Executive Programme in Management Studies Digital Business Track
The Effect of Platform Performance (PP) on Third Party App Developers Retention (TPA),
While Considering Platform Competition
Student name: Sofiane Lahmadi
Student Number: 12102741
Date submission: June 12, 2020
Name Supervisor: Prof. Dr. Peter Van Baalen
Statement of Originality
This document is written by Sofiane Lahmadi who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the
work, not for the contents.
This thesis was a great opportunity for me to take a deep dive in the world of academic research and to contribute to the retention theory within the platform industry.
I would like to take this opportunity to thank Mr. Prof. Dr. Van Baalen for the supervision, the precise remarks and the guidance to make this thesis possible.
My personal thank goes to my wife Dorra Azib and my study comrade Sonny Stroex-Carr for their support, proof-reading and help they granted.
In addition I would like to thank all my participants for the trust and the time they devoted to this work.
It was a pleasure working with you all.
Table of contents
List of figures and tables ... 5
Glossary ... 6
Abstract ... 7
1. Introduction ... 8
1.1 Background ... 8
1.2 The unit of analysis ... 11
1.3 Research questions ... 12
1.4 Research method ... 13
1.5 Thesis structure preview ... 14
2. Literature review ... 15
2.1 Introduction to the Platform ... 15
2.2 Platform Performance ... 18
2.3 Third Party App developers Retention ... 19
2.4 Contribution to existing literature ... 21
2.5 Conceptual model ... 22
3. Methodology and data ... 24
3.1 Research design ... 24
3.2 Sampling... 25
3.3 Data collection and interview design ... 28
3.4 Ethics and quality ... 29
3.5 Analysis and procedure ... 30
4. Findings and results ... 34
4.1 Platform Performance ... 34
4.2 TPA Retention ... 35
4.2.1 The Extend mechanism ... 35
4.2.2 The Exit mechanism... 38
4.2.3 The Hybrid mechanism ... 40
4.3 Platform competition ... 41
4.4 Conceptual model ... 44
4.5 Conclusion results ... 45
5. Discussion ... 47
5.1 Answer to the research questions ... 47
5.2 Theoretical contribution ... 49
5.3 Practical contribution ... 50
5.4 Limitations and future research... 51
6. Conclusion ... 53
7 Reference list ... 55
Appendix ... 58
List of figures and tables
Figure 1. The Hybrid platform Figure 2. The platform ecosystem
Figure 3. Conceptual framework; the effect of PP on TPA retention, while considering Competing platform
Figure 4. Source codes (Saunders et al., 2016, pg. 583) Figure 5. Thematic analysis approach
Figure 6. Themes labeling
Figure 7. Causality between platform performance and the Extend mechanism Figure 8. Causality between platform performance and the Exit mechanism Figure 9. Causality between performance and the Hybrid mechanism
Figure 10.Conceptual model; the effect of Platform Performance on TPA retention, while considering platform competition
Table 1. Participants list Table 2. Platform owners list
Table 3. Platform performance components
AI: Artificial Intelligence, section 5.3
AWS: Amazon Web Services, section 2.1.1 COVID-19 3.1.1 The CoronaVirus Disease
CRM: Customer Relationship Management, section 2.1.1 ERP: Enterprise Resource Planning, section 1.2.
HPaPAAS. High-Productivity application Platform as a Service, section 3.1.1 IAAS: Infrastructure As A Service, section 3.1.1
IOS: iPhone Operating System, section 2.2 ISV: Independent Software Vendor, section 1.2 PAAS: Platform As A Service, section 3.1 PP: Platform Performance, chapter 1 SAAS: Software As A Service, section 2.1.1 TPA: Third Party App developer, chapter 1
The dearth in academic knowledge on Third Party App developers (TPA) retention is the primary motivation of this writing. From that perspective, Kim et al. (2016) concluded in their research that both ‘TPA desire’ and ‘TPA dependence’ on the platform lead to a TPA continued participation in a platform and thus retention. This paper broadens the work of Kim et al. (2016) by investigating other choices a TPA has and by giving insight in the role of the Platform Performance (PP) in these choices. Hence the goal of this research is to answer the research question: what is the effect of platform performance on the TPA choice to extend, exit or mix several platforms?
The answer to the research question is summarized in the following findings:
First, this study demonstrates that platform performance (Cusumano et al., 2018; Parker et al., 2016; Zhou et al. 2018 & Tiwana, 2014) do have an effect on a TPA choice. Four specific performance sub-factors are analysed: innovation, growth, governance and technology.
Second, the findings conclude that: 1) Innovation (as in lock-in & engagement), growth and governance have a direct effect on the ‘extend’ mechanism. 2) Governance and technology have a direct effect on the exit mechanism and 3) Governance (as in lock-in) can lead to a hybrid mechanism. The third finding indicates that when a competitor platform outperforms the primary platform, in innovation or growth, the choice will shift from extend to exit mechanism, which tells us that platform competition matters excessively to the TPA choice.
These findings are significant as they serve TPA strategic decision makers by providing them with: i) a detailed insight into the most important platform performance indicators and ii) guidance on how to consider an exit strategy, especially when competitor platforms act aggressively toward the primary platform.
The increasing interest in the platform industry and the lack of knowledge and academic work on Third Party App (TPA) developers retention, is the elementary motivator of this study.
This empirical research is therefore concerned with the effect of Platform Performance (PP) and Third Party App (TPA) retention within the software platform business. The qualitative approach aims to get a deeper understanding of the phenomenon ‘TPA retention’ by investigating the choices a TPA decision maker may have. From that perspective this study investigates the role of platform performance on three different outcomes: the extend mechanism (a TPA continues to participate in its primary platform), the exit mechanism (a TPA decides to leave a platform), the hybrid mechanism (a TPA decides to mix several platforms at the same time).
Additionally, the effect of a competitor platform on the relationship between platform performance and TPA retention will be examined as a moderator.
This chapter will cover the background information, the unit of analysis, the research question broken down into sub-questions, the research method and the paper structure.
