IT’S IN THE DATA
A multimethod study on how SaaS-businesses
can utilize cohort analysis to improve marketing
decision-making
Gustav Fridell
Saam Cedighi Chafjiri
Supervisor: Martin Hoshi Larsson Examiner: Jakob Rehme
Copyright
The publishers will keep this document online on the Internet – or its possible replacement – for a period of 25 years starting from the date of publication barring exceptional circumstances.
The online availability of the document implies permanent permission for anyone to read, to download, or to print out single copies for his/her own use and to use it unchanged for non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional upon the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security, and accessibility.
According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement.
For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its www home page: http://www.ep.liu.se/.
Abstract
Incorporating data and analytics within marketing decision-making is today crucial for a company’s success. This holds true especially for SaaS-businesses due to having a subscription-based pricing model dependent on good service retention for long-term viability and profitability. Efficiently incorporating data and analytics does have its prerequisites but can for SaaS-businesses be achieved using the analytical framework of cohort analysis, which utilizes subscription data to obtain actionable insights on customer behavior and retention patterns. Consequently, to expand upon the understanding of how SaaS-businesses can utilize data-driven methodologies to improve their operations, this study has examined how SaaS-businesses can utilize cohort analysis to improve marketing decision-making and what the prerequisites are for efficiently doing so.
Thus, by utilizing a multimethodology approach consisting of action research and a single caste study on the fast-growing SaaS-company GetAccept, the study has concluded that the incorporation and utilization of cohort analysis can improve marketing decision-making for SaaS-businesses. This conclusion is drawn by having identified that:
❖ The incorporation of cohort analysis can streamline the marketing decision-making process; and
❖ The incorporation of cohort analysis can enable decision-makers to obtain a better foundation of information to base marketing decisions upon, thus leading to an improved expected outcome of the decisions.
Furthermore, to enable efficient data-driven marketing decision-making and effectively utilize methods such as cohort analysis, the study has concluded that SaaS-businesses need to fulfill three prerequisites, which have been identified to be:
(1) Management that support and advocate for data and analytics;
(2) A company culture built upon information sharing and evidence-based decision-making; and
(3) A large enough customer base to allow for determining similarities within and differences between customer segments as significant.
However, the last prerequisite applies specifically for methods such as or similar to cohort analysis. Thus, by utilizing other methods, SaaS-businesses might still be able to efficiently utilize data-driven marketing decision-making, as long as the first two prerequisites are fulfilled.
Table of Contents
1 Introduction ... 6
1.1 SaaS and the subscription-based pricing model ... 6
1.2 The importance of data and metrics in SaaS-businesses ... 7
1.3 Turning data into actionable insights using cohort analysis ... 8
1.4 Utilizing data for marketing decision-making ... 8
1.5 GetAccept and the uprising of digital sales enablement services... 9
1.6 Defining “improved” decision-making ... 10
1.7 Problem description and purpose ... 10
2 Theory ... 12
2.1 The value of information in decision-making ... 12
2.2 The marketing decision-making process ... 13
2.3 The importance of data-driven decision-making in marketing ... 14
2.4 Prerequisites for efficient data-driven marketing decision-making ... 15
2.5 Guidelines for ensuring managerial support for data and analytics through efficient integration into marketing decision-making ... 16
2.6 Service retention and the issue of limited external data availability ... 19
3 Cohort analysis ... 20
3.1 Using cohort analysis to derive actionable insights from readily available customer data20 3.2 Seven-step iterative framework for hypothesis testing with cohort analysis ... 22
3.3 Segmentation and selection process of metrics with the Three Engines of Growth ... 23
4 Methodology ... 31
4.1 Multimethodology approach ... 31
4.2 Single case study ... 32
4.3 Action research ... 35 4.4 Semi-structured interviews ... 37 4.5 Internet surveys ... 38 4.6 Research quality ... 41 5 Empirics... 43 5.1 Unfreezing ... 43 5.2 Changing ... 49 5.3 Refreezing ... 53 6 Analysis ... 57
6.1 RQ 1: Prerequisites for efficient data-driven marketing decision-making ... 57
6.2 RQ 2: Effects of incorporating cohort analysis into marketing decision-making ... 61
7 Conclusions ... 63
8 Discussion... 64
9 References ... 65
10 Appendices ... 70
A. Interview guide for semi-structured interviews during the unfreezing phase ... 70
B. Internet survey regarding GetAccept’s fulfillment of factors for ensuring a data-enabling culture ... 72
List of tables
Table 1 – Jocumsen’s (2004) summary of the different steps included in the marketing
decision-making process ... 13
Table 2 – Overview of relevant metrics for the Sticky Engine ... 25
Table 3 – Overview of relevant metrics for the Virality Engine ... 27
Table 4 – Overview of relevant metrics for the Paid Engine ... 28
Table 5 – Overview of different research methods and their relevance depending on the situation (Yin, 2009) ... 32
Table 6 – Troisi et al.’s (2019) synthetization of Cumming et al.’s (2016) summary of the main variants of Lewin’s (1947) change model ... 36
Table 7 – Fowler’s (2014) synthesis of advantages and disadvantages of internet surveys ... 41
Table 8 – Fowler’s (2014) synthesis of advantages of using a self-administered and/or computer-assisted survey ... 41
Table 9 – Overview of semi-structured interviews during the unfreezing phase ... 43
Table 10 – Overview of answers to internet survey regarding GetAccept’s fulfillment of factors for ensuring a data-enabling culture ... 45
Table 11 – Overview of the conducted cohort analyses ... 49
Table 12 – Results from the Touch versus No Touch customers cohort analysis ... 50
Table 13 – Results from the customer size cohort analysis ... 51
Table 14 – Results from the engaged versus unengaged customers cohort analysis ... 52
Table 15 – Results from the sales versus HR customers cohort analysis ... 52
Table 16 – Results from the Sales Presentation versus Contract Focused customers cohort analysis ... 53
Table 17 – Overview of semi-structured interviews during the refreezing phase ... 54
Table 18 – Overview of GetAccept’s fulfillment of factors for ensuring a data-enabling culture .... 59
Table 19 - Overview of GetAccept’s adherence to guidelines for ensuring managerial support for data and analytics through efficient integration into marketing decision-making... 60
List of figures
Figure 1 –Illustration of Guo & Ma’s (2018) explanation of the differences between the traditional perpetual pricing model and the subscription-based SaaS pricing model ... 71
Introduction
Chapter 1, Introduction, gives an overview of SaaS-businesses and the subscription based pricing model, the topic of cohort analysis within marketing decision-making, and presents the purpose and research questions of the study.
