D EVELOPING A T AXONOMY FOR R EVENUE M ODELS OF P LATFORM B USINESS M ODELS

Platform business models like Uber Ride or Airbnb Lodging enable innovative business models by operating digital platforms to connect providers and consumers of products and services in two-sided markets. A particular challenge with platform business models is designing an appropriate revenue model to capture value. This paper presents a taxonomy that classifies the different dimensions and characteristics of revenue models for platform business models. A proven taxonomy development method is used that includes a review of current literature related to platform business models. The taxonomy provides a comprehensive classification of platform revenue models and is applied to a real-life case. The results of this paper include a UML class model and a final taxonomy with 14 dimensions and 64 characteristics. The paper contributes to the design process of novel platform business models and expands the understanding of how digital platforms can generate revenues.


Introduction
The significance of digital platforms continues to grow, and companies such as Uber Ride, Airbnb Lodging, Spotify Music, and eBay Marketplace have established innovative platform business models.Regardless of the industry, every company must make strategic decisions about their business model to stay competitive (Parker et al., 2016).The emergence of platform business models raises the question of what competitive advantages a company can achieve with its own business model and underscores the need for design knowledge to innovate novel (platform) business models.The motivation of this paper is based on a research preview from Bartels & Gordijn (2022) and addresses the design of systematic revenue models for platform business models.We provide a taxonomy that classifies relevant dimensions and characteristics of revenue models for platform business models.The research question for this paper is as follows: Which dimensions and characteristics can be used to describe revenue models of platform business models?
To answer this research question, we used a taxonomy development process following Nickerson et al. (2013) and extracted relevant dimensions and characteristics from a literature review.We also present a use case of the Smarte.Land.Regionen (SLR) platform, a digital solution-brokering platform for German counties, where the proposed taxonomy was applied to design a possible revenue model.In follow-up research, the taxonomy will be developed into a design tool to help practitioners create platform business models more systematically.This paper is structured as follows.In section 2, we introduce key terms and relevant related work.Section 3 presents the research design of the taxonomy development process and section 4 shows the taxonomy we created.Section 5 presents the use case to which the taxonomy was applied.Finally, section 6 presents our discussion, limitations, and an outlook on future work.

Theoretical Background
In our understanding, a platform business model is characterized by four aspects adapted from the definitions of Koch & Krohmer et al. (2022), Gordijn & Wieringa (2021), and Täuscher & Laudien (2017): (1) A platform business model describes the concept of how economic value is created, distributed, and consumed in a network of parties, called a digital ecosystem.(2) It creates value through a digital platform, operated by a platform operator (i.e., asset broker), which connects at least two market sides -asset providers and asset consumers.
(3) It brokers assets such as products or services via its digital platform.(4) A digital platform can serve as the hub of a digital ecosystem consisting of companies working collaboratively and competitively to meet customer needs (Moore, 1996).The revenue model is part of the value capture dimension of a business model and clarifies which monetization mechanisms are used to generate revenues.Accordingly, the revenue model of a platform business model, as we understand it, explains how revenue can be generated by enabling brokering services via a digital platform connecting asset providers and asset consumers.A taxonomy is defined as a structure above the technical terms of a subject area (Freichel et al., 2021a).In this paper, a taxonomy is considered a form of classification of relevant dimensions and characteristics for revenue models of platform business models.The development process of our taxonomy for revenue models of platform business models follows the guidance of Nickerson et al. (2013) as a well-structured methodology for researchers who intend to develop taxonomies step by step.The literature review, development process, and data presented in this paper are fully documented and can be found here: Bartels et al. (2023).As shown in Figure 1, the purpose and object of the taxonomy is defined in Step 1.
In Step 2, the ending conditions are set, i.e., the criteria that the taxonomy must meet in order to be accepted.For the development of the taxonomy, Step 3, Step 4, and Step 5 are repeated in two conceptual-to-empirical iteration cycles.After applying the taxonomy to a real-life case, all ending conditions in Step 6 are met.

Determination of meta-characteristics and ending conditions
The purpose of our taxonomy follows the research preview of Bartels & Gordijn (2022), and revenue models of platform business models such as the revenue models of eBay Marketplace, Airbnb Lodging, or Uber Ride form our object of interest.The configuration aspects of these revenue models (e.g., $0.35 insertion fee per listing on eBay) are determined as the relevant meta-characteristics of our taxonomy.
Following Nickerson et al. (2013), we consider objective and subjective ending conditions that must be met for the taxonomy to be accepted: (1) The taxonomy must comprise the main dimensions and characteristics of a revenue model for platform business models, and ( 2) no new dimensions or characteristics should be added in the last iteration.
Subjectively, the taxonomy must be (3) meaningful without being unwieldy or overwhelming and ( 4) extensible in order to add new dimensions or characteristics.Finally, ( 5) each dimension and characteristic must provide useful explanations about the object (explanatory).