Van Alstyne (2019) has made an interesting comparison between the most dominant firms worldwide and the top seven platforms. He argues that these platforms, together $4.8 trillion worth of market capitalisation worldwide, are outperforming both the bank and oil sector together (Exhibit 1). These giant platforms are by far the most successful firms worldwide, spread out through several industries such as social networks (Facebook and LinkedIn), e- commerce (Amazon and Alibaba), payment (paypal and Apple Pay), mobile (Apple store and Android) and software Cloud platform (Salesforce and Microsoft) (Cusumano, 2019).
9 Central to this, is the trending use of the platform technology by Third Party App developers (TPA) to build their apps. Cusumano (2019) argues that this is due to software vendors opening their technology for innovation. For example, in 2005, Salesforce opened its AppExchange for external users, followed by its Force.com platform in 2008. During the same period, Amazon introduced the Amazon Web Services (AWS) and in 2010, Microsoft introduced the Azure platform (Cusumano, 2019). This trend made it accessible and simple for TPAs to build their apps using specific platform language and to profit from selling their apps on the platform Marketplaces like AppExchange or Google and AWS Marketplace.
The attractiveness of the platform to TPAs created value to the platform owner (Kim et al., 2016), which resulted in a competitive environment between platform owners. Several examples can illustrate competition between platforms. For example, Apple is competing with Android (operating systems), Salesforce with Microsoft (PAAS) and Amazon with Google and Alibaba (IAAS). The platform competition and the increasing number of platforms (more than 500) could lead to innovation as well as to challenges for TPA decision makers. These challenges could be the difficulty to judge whether a TPA is using a well performing platform, and if not, when and how to switch quickly to other platform innovations.
From that perspective, this research will address three factors related to the TPA retention. 1) the TPA choice to stay with its primary platform or switch to a competitor platform.
2) the role of platform performance in these choices. 3) the effect of a competitor platform entry on the relationship between performance and TPA retention. To address these factors, the following generic research question will summarize the main goal of this research:
‘’What is the effect of platform performance on the TPA choice to extend, exit or mix several platforms?’’
To answer this research question an initial literature review, as covered in chapter 2, has resulted in the following information: The platform performance phenomenon has been examined by many scholars and academics (e.g. Zhou el al., 2018; Parker et al., 2013). Some
10 examples of these indicators are: financial performance studied by Cusumano et al. (2018), control mechanisms & governance studied by Tiwana (2016) and switching cost or vendor lock- in (Tiwana, 2014). The large and accessible amount of literature on performance sub-factors will be used in this research.
Despite the popularity of platform performance, the TPA retention factor is judged to have an extensive need for a deeper examination. There is a clear shortage of academic literature related to the TPA retention phenomenon, with some exceptions, like the study conducted by Kim et al. (2016). Kim, Kim and Lee (2016) examined the ‘retention’ by studying the ‘relationship’ between a third party app developer and the platform owner. In their framework, they identified two reasons for a third party app developer to continue participating in an existing platform. 1) the dedication mechanism and 2) the constraints mechanism. In this context, the dedication is based on the desire to participate in the platform, for example loyalty, and constraints are based on the restriction to leave a platform, for example termination costs.
These findings are very significant as they construct the starting point of this paper.
From a theoretical point of view, no additional academic work, besides Kim et al. (2016) research, has been identified in relation to TPA retention. This leaves us with a new research area and a motivator for this study to identify and understand the ‘retention’ phenomenon from a TPA perspective. From that viewpoint, this research can lead to interesting contributions to the theory of retention. From a practical point of view, this study is relevant to practitioners as well. It provides TPA decision makers with tools that can be used to make the right choice by considering platform performance and competitor platforms influences.
To ascertain that qualitative depth is achieved and to give the research a context, a case study will be introduced in the following section.
1.2 The unit of analysis
This research is inspired by a professional real life case study. An introduction of this particular firm called Third Party Apps Developer (TPA) aims to give the research a context.
This particular example will identify practical questions asked within the firm in relation to the Platform Performance (PP). These questions and choice evaluations will not be isolated to this particular firm and could therefore be judged as relevant to a considerable amount of TPSs worldwide.
Company S is a private independent software and service company with a focus on customized solutions. The company has a rich history deriving from the former ERP supplier. It was considered to be the second largest Enterprise Resource Planning (ERP) systems vendor during the nineties. Due to its success, the ERP software was acquired by a global enterprise firm. After the acquisition, Company S has merged and during the last decade, the firm has been facing serious difficulties to differentiate itself from competitors. By focusing on building customized apps on the top of (ERP) systems, management hopes to bring back the success of the nineties. Apps are built in different platforms by using heterogeneous technologies mandated by the platform owner such as Salesforce, Google and Amazon.
Lately, the company has disclosed strategic challenges, which led to several discussions between the board of directors and the shareholders on how to achieve growth and success by improving its platform business strategy. Although platform adoption can lead to innovation and market access, investing in several platform technologies leads to high costs, complexity and loss of focus and intellectual property.
“Platform strategy shares same downsides with open innovation such as increased management process coordination cost and potential loss of key knowledge control or intellectual property spillover” (Enkel et al., 2009; Müller, 2013; Veer et al., 2013).
12 Knowing that platform owners transfer fully their financial risk to the app developer as they are the ones to invest in the technology, a legitimate question of a TPA-manager should be: am I still using the right platform? Do I have to focus on only one platform or should I consider benchmarking other platforms?
This research aims to answer these questions to enable Independent Software Vendors (ISVs) to keep creating value for their shareholders and end users.
Using a real life example will simplify access to interviewees and will facilitate depth. Achieving depth in the research design and using real life examples can lead to:
‘rich, empirical description and the development of theory.’ (Dubois & Gadde, 2002;
1.3 Research questions
Despite the interest in the platform business, few academic researchers have covered the relationship between Platform performance (PP) and Third Party App Developers (TPA) retention . Hence the generic research question:
‘’What is the effect of platform performance on the TPA choice to extend, exit or mix several platforms?’’
The research question could be divided in two parts. The first part of the research focuses on the relationship between platform performance and TPA retention. This relationship will be treated according to three different outcomes, hence the sub-questions:
1: How would platform performance motivate a TPA to extend participation in a platform?
2: How would platform performance be a reason for a TPA to leave a platform?
