the chart. "It’s better to keep it simple and comprehensive than confusing the reader" says a respondent who shares thought with another respondent who states that "too much information is never good and always causes problems". Then what are the advantages and disadvantages of detailed SLA’s? The general opinion of all respondents except one, tells us that that too much information could be confusing for the reader. 3 respondents think that scenarios could be useful to inclu