1.1
SaaS and the subscription-based pricing model
Guo & Ma (2018) describe how the emergence and growth of Software as a Service (SaaS) has fundamentally changed how software can be delivered, used and managed. Instead of licensing software perpetually and charging for it once, SaaS-vendors host the software, continuously upgrade the quality of it and charge customers on a recurring basis (Guo & Ma, 2018). The SaaS distribution model is increasingly becoming the de facto standard, something that can be seen in a study conducted by Blissfully (2019) which states that 68% of organizations are mostly or completely SaaS-driven at this point, and 23% operate solely using SaaS apps today. Guo & Ma (2018) describe that one of the main characteristics that sets SaaS apart from traditional perpetual software distribution is the pricing model. The authors explain that the perpetual model is defined by charging for the software once where SaaS-vendors instead use a pay-as-you-go, subscription-based model. Furthermore, by adopting this model, SaaS-vendors can tier the pricing of the software to the quality and usage of it, thus enabling more customer centric pricing (Gohad, Narendra, & Ramachandran, 2013). This implies that if the quality of the software increases over time or the customer want to increase their usage level of it, the vendor can match their pricing accordingly.
What can be drawn from Guo & Ma’s explanation of the different models is that by using the traditional perpetual model to turn a profit, the price of the software needs to be set higher than what it costs to acquire a customer. In contrast, by using a subscription-based model in SaaS, the price needs to be set so that the sum of all future recurring payments will be higher than the costs of acquiring a customer. One advantage of the subscription-based model in SaaS is thereby that, if the customer remains for longer than the payback period, the potential earnings from it increases indefinitely. This is illustrated in Figure 1 below where the subscription-based SaaS pricing model is able to deliver higher accumulated profits over time, but is not able to pay back the customer acquisition cost as quickly as the traditional perpetual pricing model.
Figure 1 –Illustration of Guo & Ma’s (2018) explanation of the differences between the traditional perpetual pricing model and the subscription-based SaaS pricing model
In addition to the subscription-based pricing model, another aspect that sets SaaS apart from other industries is the reliance on industry-specific data and metrics (Areskoug, 2020).
1.2
The importance of data and metrics in SaaS-businesses
Lord Kelvin famously said during a lecture that “If you can not measure it, you can not improve it” (1883). Ever since, this mindset has become increasingly important, and is nowadays one of the most important foundations for decision-making in today’s businesses (Croll & Yoskovitz, 2013). This is described by Bernard Marr (2015), where he analyzed how some of the world’s biggest companies such as Apple, Amazon and Google use data to improve their businesses. By using data, Marr explains how the most successful companies nowadays have a factual foundation for every decision they make, meaning they have the ability to improve every aspect of the organization in a way that just a few years ago was not possible.
As explained in the SaaS pricing model by Guo & Ma (2018) above, SaaS differs from other businesses because the revenue for the service comes over an extended period of time, the lifetime of the customer. Prevalent SaaS entrepreneur David Skok (2018) expands on this subject by emphasizing that traditional business metrics fail to capture the key factors that drive performance in SaaS-businesses, such as customer acquisition cost (CAC) and customer lifetime value (LTV). Skok explains that understanding these metrics, among others, is crucial for making the right decisions that will bring a SaaS-business forward. Fortunately, due to hosting and supplying the software, Areskoug (2019) states that SaaS-businesses have better preconditions than any other type of business to track, measure, and utilize these
metrics, something commonly done using the analytical framework of cohort analysis.
1.3
Turning data into actionable insights using cohort analysis
A cohort is generally defined as “a group of individuals having a statistical factor
(such as age or class membership) in common in a demographic study”
(Merriam-Webster, 2020). SaaS industry expert Ben Murray (2019) explains how the definition differs a bit for SaaS-businesses, where a cohort instead refers to a subset of customers who share a common characteristic, usually the month of acquisition for the vendor.
Cohort analysis is explained by Murray (2019) as an analytical framework that is used to study these cohorts and show how their behavior such as retention or service spending decrease or increase over time. In corroboration with Murray, Schweidel et al. (2008) state that while retention is a key construct for service-providers, they seldom have enough external data to fully understand their customers’ behaviors. The solution for this, the authors argue, is to use cohort analysis since it enables companies to utilize their internal data in favor of their limited external data to understand the underlying factors of the customers’ behaviors. Murray (2019) further elaborates on this, stating that by understanding the reasons behind the customers’ behaviors, SaaS-businesses can obtain actionable insights to, for example, improve customer retention and customer lifetime value.
Even though cohort analysis is just one out of many methods businesses can use to gain insights about their customers, the general concept of utilizing data has over time become increasingly important within marketing decision-making.
1.4
Utilizing data for marketing decision-making
Traditionally, companies used to base their decision-making on internal knowledge and “gut-feelings”, where HiPPOs (Highest Paid Person’s Opinions) would lay the foundation of the decision and validate it (Brynjolfsson, Hitt, & Kim, 2011). With today’s vast availability of data, companies can now gather detailed knowledge which reveals extensive insights about their customers and competitors (Brynjolfsson, Hitt, & Kim, 2011). Thus, the pressure has also increased on managers to justify their marketing decision-making in regards of the value they create and the performance shown (Jeffery, 2010). This is where the data comes in.
Croll & Yoskovitz (2013) argue that all of the previous assumptions, intuitions and gut-feelings will wither when data and analytics (i.e., the discovery, interpretation, and communication of meaningful patterns in the data) are applied, and that data-driven decision-making is a cornerstone for success moving forward. This is further strengthened by Jeffrey (2010) who states that organizations which embrace and adapt marketing metrics and a data-driven marketing culture have a competitive
advantage and show significantly better financial performance than their competitors.
However, since no company is another one equal, it is important to understand its specific context to see how data can be incorporated, utilized, and capitalized upon. This especially since there is clear evidence of organizational prerequisites for efficiently utilizing data and analytics in marketing decision-making, as stated by Johnson et al. (2019), LaValle et al. (2011), and Wedel & Kannan (2016).
1.5
GetAccept and the uprising of digital sales enablement
services
The SaaS-company GetAccept was founded with support from the startup incubator Y-combinator in 2015 and has since had a vigorous growth journey (Areskoug, 2019), being named the 4th fastest growing SaaS company in the world (SaaS Mag, 2019).
Today, GetAccept has a yearly revenue of $4.5 million along with 100 employees, up from 30 one year prior (Areskoug, 2019).
GetAccept’s business is within sales enablement where they provide a service helping customers close more deals, specifically by providing a digital platform simplifying the contract writing, managing, tracking, and signing procedure (Areskoug, 2019). Through the rise of digitalization, multiple companies provide similar services such as GetAccept, meaning the competition becomes more intense over time (Areskoug, 2019). Therefore, being able to efficiently acquire new customers and retain existing ones becomes increasingly important as time goes on (Thulin, 2020). However, maintaining growth is not easy, and GetAccept does not differ. The individual efficiency of each sales person has gone down and more importantly, service retention has not been kept at a desirable level (Areskoug, 2020). Skok (2018) uses the analogy of a “leaky bucket” which explains how in SaaS, if you cannot retain your customers, after a while it does not matter if you acquire new ones since the additional revenue from new customers will not compensate for the loss of revenue from lost customers. Hence, Skok argues that the key factors of running a successful SaaS business is not only to (1) acquire new customers, but more importantly also to (2) retain customers, and (3) monetize customers.