3: How would platform performance be a reason for a TPA to mix several platforms?
The second part of the research will be devoted to the moderator ‘platform competition’, which is predicted to have an effect on the above relationship. Therefore the sub-question:
13 4: Does a competitor platform have an effect on the relationship between platform performance and TPA retention?
These questions could be used as a guideline to collect data from existing literature, platform performance, and from primary data through observations, TPA retention.
1.4 Research method
This research will be conducted in a qualitative and exploratory method by collecting primary data from the TPA community within the Software platform business. The qualitative approach is the most appropriate as using a study case and conducting semi-structured interviews will ascertain depth and a profound understanding of the phenomena ‘TPA retention’
as well as trust through human interaction.
Due to the lack of academic material, especially on TPA retention, an abductive approach will be used in this research. This means that the researcher will be moving back and forth between theory and collected data to find patterns that could be generalized. In this context platform performance will be treated as theory driver and TPA retention as data driven method.
1.5 Thesis structure preview
2. Literature review
This chapter is dedicated to the initial literature research with three focus areas.
1) Introduction to the terminology and typology of the platform from different angles, 2) Introduction to the independent factor of this research ‘Platform Performance and 3) Introduction to TPA retention, which will be treated as an outcome factor.
2.1 Introduction to the Platform
One of the most remarkable outcomes of an initial literature review is the distinctive types of platform identified by previous literature. As a result of the quick technological changes and severe competition, these types of platforms are subject to evolutionary changes.
Due to the expanding interest in the platform economy and the wide range of interpretation of the term platform, a distinction between the different types of platforms can provide the reader some contextual insight.
The most common types of platforms have been introduced by several researchers, among whom (Evans & Gawer, 2016), are classified into four different types: innovation platforms (e.g. Salesforce), transaction platforms (e.g. Uber), integration platforms and investment or payment platforms (e.g. Paypal).
Some researchers have chosen for a complex platform typology studies (e.g. Parker, Van Alstyne, & Choudary, 2016) and others identified only two types: the transaction platform and the innovation platform (e.g. Cusumano, Gawer & Yoffie, 2019). The intersection between the latest two is called ‘Hybrid platform’ as illustrated in figure 1 below. Whether they emerge from transaction to hybrid or from innovation to hybrid, they are seen as the most valuable platform companies worldwide (Cusumano et al., 2019), including names like Google, Microsoft,
16 Salesforce and Amazon. This research will be mainly focusing on these software-based hybrid platform types.
Figure 1. The Hybrid platform (source; Cusumano et al., 2019)
2.1.1 Platform Competition
Due to the importance of the platform competition in this research, below information should be of interest to clarify the origins and different categories of software platforms.
In his book ‘Platform Ecosystems’, Tiwana (2014) used the Red Queen effect to emphasize on the evolutionary element when he quoted:
“the evolutionary pace of a platform is relative to its rivals.” (Tiwana, 2014).
Cusumano (2019) argues that this evolution is due to software vendors opening their Cloud infrastructure for third party developers to innovate and build upon their hosting
17 environment. Compared to the competition between Apple store and Google Play in the Smartphone technology, software vendors have also been dealing with competition. So did Microsoft, Google, Salesforce and Amazon open their cloud-technology for app developers which generated a race after the TPAs. In 1999, Salesforce launched a CRM SAAS version (Software As A Service). In 2005 they launched the AppExchange to stimulate third parties to develop and sell apps through their platform and in 2008 Force.com platform. In that year Amazon, with more than 400,000 developers using Amazon Web Services (AWS), is considered the most used platform (Cusumano, 2019). According to Cusumano (2019) it is due to the independence of the platform from the Amazon products. Azure was in 2010 the first step Microsoft took in the platform service by making it possible for TPAs to use other frameworks than Microsoft .NET such as Java and Python. The Azure business has shown an increase of 70% in 2018 and 90% in 2019 which makes Microsoft the biggest threat in the platform business (Cusumano, 2019).
2.2 Platform Performance
This section covers the factor ‘Platform Performance’ from different approaches examined by existing literature and former researches and will give direction to this research.
The rich amount of literature available around this topic will be structured from different angles to demonstrate the profound meaning of different mechanisms and models introduced such as platform growth (financial performance indicators), control mechanisms and value added.
Cusumano, Gawer and Yoffie (2018) have conducted a longitudinal study on platform performance and collected exploratory data between 1995 and 2015. The main focus was the financial performance platform owners and compared their results to traditional pipeline business companies. They examined data of 46 platform companies, of which 19 are innovation platforms and 27 are transaction platforms (exhibit 2). These are companies providing cloud, software and infrastructure platform products and services, such as Apple (IOS and personal computer business), Microsoft, Google, Amazon and Salesforce. The performance indicators they measured are: 1) operating profit margins, 2) market value and 3) sales growth rates (Exhibit 3). In addition they compared strong performing platforms with weak performing ones and from that perspective they agreed with the findings of (Parker, Van Alstyne & Choudary, 2016) that well performing platforms outperformed others in the following factors: 1) platform openness, 2) engagement with TPAs, 3) spillover, 4) launching the right side (TPA or end-user), 5) critical mass, 6) imagination (continuously innovating).
Zhou et al. (2018) added to previous research the multi-sided nature of the platform business and examined the platform performance from two sides: 1) the platform owner side and 2) the supply-demand side. From a platform owner perspective, they related performance to: 1) pricing, 2) openness and 3) control. From the supply-demand they concluded that three factors can affect the platform performance. These are: 1) the quality, 2) the quantity and 3) the
19 diversity of i) new TPAs and ii) TPA-updates. They also concluded that these effects are influenced by the entrance of a competitive platform.
Maurer and Tiwana (2012) have conducted their research by studying the platform ecosystem in the mobile computing industry. They also found out that platform performance depends on both sides of the platform and that this performance is related to the control mechanism orchestrated by the platform owner. In their analysis they argued that the control mechanism can lead to technology integration and app differentiation, which attracts more users and could be seen as a well performing platform.