First cycle: Literature research and classification
To get a data basis for the creation of the taxonomy, we conducted a literature review on revenue models of platform business models.The databases of Scopus, Web of Science, IEEE Xplore, ACM Digital Library, Google Scholar, and Dimensions were searched using the following string: (ecosystem OR platform) AND (business model OR value capture OR revenue model OR profit model).This resulted in a total of 930 papers.
The screening process of titles, abstracts, and full text was guided by the definition of inclusion and exclusion criteria.Of the total of 930 papers, 29 papers were included based on the following inclusion criterion: The paper focuses on relevant dimensions or characteristics of revenue models for platform business models (IC).In addition, five more papers were added to the included results, as we consider them relevant:  (2020).A total of 34 papers were thus used for developing the taxonomy.The remaining 901 papers were excluded based on the following exclusion criteria: 204 papers were duplicates of another paper (EC1), 30 papers were not in English (EC2), six papers were less than three pages (EC3), 13 papers were not research papers (EC4), 41 papers were not accessible even after contacting the authors (EC5), and 607 papers did not meet the inclusion criteria (EC6).The full-text review of the 34 included papers resulted in a total of 68 dimensions and 258 characteristics for revenue models of platform business models.The review process of the literature search with each criterion is documented here: Bartels et al. (2023).
To synthesize the data, a classification was created as a concept matrix according to Webster and Watson (2002).First, all dimensions were sorted alphabetically by title, studied based on the descriptions, and coded using our own classifications.Of the 68 dimensions examined from the literature, nine dimensions could not be classified -the remaining 59 dimensions were grouped into nine self-coded dimensions.
Figure 2 gives an overview of the selected revenue model dimensions derived from the literature.The concept matrix summarizes the comprehensive classifications for revenue models of platform business models on the left side (A) while showing relevant dimensions for revenue models on the right side (B). Figure 2 shows that nine dimensions could be extracted based on 27 papers.Here, "revenue model", "revenue stream", "revenue source", and "pricing model" are frequently used as relevant dimensions.However, the initial taxonomy derived from the concept matrix did not meet the ending conditions, as the "pricing model" dimension had a strong overlap with "price mechanism", "price discovery", and "price discrimination".Therefore, in the second iteration cycle, the dimension was deleted to avoid redundancy.

Second cycle: Meta-model and taxonomy revision
In the second iteration of taxonomy development, we created a UML class model to express the relationships of the revenue model dimensions for platform business models within the taxonomy in a transparent way.We consider this step to be useful for designing a taxonomy holistically and ensuring its meaningfulness.The metamodel in Figure 3 illustrates the relationships between eight classes depicting the dimensions of the taxonomy.An asset broker and operator of a revenue model (e.g., the platform provider of the eBay marketplace) may have multiple "revenue model types", each having a "revenue source" (who is monetized?)and a "revenue stream" (how to monetize?).This triangular relationship is crucial in our opinion and is also confirmed by the literature, as demonstrated in Figure 2. The pricing components, including "price discovery", "payment frequency", "pricing mechanism", and "price discrimination", always refer to an individual "revenue stream".The pricing model as a dimension is not explicitly included in the metamodel, as it is either redundant to the existing dimensions or can be considered as a combination.The classes shown in Figure 3 were adopted as dimensions in the second iteration.