The analogy of the leaky bucket is easily understood if set in relation to the pricing model of SaaS-business explained by Guo & Ma (2018). The pricing model shows that there is a payback time for turning a profit on a customer in SaaS, meaning if they churn too early, the company will have made a net loss on that customer. On the other hand, if the customer stays for longer than the payback time, there is an infinite upside to the profit for the company.
By combining GetAccept’s less than desirable service retention and the analogy of the leaky bucket, it is evident why one of GetAccept’s main priorities is to understand how they can improve their marketing decision-making in order to improve the retention of their customers.
1.6
Defining “improved” decision-making
In order to analyze how the incorporation of cohort analysis can improve marketing making, it is important to define what constitutes “improved” decision-making. Since “improved” hardly can be discretely defined, this study takes inspiration from the definition used in the traditional economic and mathematical models of Game Theory, originally introduced by John von Neumann (1928) and expanded upon by Neumann and Morgenstern (1944; 1947). Neumann and Morgenstern’s work introduced concepts such as expected utility, which allowed for studying the strategic interaction among rational decision-makers under uncertainty (Myerson, 1991). Hence, the definition of “improved” decision-making this study uses is:
Decision-making which, under altered circumstances (i.e., new information presented), has a better expected outcome compared to the original state.
Similar to the concept of “improved” being hard to define, the concept of “better outcome” is inherently subjective and can be defined in a magnitude of ways (higher revenue, more customers, better retention, etc.), which is why this will be defined in conjunction with GetAccept. Note that this definition is based upon the expected outcome, which means that the actual outcome does not need to be measured, instead GetAccept’s expectations of the outcome is sufficient.
Therefore, to establish if cohort analysis can improve marketing decision-making, the opinions of decision-makers at GetAccept will be analyzed in alignment with the definition stated above. However, due to the obvious pitfalls of only relying on subjective opinions, the study will also analyze how GetAccept view their access to information change as a result of incorporating cohort analysis into marketing decision-making. This, since literature suggests that making decisions on a more informed basis usually yields a better expected outcome, as will be explained in chapter 2.1. Thus, if it can be said that GetAccept have a better foundation of information to base decision-making upon, it can be argued that its decision-making has improved. Lastly, since decision-making not only can be improved from a result-oriented perspective, but also from a processual perspective, changes in GetAccept’s marketing decision-making process will also be analyzed, as a result of the incorporation of cohort analysis.
1.7
Problem description and purpose
As presented earlier, utilizing data and analytics within marketing decision-making is today crucial for a company’s success, this especially for SaaS-businesses considering the subscription-based pricing model and the analogy of the leaky bucket. One way to do this, which SaaS industry experts such as David Skok (2018) and Ben Murray (2019) describe as the industry standard is to utilize the analytical framework of cohort analysis. However, as noted by Johnson et al. (2019), LaValle et
al. (2011), and Wedel & Kannan (2016), organizational prerequisites exist for efficient utilization of data-driven methods such as cohort analysis. Thus, there is of interest to examine how cohort analysis can be used to improve marketing decision-making for SaaS-businesses and investigate what the prerequisites are for efficiently doing so. Hence, the purpose of this study is to:
Examine how SaaS-businesses can utilize cohort analysis to improve marketing decision-making.
This will be investigated through:
❖ Research question 1 – What are the prerequisites for efficient data-driven
marketing decision-making?
❖ Research question 2 – How does the incorporation and utilization of
cohort analysis affect marketing decision-making?
The study is conducted upon the fast-growing SaaS-company GetAccept, with the aim of drawing conclusions applicable for SaaS-businesses in general.
2
Theory
Chapter 2, Theory, presents the theoretical frame of reference on which the empirics and analysis build upon. The chapter discusses areas including data-driven marketing decision-making, prerequisites, and guidelines to succeed with it, along with an introduction to the area of service retention to emphasize the value of cohort analysis.
2.1
The value of information in decision-making
As introduced in chapter 1.6, one of the aspects to be analyzed is how GetAccept’s access to information change as a result incorporating cohort analysis into their marketing decision-making. The reason being the logical assumption that gaining access to more informative information should be beneficital in a decision-making setting. However, this logical assumption is also supported by academic evidence.
In their article regarding the effects of data-driven decision-making on company performance, Brynjolfsson et al. (2011) state that the seminal work of Blackwell (1953) has laid the foundation for modern theories regarding the value of information. One of Blackwell’s contributions to this field was proving that a rational decision-maker should achieve a higher expected payoff if acting under one imperfect information set better (i.e., “more informative”) than another one. In this perspective, Brynjolfsson et al. (2011) state that improved information always, at least slightly, improves company performance.
Blackwell’s theories are of course theoretical which is why it can be argued that the conclusions do not always apply in practice. However, Porat & Haas (1969) studied this very phenomenon of the impact of information on decision-making and concluded that, in practice, more information does lead to more accurate levels of goal-setting and decision-making.
Of course, it can be argued that an increased amount of information does not always lead to better decisions. For example with the case of information overload, when the decision-maker receives more information than it can process (Milford & Perry, 1977), it is likely a reduction in decision quality will occur (Speier, Valacich, & Vessey, 1999). However, it is hard to argue that cases where the information set about a phenomenon goes from nothing to something, for example with the help of cohort analysis, constitutes as information overload.
Therefore, in the context of this study, the theory suggests that gaining access to more informative information generally does lead to improved decision-making, strategic or otherwise. Furthermore, in a strategic setting, it is generally agreed upon that information leads to better decision-making (e.g. through an increased competitive advantage), as described by prominent authors Michael Porter and Victor Millar (1985).
2.2
The marketing decision-making process
As mentioned in chapter 1.6, one asepct to be analyzed is how the incorporation of cohort analysis affects GetAccept’s marketing decision-making process. The purpose being that if improvements in the process can be observed, it can be argued that GetAccept’s decision-making has improved. However, to properly conduct such an analysis and draw such conclusions, an understanding of the marketing decision-making process is necessary.
Harrison & Pelletier (1996) describe strategic decisions as highly complex and important because of their impact and/or long-term implications on the organization. Kotler (2000) describes strategic marketing decisions as management of portable relationships, attraction of new customers and delivering satisfaction to them. Furthermore, Kotler (2000) and Corman et al. (1996) underline the importance of strategic marketing decisions since they are a prerequisite for companies to survive, grow, and deliver satisfactory results.
In order to understand how companies can improve their marketing decisions, Jocumsen (2004) states that it is important to have an understanding of the strategic decision-making process. He further argues that all strategic decision-making processes, marketing or not, generally include the same steps. Thus, by reviewing both strategic and marketing literature, Jocumsen has synthesized this process into the steps which are presented in Table 1 below, along with the supporting literature.