In more recent writing, Kim and Tiwana (2016) have concluded that both formal control mechanisms (input, output and process control) and informal control define the degree of platform performance and in his book ‘Platform Ecosystems’ Tiwana (2014) defined the core principles of a platform as in; 1) governance (among others control mechanisms), 2) architecture and 3) balance control.
Another look at platform performance is the platform added value. A major differentiator of platform performance is the value a platform owner adds to its ecosystem. Boudreau (2017) agrees with the findings of Boudreau & Hagiu (2009) and Gawer & Gusumano (2002) that opposite to the traditional business, the platform owner should be at the center of its multi- relationships as a orchestrator of values created in the ecosystem. Tiwana (2014) has identified two major values a TPA can gain from using a platform: technology and mass access.
2.3 Third Party App developers Retention
Surprisingly, the TPA retention has not been considered by existing literature, which makes this research challenging and interesting at the same time. To settle an acceptable definition of the term ‘retention’, an online search revealed the following explanation ‘retention is the continued possession, use, or control of something.’ (online Cambridge dictionary).
20 Different frameworks of the retention theory are found to be relevant studies and mostly agreed on the economic importance of retention (Grönroos, 1990; Dawkins & Reichheld, 1990).
Basically, retention is when a user keeps using products or services of an external supplier and does not leave or switch to a competitor. It is argued that retaining a ‘relationship’ is more profitable than the acquisition of new ones and that retention adds more value to the business (Chrstopher, Payne & Ballantyne, 1991; Peck et al., 1999). The main argument is that acquisition is much more expensive than retention (Reichheld & Kenny, 1990; Gallo, 2014) and that retention will be transformed in more value to the business in the long run (Vandermerwe, 1996).
The most relevant TPA retention study is the one conducted by Kim, Kim & Lee (2016).
Kim, Kim and Lee (2016) treated the term retention from the perspective of the ‘relationship’
between a platform owner and a TPA. This valuable empirical research is platform-centric and has identified two specific mechanisms that motivates a TPA to continue participating in a platform. They argue that it is important for the platform owner to understand how this relationship is perceived by a TPA and that this knowledge can shield the platform owner from potential competitor platforms. Their dual model framework predicted that retention is related to two factors: 1) dedication and 2) constraints, earlier introduced by (Kim & Son, 2009). In that context dedication happens when a TPA desires to participate in a platform and constraints happen when a TPA continues the participation because they are obliged to do so (e.g. due to start-up costs or high learning costs). The variables they identified are two sided. The first part is in relation to the ‘desire’ to participate in a platform. The desire, they argue, is related to the following variables: 1) revenue attractiveness, 2) market demand, 3) development tools, 4) online forum and 5) review fairness. The second part is in relation to the ‘need’ to participate in a platform, based on the following variables: 1) learning and 2) setup activity. These variables lead to termination costs constraints and will result in dependence on the platform. According to
21 their statistical testing and findings, all these factors, except for online forums, have an effect on a TPA continuing participation on a platform whether due to dedication or constraint.
Opposite to continuous participation, one may be interested in the reasons of a platform owner to fail retaining its TPA or, in other words, the reasons for a TPA to leave a platform.
Keaveney (1995) has identified 8 factors to leave a platform: 1) pricing, 2) inconvenience, 3) service failures, 4) service encounter failures, 5) employee’s responses to service failures, 6) competitor attraction, 7) ethical problems and 8) involuntary switching and seldom-mentioned incidents. This research is leaning toward the end-user, but one may consider these variables as definitely relevant to TPAs serving these end-users and listening to their desire.
Tiwana (2014) argues that the increasing number of apps and the evolutionary technological pace of several apps can make integration a complex task. Cusumano et al.
(2018) gave the example of Apple and argued that managing two millions apps could be a very complex and challenging task for the platform owner and this could result in customers dissatisfaction (Zhou et al., 2018) and TPAs leaving the platform. Bad architecture due to complex innovation (Parker et al., 2018) that can’t be managed by the platform owner is one of the main reasons for a TPA to leave a platform. However, partitioning and integration architecture can make complexity manageable (Tiwana, 2014) and lead to retention. Besides architecture and governance (Cusumano 2019), as revealed in the previous section, other aspects could be considered by this research are: engagement of the platform owner with its TPAs and the vendor lock-in & switching costs (Tiwana, 2014).
2.4 Contribution to existing literature
The platform phenomena has been classified as being a totally different business model compared to the traditional pipeline business. Due to the multi-sidedness of this model, where supply meets demand, a solely perfect balance between these three elements of the platform
22 business as illustrated in the figure2 could lead to success. Hagiu and Wright (2015) define this as a ‘’system of matching supply and demand’’.
Figure 2. The platform ecosystem
This research contributes to existing literature by looking at the retention phenomena from the supply side, which enjoyed rare attention. This research will thus examine the third party app developer as an important organ within the platform business and their choice to stay or leave a platform. Without TPA there is no platform and without supply there is no demand.
“The attractiveness of a platform to prospective end-users is therefore influenced by the availability and diversity of apps that complement it” (Tiwana, 2014, P. 66)
It is argued that there is no evidence of a prior study disclosing a TPA retention theory based on platform performance. This makes this empirical study an interesting and unique reading material at the same time for both scholars and practitioners.
2.5 Conceptual model
This section will discuss the initial conceptual model based on the research question and the literature review resulting in the model factors as shown in figure 3 below.
With reference to the literature review and the identified gap, it should be clear that the writer wants to examine the ‘Third Party App developer retention’ as the outcome factor in relation to ‘platform performance’. From that perspective, figure 3 reads as follows: platform performance could have an effect on a third party app developer retention. This effect could
23 have three possible outcomes: 1) a TPA may choose to extend the use of its primary platform, 2) a TPA may wish to exit the platform or 3) it may mix several platforms at the same time.
These three elements will construct the three first propositions indicated in the research question and illustrated in the below figure as P1, P2, and P3.
The second expected relationship is the effect of a competitor or entry competitor on the relationship between TPA retention and platform performance, illustrated as ‘Competing P.’ and mentioned in the research question as the fourth proposition of this study (P4).
Figure 3. Conceptual framework; the effect of PP on TPA retention, while considering Competing platform
3. Methodology and data
This chapter will serve as a general plan on how to answer the research question by covering the following sections: research design, sampling method, data collection, ethics and data analysis.
3.1 Research design
This research aims to study the effect of platform performance on the third party app developer retention. To be able to answer this question, a mono-method approach is adopted in an abductive manner. Abductive means using existing literature as a groundwork for the platform performance and using the collected data to explore the TPA retention.
The data is generated by TPAs with an office in Europe or participating in a platform with an office in Europe. The source of the voluntary participants is the interviewer's existing network and through social media like LinkedIn. TPAs within the Salesforce, Google, Amazon and Microsoft community were mainly targeted. TPAs are chosen as a source of data, due to their direct or close involvement with the decision to leave or stay with the platform.
TPAs participating in only one platform like Gen25 and ComplianceQuest are purposely selected. The input of this first group of participants is highly valuable as their choice for one platform could reveal interesting insight on the investigated retention outcome in relation to performance. The flip side effect of the mono-platform choice is that this input may bias the result, as the decision is already made to be loyal to one platform, no matter the performance.
Surprisingly, this was not the case, as even some of these monogamous TPAs have ever switched platforms or are considering that, like the Rootstock example discussed in chapter four. This group has revealed interesting findings toward their strategy and even the role of competition in their choice.
25 The second group consists of TPAs participating in several platforms at the same time, like the introduced case study (Company S) or the global company Deloitte. The input of this group is interesting as its choice for multiple platforms may reveal important information on the effect of performance and platform competition of that choice. Major participants, as shown in table 1, are c-level employees involved in the strategic decision making and have shown ability to discuss retention, performance and platform competition on a strategic level.
Due to the competitive environment between TPAs and the sensitive company information that will be shared with the interviewer, a full trust is needed. Therefore it is chosen for the qualitative approach as it guarantees trust through human interaction, full transparency and an agreement to participate beforehand.
In addition, the relationship between performance and TPA retention is rare, could be very complex and can get very detailed. It is therefore proven that the qualitative method is the best way to: 1) achieve depth, 2) get a detailed identification of a specific performance indicator and 3) relate this particular indicator to the TPA choice (leave of extend).
The interviewer adapted the interpretivism as philosophy, where the content of the answers of the interviewees depends on their interpretation and meaning of argumentation. The open questions lead to deeper and multi-dimensional levels of the real meaning of statements or announcements made by the interviewee and which, in general, lead to a new insight.
“Qualitative research is an amorphous, multi-dimensional field which forbids any easy single definition or set definitions” (Marisson, 2014, p. 328).
The focus area of the sampling is based on third party app developers (TPA) within the software platform ecosystem. There are thousands of TPAs related to more than five hundred platforms worldwide. Accordingly, a selection of a hundred achievable TPAs is made and
26 targeted for interviews. The voluntary participation had three sources; 1) employment, 2) Social Media and 3) network as shown in table 1. To support the generalization of this study, a primer selection of these targeted TPAs is needed to comply with certain imperatives. Some of the TPAs need to be generic app developers using for example AppExchange, where platform dependency could be tested. While others are customized app developers (e.g. Vanenburg Software or Pragmatiq) using different platforms. Second, some of these TPAs needed to be start-ups (e.g. SalesEngineers) and others needed to be incumbent (e.g. Rootstock). This is needed to test the capability and willingness to switch platforms based on company size.
This study succeeded to interview seventeen participants as shown in table 1 below.
Sixteen of these were TPAs complying with research requirements and one platform participant (Google) used as validation of the retention and Performance from the platform owner vision and to give access to Google ecosystem. While the interviewees targeting method was successful, it must be noticed that it is time consuming and does not always lead to interviews (in such a short period).
27 Table 1. Participants list
The diversity of the platform owners plays a significant role in the differentiation between several platform types such as PAAS (e.g. Salesforce and Microsoft), IAAS (e.g. Google and Amazon) and to make generalization possible (as shown in table 2).
Table 2. Platform owners list
Platform owners used by participants
DocuSign Betty Blocks
3.3 Data collection and interview design
The interviews are semi structured with an average of sixty minutes per interview.
An extract of the structure (exhibit 6) shows the main elements of the open questions in relation to the propositions and research factors; platform performance, third party app developers retention and platform competition. Every interview started with a short introduction of five minutes, a general discussion about the business and the interviewee background to set the scene and make the interviewee comfortable.
The structure of the interviews is kept intact during all interviews as it has shown effectiveness and focus on core research sub-questions. These fundamentals are: 1) the TPA platform choice, i) to stay with the primary platform, ii) to leave a platform, iii) to use several platforms at the same time, 2) the effect platform performance of these choices and 3) the role of competitor platform on the investigated relationship.
These questions represented the main focus and took between thirty and forty five minutes. The question-set helps to make sure no research elements are left out and the open question helps to identify aspects and interpretations that haven’t been considered and could be crucial for the TPA retention theory. “Theory-oriented research is research where the objective is to contribute to theory development. Ultimately, the theory may be useful for practice in general.” (Dul & Hak, 2008, p. 31). While having a structure has shown positive results, it must be noticed that the depth could be lost due to time constraint and focus on the question structure. For example during the introduction phase, when participants disclosed valuable but unstructured information that could lead to deeper discussion and may end up losing focus on the topic.
Every interview ends by thanking the interviewees and giving them space for general feedback, reflection or additional information. Besides the content also the context was taken into consideration by collecting information like, location, feeling, and behaviour in a form of
29 notes. The interest for the finding of the research was clearly articulated by the majority of the interviewees. Due to the pandemic spread of the COVID-19 virus all the interviews were conducted online through video conferencing, which helped reaching out to international participants from France, Brazil, England and Sweden.
3.4 Ethics and quality
The research is conducted through human and organizational participation, where ethical concerns were considered at all times, every time and during every interview. The ethical methods based on three elements: Fairness, integrity and confidentiality (Maw & Khin, 2014).
Fairness: all the interviews were conducted in full transparency by explaining the reason and topic of the research and full agreement of the candidate to participate to record the interviews.
Integrity: before every interview a formal or an informal accord is given to the interviewer, based on an e-mail or a call, by the organisation or the participant. The topic and examples question were shared with the participant to manage expectation and guarantee integrity. Every interview was recorded after full agreement of the participant. Also the reason for recording the meeting was clarified and understood by every participant.