Table 1 – Jocumsen’s (2004) summary of the different steps included in the marketing decision-making process
Steps Literature support
Acknowledgement of existence
(Johnson & Scholes, 1999), (Aaker, 1998), (Ferrell, Lucas, & Luck, 1994), (Wheelan & Hunger, 1992), (Fahley, 1981)
Decision emergence / need
(Aaker, 1998), (Stacey, 1996), (Robbins, 1994), (Nutt, 1993), (Van de Ven, 1992), (Vecchio, Hearn, & Southey, 1992), (Mintzberg, 1987)
Diagnosis / initial
intelligence gathering (Johnson & Scholes, 1999), (Nelson & Quick, 1997) Decision criteria and
weights (Nelson & Quick, 1997), (Robbins, 1994)
Data collection / analysis (Johnson & Scholes, 1999), (Ivancevich, Olekalns, & Matteson, 1997), (Mintzberg, 1987)
Alternatives development / evaluation
(Johnson & Scholes, 1999), (Aaker, 1998), (Mowen & Gaeth, 1992)
Final detailed assessment (Johnson & Scholes, 1999), (Aaker, 1998), (Harrison E. , 1996)
Commitment (Johnson & Scholes, 1999), (Aaker, 1998), (Eisenhardt & Zbaracki, 1992), (Quinn, 1980)
Are steps conducted sequentially or iteratively?
Iterative: (Carson, Cummins, O'Donnell, & Grant, 1998), (Gibb
& Scott, 1985), (Van Hoorn, 1979). Sequential: (Johnson & Scholes, 1999), (Ferrell, Lucas, & Luck, 1994), (Aaker, 1998)
Furthermore Jocumsen (2004) states that company-specific factors such as decision type, manager characteristics, and business characteristics all influence the strategic decision-making process. This implies the process can differ between companies, because of which GetAccept’s specific practices will be outlined in conjunction with decision-makers at GetAccept.
2.3
The importance of data-driven decision-making in
marketing
Since the focus of this study is to evaluate the benefits of cohort analysis in a marketing setting, it is important to understand the value such data-driven methods can provide for organizations.
Jie Lu et al. (2019) state that to overcome the challenges that are present in the big data era, both researchers and practitioners emphasize the importance of making decisions that are backed up by data. This is exemplified by prominent marketing author George S. Day (2010) who says that “The first contributor to a superior
growth record is deep market insights and foresight.” and by Google’s ex-CEO and
executive chairman Eric Schmidt (2014) who says that you should “Bet on technical
insights, not market research.”
Moreover, Johnson et al. (2019) underline the benefits of using data to drive decisions specifically in the field of marketing. The authors state that by analysis of large amounts of data, marketers can make useful inferences about customers and competitors, such as better understanding their costs, sales potential, and emerging marketplace opportunities. This is further corroborated by Brynjolfsson et al. (2011) who, through a quantitative study of 170+ publicly traded firms, found that the ones that adopted data-driven decision-making showed higher productivity and market value than the ones that did not.
Jeffery (2010) further emphasizes the importance of data-driven marketing by pointing out that managers, more and more need to (1) justify their marketing
spending, (2) show the value they create for the business, and (3) radically improve marketing performance. One crucial key to enable this in marketing, Jeffery argues,
is to understand some essential metrics, such as churn, customer satisfaction, profit,
payback, and customer lifetime value. A further explanation of these metrics, among
2.4
Prerequisites for efficient data-driven marketing
decision-making
As stated in chapter 1.4, organizations differ on several aspects, and likewise their ability to efficiently utilize data in decision-making. Thus, in order to achieve the purpose of this study, it is important to identify what organizational prerequisites might exist for efficiently being able to utilize methods such as cohort analysis in marketing decision-making.
Johnson et al. (2019), LaValle et al. (2011), and Wedel & Kannan (2016) all describe what they consider to be the most important aspects in an organization’s (in Johnson et al.’s and Wedel & Kannan’s case a marketing organization’s) journey to become data-driven. Although these perspectives differ somewhat, all of them recognize two distinct prerequisites for being successful in this journey: (1)
management that support and advocate for data and analytics, as well as (2) a company culture built upon information sharing and evidence-based decision-making.
2.4.1 Ensuring managerial support for data and analytics
Young & Jordan (2008) state that top management support is the most important factor for project success. This holds true for the path to become data-driven, as LaValle et al. (2011) describe lack of managerial support as one of the biggest obstacles in adopting analytics.
By conducting a survey of nearly 3 000 executives, managers and analysts working across more than 30 industries and 100 countries, LaValle et al. (2011) have identified a strong correlation between company performance and analytics-driven management. This is reinforced by earlier presented theory stating the correlation between utilizing data-driven insights and company performance (Brynjolfsson, Hitt, & Kim, 2011), in conjunction with Johnson et al. (2019) who describe how support from company management is a requirement for being able to exploit the advantages of analytics. This is further reinforced by Wedel & Kannan (2016) who describe that even though technology, organizational structure, and skilled analysts are all requirements for utilizing data, analytics, and data-driven decision-making, this cannot happen without leaders that recognize the importance of it.
To ensure managerial support, LaValle et al. (2011) state that you need to prove that the transformation of becoming data-driven meets each of these three critical management needs: (1) reduced time to value, (2) increased likelihood of
transformation that is both significant and enduring, and (3) greater focus on achievable steps. The authors further present five guidelines, outlined in chapter 2.5,
which enables the fulfillment of these requirements.
2.4.2 Ensuring a data-enabling culture
Wedel & Kannan (2016) state that a company-wide culture of evidence-based decision-making is the primary precondition to efficiently adopt analytics and
become data-driven. This is also stated by LaValle et al. (2011) who put the importance of a data-enabling culture on par with the importance of getting managerial support, as described earlier. This data-enabling culture is exemplified by Johnson et al. (2019) as one that does not rely on conclusions without data to back it up, by LaValle et al. (2011) as one that encourages the sharing of information, and by Wedel & Kannan (2016) as one that can be summarized with the cliché saying of “In God we trust; all others must bring data.”.
Stater et al. (2011) state that overall firm performance is influenced by how well the cultural orientation complements its strategy. However, enabling such a culture is no small task, as described by Johnson et al. (2019) who state that to successfully transform into a data-enabling culture, a shift may be required in how the organization thinks about itself holistically. To counter this issue, Grossman & Siegel (2014) have developed a framework for organizing and enabling analytics within an organization. The framework includes four factors which help facilitate a data-enabling organizational culture, and is summarized below:
(1) Analytics should be viewed as an organizational function and there should be an analytics department or unit to support this function;
(2) Analytics should be integrated into the corporate strategy; (3) Senior leaders should advocate for analytics; and
(4) Data, both internal and external, that can provide value should be used.
2.5
Guidelines for ensuring managerial support for data and
analytics through efficient integration into marketing
decision-making
As mentioned in chapter 2.4.1, managerial support for data and analytics is ensured by proving that the transformation of becoming data-driven meets the needs of management. Throughout the literature review, several important guidelines were identified to succeed with this, and while no author provided a literature-based synthesis of these factors, LaValle et al.’s (2011) interpretation was the most comprehensive. Thus, LaValle et al.’s five guidelines along with the other authors’ additions are presented below, and include to: (1) focus on the biggest and
highest-value opportunities, (2) within each opportunity, start with questions, not data, (3) embed insights to deliver value, (4) keep existing capabilities while adding new ones, and (5) use an information agenda to plan for the future.