Confidentiality: due to the competitive nature between the TPAs, specific information disclosed by interviewees and considered to be confidential is removed right away (if asked by the candidate) or treated as generic statements to contribute to the findings.
The quality of this research is assured by following several protocols based on certain qualitative research criteria, like the ones defined by Frambach, Vleuten and Durning (2013) and Perry and Perry (2017).
Credibility: this study aimed to be trustworthy and based on credible findings by using different TPAs besides the study case participants and by interviewing multiple cloud platforms users. Also triangulation tactic has been implemented by getting definition confirmation not only
30 from a TPA point view but also the platform owner. Every interview ended with asking participants for feedback and interpretation of the investigated relationship.
Transferability: the researcher tended to collect data from several participants from different countries to apply transferability of the findings in different settings and to make the findings meaningful to the TPA community worldwide, not only applicable for the Dutch market. In addition these findings are regularly reflected to existing literature to make sure the link with the literature is visible.
Dependency: the findings have been consistent during the interview process and this fact (in relation to the research question) has been checked after every five interviews until saturation is reached. Likewise for the continuous review of the collected data to make sure no elements are left out. By being open new findings from collected data is added to the final framework.
Confirmability: researcher tended to be as objective as possible by not reacting to silence and leaving participant time to answer questions without interpretation or possibility to bias the data. In addition every interview had its own memo to reflect on the interviewee evaluation and area of improvement to make sure objectivity is reached.
3.5 Analysis and procedure
This study is characterized by its deductive approach when relating to existing literature on ‘Platform Performance’ and its inductive approach toward a TPA retention theory that is kept unexplored. Therefore the choice for a Thematic analysis as a flexible and thematic method, usually adopted in a qualitative research (Braun & Clarke, 2006), is the best match for this academic study.
Interviews were fully online, recorded material is transcribed manually and fed to ATLAS.ti for analysis and coding. About eighty percent of the interviews are conducted in Dutch
31 and the remaining in English. The manual transcript is fully translated to English, which comes always with a marginal interpretation due to translation. This process is time consuming, but very thoughtful as it gives full insight to the collected information and leads to familiarity with own data, resulting in seventeen documents. The contextual notes made before, during or after the interviews are captured and added to the final transcript. While individual memos are attached to every document containing general summary and specific details of the interview, the interviewee behavior and important quotes. The coding structure, as shown in figure 4, is divided into two categories: 1) data driven codes; defined during the research and 2) theory driven codes; derived from existing platform theory.
Figure 4. Source codes (Saunders et al., 2016, pg. 583)
By adapting a Thematic analysis (Gioia et al., 2012), the collected data is categorized by similarity (exhibit 8). Some of these are power or proof quotes, ranked at first order concept, which are thematically arranged by resemblance under the second order themes. The highest level is the aggregate dimension related to the research question as illustrated in figure 5 below.
32 Figure 5. Thematic analysis approach
The themes labeled in the second order resulted in fifty eight codes (exhibit 7) and sub- codes and three main groups (aggregate dimensions): 1) the effect of platform performance on the TPA choice (extend, leave or hybrid), 2) TPA choice (extend, exit or hybrid) and 3) the effect platform competition on the relationship between performance and TPA retention, as shown in figure 6.
33 Figure 6. Themes labeling
The data collected during interviews was rich and valuable. The amount of the codes exceeded at first seventy codes and reduced to under sixty codes, which helped focusing on the research questions, for example the codes community and ecosystem were merged to one code and the same thing applied to lock-in and switching costs (Exhibit 6).
During the first five interviews a big amount of quotes are collected and labeled, which was valuable and time consuming. Based on learning and focus on the research question, the second round of interviews led to the consolidation of comparable quotes to achieve effectiveness.
4. Findings and results
This chapter will present a detailed analysis of the findings of this research based on 17 interviews conducted during the period March to May of 2020.
The main findings are summarized in the research model in figure 10, where the TPA choice is illustrated as a reflection on the platform performance and divided into three outcome mechanisms (unique to this study and inspired by the work of Kim et al. (2016). 1) The extend mechanism. 2) The exit mechanism. 3) The hybrid mechanism.
The results will be presented in a systematic manner based on power and proof-quotes, generated by the participants as a groundwork. This thematic data description will be contextually bound and will focus on three main factors: platform performance, TPA retention and competitor platform. This will help answering the research question and to summarize the conceptual model of this study.
4.1 Platform Performance
Platform performance has been examined in a deductive manner by using several elements of the platform business revealed and presented in the literature review chapter.
These elements are used to give structure to the interviews and to make sure that these are validated with the participants knowledge. This study narrowed down the performance factors to four main sub-factors as presented in table 3 below. These sub-factors are innovation, growth, governance and technology as presented in detail in the table 3. The thematic consolidation of these sub-elements will help identify in extreme precision the relationship between every sub- element and the appropriate TPA choice (extend, exit or hybrid).
Table 3. Platform performance components
4.2 TPA Retention
The findings presented in this section are in relation to the outcome factor ‘TPA retention’. This factor will be examined in three separate mechanisms in relation to platform performance: the mechanism of extending participation in a primary platform ‘the extend mechanism’, the mechanism of leaving a platform ‘the exit mechanism’ and the mechanism of mixing several platforms at the same time ‘the hybrid mechanism’.
4.2.1 The Extend mechanism
The collected data in this section resulted in seventy-three quotes and examples related to the ‘Extend mechanism’ and that could be summarized as follows:
A TPA will extend the use of a platform due to three major factors in relation to platform performance: 1) innovation 2) growth and 3) governance. This study concludes that technology is not a differentiator and has therefore no significant effect on TPA choice.
Figure 7 below gives a summary of the ‘extend mechanism’ in relation to platform performance factors which will be analysed in details.
36 Innovation has shown to be an important performance sub-factor, with direct effect on the
‘extend’ mechanism (P1). Due to the intensive platform competition and the fast past of technology, good and regular innovation leads directly to TPA retention, as quoted below.