2.5.1 Focus on the biggest and highest-value opportunities
LaValle et al. (2011) state that while starting to adopt data-driven decision-making, companies should start with their highest priority challenge where analytics will have
an impact. This aligns with Bose (2009) who states that the choice of first initiative is critical and must be in a potentially high-impact area to be able to act as a catalyst for accelerating subsequent adoption of analytics.
LaValle et al. (2011) argue that by prioritizing this way, this will: (1) make it easier
to get management invested, since it is a priority for them, and (2) make it easier to justify the initiative, since the potential reward is significant and contributes to major
business goals. This is reinforced by Aldea et al. (2019) who describe how organizations prioritize projects on the basis of how well they are expected to directly and efficiently contribute to the achievement of organizational goals, given resource and capability constraints.
2.5.2 Within each opportunity, start with questions, not data
LaValle et al. (2011) describe how one common pitfall during the process of adopting analytics is that companies often start by gathering all available data before starting the analysis. The authors describe that this often leads to an all-encompassing focus on data management which leaves little time, energy, and resources to understand its potential uses. This is also illustrated by Sherman (2015) who describes how too much data can inhibit an organization’s ability to properly analyze and act upon it.
LaValle et al. (2011) argue that this obstacle of data overflow can be overcome by first defining the insights and questions needed to meet the business objective, and then identifying which data is needed.
2.5.3 Embed insights to drive actions and deliver value
LaValle et al. (2011) and Bose (2009) emphasize the importance of making insights actionable to be able to obtain value from them. Bose elaborates on this by stating that when sharing the results with managers and decision-makers, they need to be delivered in a format that even non-technical users can understand and by using everyday business terms to explain the results. This is corroborated by LaValle et al.’s (2011) view of actionable insights, which they describe as such that can be readily understood and acted upon.
Bose (2009) states that a key for successfully making insights actionable and valuable to the end-user is to make sure they are simple, concise, readable, and usable. Bose mentions a couple of methods on how to achieve this, which includes using support tools like dashboards, reports, and visualization systems. This aligns with LaValle et al.’s (2011) more explanatory description on how to achieve this, which includes the following techniques along with examples:
❖ Dashboards – Use dashboards to visualize not only last-quarter sales, but also
show predictions on what sales could be next quarter under certain conditions, such as a new media mix, a price change, a larger sales team, or even a major weather or sporting event.
❖ Automatic recommendations – Based on different scenarios, automatically
recommend the best approaches, such as the best media mix to introduce a specific product to a specific segment, or the ideal number of sales representatives to assign to a new market.
❖ Use cases – Produce use cases to illustrate how to embed insights into business
applications and processes.
2.5.4 Keep existing capabilities while adding new ones
LaValle et al. (2011) state that during the process of becoming data-driven through the adoption of analytics, analytics capabilities are added upstream at increasingly central levels of management. During this process, the authors argue that it is important not to subtract or deprioritize existing capabilities, but instead to focus on incorporating analytics into those core disciplines.
The authors state that one way to achieve this incorporation of existing capabilities and analytics is through a centralized analytics unit or department which share analytics resources with other parts of the organization. Sherman (2015) also advocates for such units along with Wedel & Kannan (2016) who argue that they help prioritize opportunities, obtain resources, ensure access to data and software, facilitate the deployment of models, develop necessary expertise, ensure accountability, and coordinate team effort.
2.5.5 Use an information agenda to plan for the future
LaValle et al. (2011) state that a majority of organizations have more data than they know how to use effectively, and that to effectively use this data, it must be molded into an information foundation that is integrated, consistent, and trustworthy.
The authors describe how such a foundation can be achieved through an information agenda that provides a high-level road map for information which includes areas such as information governance policies and tool kits, data
architecture, data currency, data management, integration and middleware, and analytical tool kits based upon user needs. The authors further underline the
agenda’s importance, by describing it as a key enabler for analytics initiatives, since it provides the right information and tools at the right time based upon business-driven priorities.
The argument for using an information agenda to plan for the future while implementing analytics is strengthened by Bose (2009) who describes that the introduction of analytics in an organization must be managed carefully and incrementally.
2.6
Service retention and the issue of limited external data
availability
To understand how cohort analysis can be beneficial for SaaS-businesses, it is important to understand the concept of service retention and how it can be affected. This since cohort analysis is based upon analyzing different customer segments’ retention patterns.
Schweidel et al. (2008) describe retention as a key construct for contractual service providers. The authors illustrate this by arguing it is essential for determining the value of existing and future subscriptions, and for making resource allocation decisions. Previous research has shown that service retention is, among other factors, positively affected by:
❖ Increased tenure (i.e., negative duration dependence) (Reichheld, 1996); ❖ Perceptions of service quality (Zeithaml, Berry, & Parasuraman, 1996); ❖ Customer satisfaction (Bolton, 1998); and
❖ Marketing activities (Lewis, 2005) such as loyalty programs and short-term promotions (Lewis, 2004).
Moreover, service retention has also shown to be affected by:
❖ Unobserved differences across subscribers, such as subscriber heterogeneity (e.g. having different preferences or needs) (Morrison & Schmittlein, 1980); ❖ Subscribers’ starting periods, for example different months (Schweidel, Fader,
& Bradlow, 2008); and
❖ Seasonality in retention patterns (Radas & Shugan, 1998).
Schweidel et al. (2008), however, argue that even though these studies have furthered the overall understanding of the antecedents of service retention and the consequences based on these factors, in reality, the external information needed to apply these insights is seldom available. Schweidel et al. state that most of the time, the data available is only the number of subscribers to a particular service, which poses difficulties in accurately modelling or forecasting service retentions patterns.
To counter this issue of limited external data availability, Schweidel et al. (2008) suggest using a method called cohort analysis which with limited external information, still enables the company to obtain actionable insights on customer behavior and retention patterns.
3
Cohort analysis
Chapter 3, Cohort analysis, presents an introduction to the analytical framework of cohort analysis as well as its uses. The chapter discusses how the framework is utilized in conjunction with an iterative version of the traditional scientific method, along with relevant metrics to build the analysis upon.
3.1
Using cohort analysis to derive actionable insights from
readily available customer data
As introduced in chapter 1.4 and discussed in chapter 2.3, using methods such as cohort analysis to utilize data within marketing decision-making has become increasingly important over time for SaaS-businesses. However, since the purpose of this study is to primarily analyze the effects of cohort analysis on marketing decision-making, only an overview of each conducted analysis will be presented in the empirics, while a more in depth explanation of the framework is presented in this chapter.