‘As long as the platform continues performing and being innovative, we will keep using the platform.’ (Participant 6)
‘Innovation and evolution can play a positive role to retain a TPA.’ (Participant 1).
2) Platform Growth
Within platform growth two main themes or sub-factors have been noted as a reason for a TPA to extend participating in a platform: the revenue growth and market access, as illustrated in the below excerpts:
‘Platform financial performance is in relation with our own performance; the better a platform does, the more customers get interested in the platform and the bigger our horizon is.’
‘Salesforce can retain a TPA when they keep selling a TPA app (….). Salesforce gives a TPA market access worldwide.’ (Participant 12)
3) Governance is quoted thirty-five times, making it the most frequently quoted factor.
Compared to other sub-element of the governance factor (as shown in table 3), lock-in and engagement model are by far the most noticeable sub-elements according to these findings.
Platform Lock-in practice, quoted twenty-seven times, represents one of the most frequently mentioned reasons to extend the use of the primary platform. For example, if you build an app on the Salesforce AppExchange (using Force.com coding language), leaving Salesforce will become a hurdle, like the following excerpt will prove.
‘The apps we build in Salesforce could be used only in Salesforce, but that is our choice.’
To give this sub-element more context an example has been shared with the interviewer.
Rootstock (Participant 16) built the biggest app on the AppExchange, based on 547 objects and for them, leaving the platform is just not an option, as stated by the Participant (16).
‘We are not in the position to switch platforms. We have no choice. Our product is built on the platform, it is therefore difficult to leave the platform, as leaving means rewriting the whole application again.’ (Participant 16)
Platform engagement model has been mentioned thirty-times in total and includes elements like direct engagement with the platform sales channel, the platform community, loyalty program and training and certification. The following excerpt expresses explicitly the direct relationship between engagement and TPA retention.
‘Platform engagement with its TPA increases the retention of the TPA and reduces the need to look at other platforms.’ (Participant 6)
The next participants give more specification on the term governance and relate it to retention.
‘The Motivation to stay with an existing platform is through an engagement model like trailhead (training facilities), where community can play an important role.’ (Participant 5)
‘Retention to us is related to partner status and loyalty program, engagement model and strategic involvement. Also common sales and marketing efforts e.g. events webinars, upscale resources.’ (Participant 9).
38 Figure 7. Causality between platform performance and the Extend mechanism
4.2.2 The Exit mechanism
The second outcome factor, of which the results will be presented, is the TPA exit mechanism. Sub-factors that have effect on the outcome factor are illustrated in figure 8 below.
To make sure that there is no correlation between our factor ‘platform performance’ and other factors like platform competition, all other factors are held constant through the interview process. Even when an interviewee tends to start discussing the effect of platform competition on the relationship for example, the question will be rephrased and interviewee will be requested to rethink his answer without having the platform competition factor his reasoning.
The exit mechanism has been acknowledged fifty-two times in total, twenty-one times less than the extend mechanism, making the amount of TPA willing to leave a platform 30% less than staying with the primary platform, for example due to the lock-in hurdle.
The study results show that a TPA will leave a platform due the two following major factors 1) competing with the platform owner (bad governance) and 2) platform technological limitations.
1) Platform competition with TPA (bad governance)
Out of the total ‘exit mechanism’ collected codes, twenty-nine quotes have mentioned platform owner competition with a TPA as a main reason to leave a platform, as this could have a direct effect on the business. These quotes are either hypothetical or based on real life examples.
‘The only reason to exit the platform is if they start competing with us in one way or another, like building their own ERP app.’ (Participant 16)
Major participants agree that bad governance of the roadmap, such as not communicating on time or bad communication with all related TPAs, is seen as failure of platform governance.
‘The roadmap of the platform owner is an important challenge for TPAs, as it could be a threat for app developers.’ (Participant 4)
Important to notice that platform owners with a direct sales interest and own products range (e.g. Salesforce) could cannibalize apps from its TPAs. This is experienced as bad governance in favor of the platform owner apps.
‘An exit strategy or exit cause for a TPA is a direct competition with Salesforce own products, like healthcare cloud and manufacturing cloud they just introduced.’ (Participant 4)
2) Platform limitation
The second most important reason for a TPA to leave a platform is the platform technological limitation. According to most interviewees the technological aspect of a platform is not a significant differentiator.
‘Technically, it is easy to leave the Salesforce platform or use different elements (like storage) of other platforms.’ (Participant 2)
40 However, a bad technological performance could lead directly to an exit mechanism, as the technological switch is not that difficult, as quoted by Participant (2) and Participant (16).
‘We have recently chosen to leave a platform due to bad technological performance, such as slow or lack of updates or security (especially after the Citrix incident). In that case, a developer has to put a lot of efforts and investments to keep the developed app working.’ (Participant 2)
‘Before Salesforce (2008), Rootstock was built in another platform called Netsuite, which we strategically left due to limitations in terms of technology as well as business ethics.’ (Participant 16).
Figure 8. Causality between platform performance and the Exit mechanism
4.2.3 The Hybrid mechanism
The Hybrid mechanism has been quoted twenty-six times and has been directly related to governance. The primary platform performance factor, leading to direct relationship with the mechanism, is the platform lock-in practice, as illustrated in figure 9 below.
Platform Lock-in represents an important reason for eleven of the sixteen interviewed TPAs to choose to mix several platforms. The main argument is that mixing platforms decreases the platform dependency. Once you are locked, switching costs will be very high.
‘Mix & Match is an interesting choice to be platform independent.’ (Participant 4)
‘Vendor lock-in is also one of the reasons not to commit to only one platform like Microsoft. If we want to overcome vendor lock-in, we have to implement less in PAAS or combine platforms’
technology.’ (Participant 10)
At the same time, the fear of high switching costs and the technological evolution is dictating app developers to combine platforms when the technology is very similar. Some TPAs choose for a combination of different components of different platforms, like building the application layer on Salesforce and the storage layer on Google (from a cost effectiveness point of view).
‘It is getting very easy to use or switch between several cloud platforms at the same time, like hybrid or multi-Cloud solutions. Especially hosting is made very easy by, for example, Anthos (Google) where you can combine platforms and technologies or just move your containers from one environment to the other.’ (Participant 5).