In alignment with Schweidel et al. (2008), Murray (2019) describes cohort analysis as an analytical framework which is used to study groups with similar characteristics over time. By utilizing cohort analysis, Murray explains that SaaS-businesses can obtain insights on the reasoning behind customer behavior related to for example service spending and retention, which then can be used to improve those parameters. This is expanded upon by Croll & Yoskovitz (2013) who explain that for SaaS-businesses, cohorts are usually segmented by month of which the customers are acquired and compared over time to gain an understanding of how different onboarding experiences relate to each cohort’s retention and spending patterns.
To properly conduct a cohort analysis and ensure valid results, Schweidel et al. (2008) describe a general working methodology that can be synthesized into four steps: (1) segment the customers, (2) plot the cohorts over time, (3) study the plots, and (4) examine underlying factors to explain factors found.
3.1.1 Step 1: Segment the customers
Firstly, Schweidel et al. (2008) suggest segmenting the customers by date of acquirement, thus creating the cohorts. This can for example be done on a monthly, quarterly, or yearly basis. Furthermore, the authors state that the reason for this segmentation lies in that different cohorts may have different properties and are thereby needed to be examined separately.
However, before creating the cohorts, the customers can be segmented by additional parameters such as geographical market or type of industry, to allow for comparison between the segments.
3.1.2 Step 2: Plot the cohorts over time
Secondly, Schweidel et al. (2008) suggest plotting each cohort over time to enable for examining how they change. As stated earlier, the authors describe retention rates as a key construct for service providers and is thereby the preferred variable of examination for the cohorts. Skok (2018) describes two main ways of creating these retention plots: (1) examine the revenue retention within each cohort over time, and (2) examine the customer retention rate within each cohort over time.
Examining the customer retention gives the business insights on how many of their customers within each cohort are staying over time. Similarly, revenue retention gives insights on how the revenue of each cohort is changing over time. For illustrative purposes, Figure 2 below showcases a customer retention plot.
Figure 2 – Example of a customer retention plot (Schweidel, Fader, & Bradlow, 2008)
In Figure 2 above, the business uses monthly contracts which in turn makes changes in retention visible after each month. However, businesses are not limited to monthly contracts and can for example use yearly instead, which in turn would make changes in retention visible only after each year passed. In that case, depending on the age of the business along with the average lifetime of its customers, using visual plots can be superfluous and it is instead sufficient to compare the raw retention numbers of the cohorts after each year passed.
3.1.3 Step 3: Study the plots
Schweidel et al. (2008) then instruct to study the patterns of the plots to identify trends or similarities. As an example, the authors mention that by studying the plot in Figure 2, there is a clear indication that the customers seem less likely to leave the service within the first three months, whereas the retention rate decreases significantly during the following months and stabilizes towards the end.
3.1.4 Step 4: Examine underlying factors to explain patterns found
To explain the patterns found in the previous step, Schweidel et al. (2008) propose to analyze each cohort in regards to potential underlying factors or drivers. For this, the authors suggest five probable factors which are presented below but emphasize that these might vary depending on the context.
❖ Duration dependence – Allows for the service churn rate for a subscriber to
vary according to the length of time they have had the service.
❖ Time-varying marketing activity – Activities such as promotions and its
effects on the cohort.
❖ Subscriber heterogeneity – The varying retention rates across subscribers and
their cohorts.
❖ Cross-cohort effects – Factors which can have an affect across multiple
cohorts.
❖ Calendar-time effects – Effects that can be linked to the time of the year in
which the customer was acquired (e.g. seasonality).
3.2
Seven-step iterative framework for hypothesis testing with
cohort analysis
In order for the cohort analysis to be conducted in an efficient manner, Murray (2019) proposes a seven-step framework which is a modification of the traditional scientific method. The framework contains the following steps:
(1) Question – Define the objective of the analysis. This can for example be to question why customers are churning.
(2) Research – Determine what information is already known. Murray states that almost all businesses have anecdotal knowledge about their customers, and that it is important to utilize it in the next steps.
(4) Experiment – Test the hypothesis.
(5) Observations – Collect data during the experiment. Murray explains this as the heart of the cohort analysis and is essentially what the analysis depends upon.
(6) Results/Conclusions – Determine if the hypothesis was correct. (7) Communicate – Share the results with internal stakeholders.
The scientific method is defined as a systematic procedure to gain knowledge through formulating a problem, collecting data and to formulate and test hypotheses (Merriam-Webster, 2020). Traditionally, the scientific method has been linear, meaning that if the hypothesis has been verified by an empirical test, it can be assessed as true (Anderson, 1983). Despite this, no finite number of empirical tests can guarantee the truth of the statement (Black, 1967). To resolve this issue, Carnap (1953) introduces the idea of “gradually increasing verification” which proposes an iterative version of the scientific model. The iterative version tests the hypothesis though an empiric test, and based on the results the hypothesis it is either tentatively accepted or declined, resulting in a new hypothesis being formed (Savitt, 1980; Zaltman, Pinson, & Reinhard, 1973).
Therefore, applying the iterative version of the scientific model to the framework presented by Murray (2019), steps 3 through 6 should be iterated over to ensure proper verification of the hypothesis before communicating the results with internal stakeholders.
3.3
Segmentation and selection process of metrics with the
Three Engines of Growth
As mentioned in chapter 2.3, Jeffery (2010) underlines the importance of tracking different metrics in order for the businesses to be able to efficiently utilize data in their operations. Skok (2018) adds to this by stating that traditional business metrics fail to capture the key factors that drive performance in SaaS-businesses, thus highlighting the importance understanding of some key SaaS-metrics, and their relevance when utilizing cohort analysis.
In order to identify which metrics are of relevance, Croll & Yoskovitz (2013) state that it is important for businesses to: (1) understand what makes a good metric and (2) which metrics to use. To tackle the first obstacle, the authors have defined four characteristics which make a metric good:
(1) A good metric is comparative – The ability to compare a metric over time, user groups or competitors to understand change.
(2) A good metric is understandable – It is crucial to be able to understand the metric in order to be able to take actions upon it.
(3) A good metric is a ratio or a rate – Ratios are easier to act on and inherently comparative.
(4) A good metric changes the way you behave – Changes in the metric should have an impact on how the business operates.
Adding to the four characteristics that define a good, Moran (2018) expands the list by describing two additional ones:
(5) A good metric is relevant – It is important that the metric aligns with the core business goals.
(6) A good metric is reliable – The metric should be technically robust and relevant over time.
To tackle the obstacle of selecting relevant metrics, Croll & Yoskovitz (2013) recommend businesses to use frameworks which helps them understand changes and which metrics to use. One such framework is the Three Engines of Growth: the Sticky
Engine, the Virality Engine, and the Paid Engine as contrived by Ries (2011). Ries
states that while all three engines are useful to drive the growth of a business, it is important to focus on one engine at the time.
3.3.1 Sticky Engine
Ries (2011) states that the focus of the Sticky Engine is to get customers to return and to keep using the product. The author mentions that if customers are not sticky, this leads to high levels of churn which becomes problematic. This is especially true for SaaS-businesses considering the subscription based business-model, which only works if retention rates are kept at an sufficiently high level, as described in chapter 1.1 and 1.5.
Ries (2011) divides the fundamental metrics for the Sticky Engine into two main categories: (1) customer retention and (2) usage frequency. In addition to those, McClure (2007) also points out engagement as a crucial metric to track in this engine. Table 2 below gives an overview of the metrics within each category followed by a more in-depth description.
Table 2 – Overview of relevant metrics for the Sticky Engine
Metric Category Description
DAU, WAU & MAU Engagement The number of daily, weekly, or monthly active users.
Stickiness Engagement The level of engagement the users show with the product.
Logo Retention
Rate Retention
The percentage of customers who remain after a certain amount of time.
Net Retention Rate Retention
The percentage of revenue the company receives from the customers who remain after a certain amount of time, compared to revenue at the time of customer acquisition. This includes the effects of churn, downgrades, and expansions.
Gross Retention
Rate Retention
The percentage of initial revenue the company receives from the customers who remain after a certain amount of time, compared to revenue at the time of customer acquisition.
Customer
Engagement Score
Engagement/User frequency
The level of customer activity and usage of the product.
DAU, WAU & MAU
The Daily, Weekly and Monthly Active Users provide an indication of engagement and activity of the users within the product (Mura, 2018). Croll & Yoskovitz (2013) state that in order for the metric to follow the guidelines on what make a metric good, as mentioned in chapter 3.3, two things need to be done:
(1) Define what an active user is – For example, this can be that the user logs in or completes a task in the product.
(2) Make it a ratio in order for it to be comparable – For example, this can be done by dividing the active users with the total user base.
Stickiness
Stickiness is an engagement metric which gives an indication on how engaged the
customers are with the product, and is defined as the ratio between daily active users and monthly active users (Mura, 2018).
𝑆𝑡𝑖𝑐𝑘𝑖𝑛𝑒𝑠𝑠 = 𝐷𝐴𝑈 𝑀𝐴𝑈
As stated in chapter 3.3, for the metric to provide value it should be comparative, which is why Miller (2019) states that it is more interesting to focus on how the metric changes over time rather than its actual value.
Logo Retention Rate
The Logo Retention Rate (LRR), indicates the percentage of customers who remain after a certain amount of time (Bernazzani, 2019). Murray (2019) states that the LRR should be examined in a time frame relevant to the volatility or stability of the customer base. To determine this, Murray suggests measuring different time frames and choosing a relevant one accordingly. The LRR is calculated using the formula below (Bernazzani, 2019):
𝐿𝑅𝑅 = # 𝐶𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑡 𝐸𝑛𝑑 𝑜𝑓 𝑃𝑒𝑟𝑖𝑜𝑑 − # 𝐶𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝐴𝑐𝑞𝑢𝑖𝑟𝑒𝑑 𝐷𝑢𝑟𝑖𝑛𝑔 𝑃𝑒𝑟𝑖𝑜𝑑 # 𝐶𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑎𝑡 𝑆𝑡𝑎𝑟𝑡 𝑜𝑓 𝑃𝑒𝑟𝑖𝑜𝑑
Net Retention Rate
Murray (2019) defines the Net Retention Rate (NRR) as the percentage of revenue the company receives from the customers who remain after a certain amount of time compared to revenue at the time of customer acquisition, after accounting for customers expanding, churning, and downgrading. The metric is calculated using the formula below:
𝑁𝑅𝑅 = 1 + 𝐸𝑥𝑝𝑎𝑛𝑑𝑒𝑑 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 + 𝐶ℎ𝑢𝑟𝑛𝑒𝑑 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 + 𝐷𝑜𝑤𝑛𝑔𝑟𝑎𝑑𝑒𝑑 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 𝐼𝑛𝑖𝑡𝑖𝑎𝑙 𝑅𝑒𝑣𝑒𝑛𝑢𝑒
Vaswani (2016) states that best-in-class SaaS-companies have a NRR of above 100%, meaning these companies will grow even if they do not acquire new customers, which occurs if the company manages to upsell customers more than they downgrade and churn combined.
Gross Retention Rate
Murray (2019) defines the Gross Retention Rate (GRR) as the percentage of revenue the company receives from the customers who remain after a certain amount of time, compared to revenue at the time of customer acquisition, after accounting for customers churning and downgrading. The metric is calculated using the formula below:
𝐺𝑅𝑅 = 1 + 𝐶ℎ𝑢𝑟𝑛𝑒𝑑 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 + 𝐷𝑜𝑤𝑛𝑔𝑟𝑎𝑑𝑒𝑑 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 𝐼𝑛𝑖𝑡𝑖𝑎𝑙 𝑅𝑒𝑣𝑒𝑛𝑢𝑒
Vaswani (2016) states that best-in-class SaaS-companies typically manage to achieve 99% GRR on a monthly basis.
Customer Engagement Score
The Customer Engagement Score (CES) measures the engagement of a company’s customers regarding its product (Mazzeu, 2018). The Corporate Finance Institute (2020) states that tracking the CES can help companies identify customers who are willing to convert from a free trial to a full product, are close to churning, and may
accept upsell or cross-sell deals. To calculate the metric, Skok (2012) presents two steps:
(1) Identify the key engagement events to track. These should relate to the benefits which the product brings to the customer and the company (e.g., the number of logins, amount of time using the product, or number of sessions per day); and
(2) Assign engagement scores to each action.
Adding to the list, Mazzeu (2018) describes a third step which is to:
(3) Prioritize each engagement event in relation to its importance. This is done by adding a weight to the actions within a range of 0 and 1, where higher weight indicates higher importance.
Combining these three steps creates the Customer Engagement Score which is calculated by the following formula (Skok, 2012):
𝐶𝐸𝑆 = (𝑤1∗ 𝑛1) + (𝑤2∗ 𝑛2) + ⋯ + (𝑤𝑛∗ 𝑛𝑛)
In the formula, w is the weight of each action and n is the number of times the action has occurred.
3.3.2 Virality Engine
The focus of the Virality Engine is to get the word out about the product, either by the company actively advertising it through different channels, or the customers advertising it for the company (Ries, 2011). Croll & Yoskovitz (2013) state that one of the most attractive factors about virality is that it compounds, meaning that if there is no churn and each customer brings in more than one new customer respectively, the user base can grow infinitely until it is saturated. Furthermore, the authors propose three key metrics for this engine which are presented in Table 3 below.
Table 3 – Overview of relevant metrics for the Virality Engine
Metric Category Description
Viral Coefficient Virality The rate of which existing customers bring in new customers.
Viral Cycle Time Virality The time it takes for one customer invite new customers.
Net Promoter Score Advertising The likeliness that a customer would recommend the product to a friend of college.
However, since this study mainly focuses on how to retain customers, and not necessarily on how to obtain new ones, the metrics presented in this engine are of little relevance and will not be explained further.