Figure 9. Causality between platform performance and the Hybrid mechanism
4.3 Platform competition
Platform competition is the only moderator factor analysed in this study. This factor has been expected to have an effect on the relationship between platform performance, of the primary platform, and the TPA retention. The data collected twenty-nine quotes from thirteen different participants, making it an important factor to analyse.
42 This research has evidence that platform competition has a positive as well as a negative effect on the examined relationship, as shown below. What makes these findings unique is the effect of platform competition that changes the mechanism direction from extend to exit, as shown in (P4) figure 10.
The collected data tells us that the effect of a competitor's innovation (Participant 5, quote 1) and growth (Participant 5, quote 2) have an effect on the relationship between performance and retention, by shifting from an extend to an exit mechanism, as mentioned by below quotes.
‘If a competitor platform outperforms current platform innovation, we need to consider a potential exit strategy.’ (Participant 5)
“You switch platforms when you start losing business from the current platform, as it is outperformed by other platforms.” (Participant 5)
A less direct quote, but also reflecting on the exit mechanism, is the following response to a question related to growth: What happens if the platform performance stagnates?
‘this depends on other competitors, their market share and disruption in the market. We stay open for other opportunities at all times.’ (Participant 7).
4.4 Conceptual model
As a holistic summary of this chapter, a thematic interpretation of the collected data is consolidated in the below conceptual model, figure 10. This model is an interesting guidance towards answering the research question and a good tool to visually clarify the relationships between every sub-factor and every outcome.
Figure 10. Conceptual model; the effect of Platform Performance on TPA retention, while considering platform competition
4.5 Conclusion results
The thematic analysis of the collected data is structurally presented in this chapter and contains the following findings, as summarized in the final conceptual model (figure 10).
First, the qualitative research generated rich data related to the factor ‘platform performance’. The data is validated during the interview process, where four sub-factors emerged: innovation, growth, governance and technology (table 3). These sub-factors are directly related to TPA choice.
Second, the findings related to the main factor of this research, TPA retention, are explained based on three distinct mechanisms: the extend mechanism, the exit mechanism and the hybrid mechanism.
1) This study demonstrated that platform innovation, growth and governance (such as vendor lock-in and engagement model), as presented in figure 10, have a direct effect on the TPA retention. This is called the TPA extend mechanism.
2) Both governance and technological limitations have a direct effect on the TPA exit mechanism, as presented in figure 10. The data also shows that a bad governance of product roadmap or interest in its own product lead to cannibalism and competition between the platform owner and TPA’s, leading to an exit choice.
3) Only governance, as in platform lock-in, is validated to have a direct effect on the hybrid mechanism. Several participants stated that due to the strong lock-in (in case of Salesforce for example) they decided to choose to use multiple platforms. This reduces the platform’s dependency effect.
Third, we learned from the analysed data that platform competition has a direct effect on the relationship between platform performance (especially innovation and growth) and TPA retention. When platform competition appears, the effect of innovation will shift from an extend to an exit mechanism. The latest is true for the effect of growth, which shifts from extend to exit.
46 In other words the competition as expected has a direct influence on the relationship between the primary platform performance and TPA choice to stay or leave a platform. The performance sub-factors that will suffer the most from competition and could be subject to outperformance are innovation and growth (e.g. sales or revenue share).
These results are interesting empirical findings, as it has not been conducted before.
This research adds value to TPA business owners by giving them a detailed clarification of the sub-factors and their precise effect on the TPA choice in relation to three possible outcome mechanisms.
Third Party App developers (TPA) retention has suffered from dearth in the existing literature as well as in the theoretical contribution. This is the primary motivation for this empirical study to investigate TPA retention within the platform business. Seventeen qualitative interviews have been conducted to answer the research question: what is the effect of the platform performance (PP) on the TPA choice to extend, exit or mix several platforms?
The findings suggest that: 1) a well performing platform is a good indicator for a TPA to extend the use of the platform, 2) a bad performing platform is an indicator for a TPA to consider an exit or hybrid strategy and 3) when a competitor platform outperforms the primary platform - in innovation or growth- a TPA should consider an exit strategy.
These findings are significant, as they provide TPAs with detailed insight into the most important performance indicators of a software platform. In addition, the findings support the consideration of an exit strategy, especially when a competitor platform outperforms the primary platform in innovation or growth. From that perspective, the exit strategy consideration means that an app developer should be very strategic about building different app layers. As a result of the increasing platform competition and the severe platform lock-in technique, agility and flexibility of app structure are highly recommended to support a quick and facile exit strategy.
5.1 Answer to the research questions
The conceptual model, illustrated in figure 10, reveals unique and interesting findings, visually summarized by associating the exact independent factor with the exact outcome factor.
What was predicted to be a simple relationship between platform performance and TPA retention (figure 3) reveals significant detailed findings on which performance sub-factor has an effect on which outcome. Furthermore, the platform competition, also predicted in the primary
48 concept, shows significant effect on the relationship between innovation or growth and a TPA retention.
To answer the research question ‘What is the effect of platform performance on the TPA choice to extend, exit or mix several platforms?’, the earlier introduced propositions will be covered individually.
Proposition 1: How would platform performance motivate a TPA to extend participation in a platform?
It is concluded that the three main factors that have direct influence on the TPA extend mechanism are: Innovation, Growth and Governance, as illustrated in figure 7. A good performing platform in one of these factors will encourage a TPA to keep participating.
Proposition 2: How would platform performance be a reason for a TPA to leave a platform?
As demonstrated in section 4.2.2, both governance and technology have a direct effect on the TPA exit mechanism (figure 8). Bad performance in one of these factors will motivate a TPA to consider an exit choice.
Proposition 3: How would platform performance be a reason for a TPA to mix several platforms?
In section 4.2.3, the data analysis concluded that only governance (platform lock-in) could have a direct effect on the TPA hybrid mechanism (figure 9). A TPA that fears the platform lock-in practice will mix platforms or extract separate app components to a second platform. For example, by building an app in Salesforce and using Google for storage.
Proposition 4: Does a competitor platform have an effect on the relationship between platform performance and TPA retention?