3.3.3 Paid Engine
The third and final engine is the Paid Engine which mainly focuses on payments from customers (Ries, 2011). Croll & Yoskovitz (2013) describe getting paid as the “ultimate metric” for identifying a sustainable business model. They further define a sustainable business model as one which makes more money from the customers, than it costs to acquire them. Furthermore, Ries (2011) states that the key metrics to examine within this engine are customer revenue, cost, and lifetime value, which are presented in Table 4 below.
Table 4 – Overview of relevant metrics for the Paid Engine
Metric Category Description
MRR/ARR Revenue The monthly or annually recurring revenue from the customers.
ARPU & ARPA Revenue The average revenue per user or account.
ACV Revenue The average annual contract value of a customer subscription.
CAC Cost The cost of acquiring a customer.
LTV Lifetime value The total economic value of a customer over its lifetime.
LTV:CAC Lifetime value/Cost The total economic value of a customer over its lifetime, in relation to its acquisition cost.
CAC-payback Cost/Revenue The time it takes for a business to recover the CAC from customer revenue.
MRR & ARR
The Monthly Recurring Revenue (MRR) and Annual Recurring Revenue (ARR) are defined as the monthly or annual revenue the company generates from its customers (Skok, 2018). Since SaaS-businesses generate revenue from a subscription that the users pay on a monthly or yearly basis (Guo & Ma, 2018), Croll & Yoskovitz (2013) argue that MRR and ARR are fundamental metrics to measure in order to understand the performance of the business.
ARPU & ARPA
The Average Revenue Per User (ARPU) and Average Revenue per Account (ARPA) are defined as the average recurring revenue which a user or account generate each month (Mura, 2018), and is calculated as:
𝐴𝑅𝑃𝑈 = 𝑇𝑜𝑡𝑎𝑙 𝑀𝑅𝑅
𝑇𝑜𝑡𝑎𝑙 # 𝑜𝑓 𝑈𝑠𝑒𝑟𝑠 𝐴𝑅𝑃𝐴 =
𝑇𝑜𝑡𝑎𝑙 𝑀𝑅𝑅 𝑇𝑜𝑡𝑎𝑙 # 𝑜𝑓 𝐴𝑐𝑐𝑜𝑢𝑛𝑡𝑠
ACV
The Annual Contract Value (ACV) is defined as the average annualized revenue of the company’s account subscription agreements (Hatfield, 2017). The advantage of ACV over ARR is that it can be used for one-time fees, and that is also accounts for the revenue of each year dependent of the contract length (Hatfield, 2017). Although its advantages, Campbell (2019) points out that the metric lacks a clear definition since different companies calculate it in different ways.
Customer Acquisition Cost (CAC)
The Customer Acquisition Cost (CAC) is defined as the total costs of acquiring a customer, and is calculated as:
𝐶𝐴𝐶 =𝑆𝑢𝑚 𝑜𝑓 𝑎𝑙𝑙 𝑆𝑎𝑙𝑒𝑠 & 𝑀𝑎𝑟𝑘𝑒𝑡𝑖𝑛𝑔 𝑒𝑥𝑝𝑒𝑛𝑐𝑒𝑠 𝑁𝑜 𝑜𝑓 𝑛𝑒𝑤 𝐶𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝐴𝑑𝑑𝑒𝑑
By calculating the CAC, SaaS-business can better understand how much of its resources it can profitably allocate for acquiring a new customer (Singh, Bhagat, & Kumar, 2012).
Customer Lifetime Value (LTV)
The Lifetime Value (LTV) of a customer is defined as the total economic value it generates for the business over the customer’s lifetime (Singh, Bhagat, & Kumar, 2012). The LTV can more accurately be calculated by discounting the customer’s future cash flows (Kumar & Reinartz, 2016), but is within SaaS-businesses usually calculated as (Skok, 2015):
𝐿𝑇𝑉 = 𝐴𝑃𝑅𝐴
𝐶𝑢𝑠𝑡𝑜𝑚𝑒𝑟 𝐶ℎ𝑢𝑟𝑛 𝑅𝑎𝑡𝑒
LTV:CAC
Putting the Lifetime Value (LTV) of a customer in relation to its cost of acquisition (CAC), creates the LTV:CAC ratio. Thus, the ratio is defined as the amount of money the company spends on acquiring a customer, in relation to the economic value it brings over its lifetime (Croll & Yoskovitz, 2013). Furthermore, Skok (2018) argues that the ratio is one of the most important metrics for SaaS-businesses, as it indicates the profitability and viability of a business in the long run. He also mentions that a
general benchmark for the metric to ensure profitability is that the LTV should be three times the CAC. The metric is calculated as:
𝐿𝑇𝑉: 𝐶𝐴𝐶 =𝐿𝑇𝑉 𝐶𝐴𝐶
CAC-payback
Skok (2018) states the CAC-payback period is defined as the time it takes for the recurring revenue from a customer to cover its CAC. He argues that the CAC-payback period is, in combination with the LTV:CAC ratio, a vital metric for a SaaS-business since it showcases the company’s profitability and cash flow. Furthermore, Skok states that top performing SaaS-businesses strive to achieve a CAC-payback period of less than 12 months.
4
Methodology
Chapter 4, Methodology, explains the multimethodology approach consisting of a single case study and action research, the different data collection methods, and how it will all be utilized in order to fulfill the study’s purpose.
4.1
Multimethodology approach
This study follows a “multimethodology approach”, also known as multimethod research, as introduced by Brewer & Hunter (1989). Multimethod research allows for the researchers to utilize multiple methodologies and pursue several goals (Troisi, Maione, Grimaldi, & Loia, 2019). This aligns with the main objective of this study, which is to study the purpose in question, something only made possible by the researchers’ active participation of conducting cohort analyses at GetAccept. Although multimethod research have similarities with the “mixed method”, Troisi et al. (2019) state that these should not be confused, since in contrast to multimethod research, the mixed method is instead characterized by the combined use of qualitative and quantitative methodologies.
On top of enabling the flexibility of combining multiple methodologies, multimethod research brings other advantages. Troisi et al. (2019) argues that one advantage of using an approach based on several methods is that it allows for obtaining results characterized by a greater degree of generalizability. Brewer & Hunter (2005) further argues for the use of multimethod research by stating that it allows for cross-validating and cross-fertilizing research procedures, findings and theories, thus ensuring higher validity of the study’s findings. This is further argued by Creswell & Plano Clark (2017) who state that every scientific research approach, theoretical and experimental, can be improved in terms of greater reliability of findings.
The two methodologies chosen, which enables the researchers to both study the purpose in question and also actively participate during the process, are:
❖ Single case study – Since it enables the researchers to investigate a
contemporary phenomenon in depth within its real-life context (Yin, 2009). In this study, the phenomenon investigated is GetAccept’s marketing decision-making, as well as its evolution as a result of the incorporation of cohort analysis.
❖ Action research – Since it allows the researchers to observe, verify, intervene,
modify, and enrich many aspects of the investigated phenomenon, through active participation (Coghlan, 2019; Kemmis, McTaggart, & Nixon, 2014). In this study, this will be utilized to incorporate cohort analysis into GetAccept’s marketing decision-making.