Taxonomy
An asset provider (e.g., Airbnb host) aims to generate revenues through a business model of its own (e.g., renting one's own apartment to travelers), which should be viewed as a separate but relevant component for describing the overall platform business model of an asset broker (e.g., the operator of the Airbnb Lodging platform).For this, the use of a digital platform by asset providers depends on their ability to generate revenues.We concluded that a revenue model for a (two-sided) platform business model can only be described holistically if both the asset broker's revenue model and the asset provider's revenue model are represented.
Consequently, the final taxonomy includes 14 dimensions, with seven dimensions covering the asset broker's perspective and the other seven dimensions covering the asset provider's perspective.The taxonomy shown in Figure 4 satisfies all relevant ending conditions.
A revenue model type of the asset broker (DB1) covers the revenue source and revenue stream through which the asset broker generates revenues.A revenue stream of the asset broker (DB2) describes how the asset broker generates revenues, i.e., the strategy the asset broker uses to monetize the revenue source through the platform.Access fees, commission fees, sale of platform services, advertising fees, listing fees, or donations may be used to generate revenue.The revenue source of the asset broker (DB3) describes who is monetized by the asset broker, i.e., the actor through whom the asset broker generates the revenue stream.Asset consumers, asset providers, or third parties can be monetized by the asset broker.The payment frequency of the platform price (DB4) describes how often payments recur for the asset broker, i.e., the frequency with which the revenue source is charged by the asset broker.Payments can appear as one-time, multiple-time, or usage-based.The price discovery of the platform price (DB5) describes who sets the platform price, i.e., whether the platform price is set by the asset broker, by asset providers, asset consumers, or by negotiations.The price mechanism of the platform price (DB6) describes the influence of supply and demand on the platform price, i.e., whether the platform price is fixed or variable.A platform price can be fixed and static or variable and dependent on further factors.If the platform price is variable, it can be subject to price discrimination.The price discrimination of the platform price (DB7) describes different platform prices, i.e., whether discriminatory factors influence the platform price to be paid.Platform price discrimination can take the form of location-based, quantity-based, or feature-based price differences.A revenue model type of the asset provider (DP1) covers the revenue source and revenue stream by which the asset providers generate revenues.The revenue stream of the asset provider (DP2) describes how the asset providers generate revenues, i.e., the strategy the asset providers use to monetize the revenue source through the platform.The asset provider can generate revenue through the platform by selling, renting, or charging a usage-based fee for the asset.The revenue source of the asset provider (DP3) describes who is monetized by the asset providers, i.e., the actor through which asset providers generate their revenue stream.Asset consumers, the asset broker, or third parties can generate revenue for the asset provider.The payment frequency of the asset price (DP4) describes how often payments recur for asset providers, i.e., the frequency with which the revenue source is charged by the asset providers.Payments for an asset can appear as one-time, multiple times, or usage-based.The price discovery of the asset price (DP5) describes who sets asset prices on the platform, i.e., whether asset prices are set by the asset broker, by asset providers, or by asset consumers.
The price mechanism of the asset price (DP6) describes the influence of supply and demand on asset prices, i.e., whether asset prices on the platform are fixed or variable.The price of an asset may be fixed or variable and depend on other factors.
If the price of an asset is variable, it can be subject to price discrimination.The price discrimination of the asset price (DP7) describes different asset prices, i.e., whether discriminatory factors influence asset prices on the platform.Asset price discrimination can take the form of location-based, quantity-based, or feature-based price differences.

Taxonomy applied to the SLR Platform
To ensure that our taxonomy will be applicable, we used a revenue model of a digital platform in a research project as a real-life case.This research project, called Smarte.Land.Regionen (SLR), aims to improve public services in rural areas through digital solutions.For this purpose, a digital ecosystem is being created that includes a digital platform at its core, called the SLR platform.The SLR platform follows the logic that the SLR platform operator (the asset broker) brokers digital solutions, e.g., mobility services (assets) provided by software companies (asset providers) to counties (asset consumers) and their citizens on its digital platform.The SLR platform was studied in an earlier work by Bartels & Schmitt (2022) as a use case for designing network effects for a platform business model.In this work, the SLR platform is used as a real-life object to test whether the taxonomy is suitable for representing a platform revenue model.As shown in Table 1, the SLR platform's revenue model defines that software providers who want to offer their digital solutions on the SLR platform have to pay a fixed access fee to the SLR platform operator on a monthly basis.In our view, the combination of "access fee" (in DB2) and "monthly" frequency (in DB4) is a subscription model, but we can express this more precisely through the taxonomy and consider it not as a standalone revenue model, but as a variant of the "access model" (in DB1).In this way, the digital solution listed on the SLR platform can be found by counties and booked for their citizens.Software companies generate revenue by offering counties their digital solutions through the SLR platform and customizing them to meet the needs of counties and citizens.

6
Discussion, Limitations, and Future Work The main contribution of this work is the creation of a meaningful taxonomy and metamodel in order to get a better understanding the revenue models used by platform business models.The research question of how to classify revenue models of platform business models is answered with a taxonomy of 14 dimensions and 68 characteristics.In their work, Täuscher & Laudien (2018) showed that 74% of platform business models use commission models as their core revenue model.Although this number is significant, it also indicates that much of the variation in revenue models is not fully understood yet.In our view, there are variants such as commission per transaction (e.g., a fee per eBay product sold) or commission per unit of usage (e.g., a fee per Uber mile driven).Our taxonomy is a first step towards gaining a more nuanced understanding of revenue models of platform business models.The proposed taxonomy offers a more precise way of describing different revenue models compared to other taxonomies that use a single characteristic, such as 'subscription' (as seen in Täuscher & Laudien, 2018).As shown in our real-life case, we achieve this level of detail by combining multiple dimensions: "revenue stream" (DB2) with "access model" and "payment frequency" (DB4) with "monthly" frequency.
Limitations.Our taxonomy focuses on revenue models as part of the value capture and does not address the value proposition and the value creation of a business model.Second, it focuses solely on platform business models with two-sided markets involving the asset broker and asset providers as actors with monetization intentions, and therefore cannot be used for one-sided or multi-sided platforms.Despite our transparent taxonomy development process (the research data can be found here: Bartels et al. (2023), there may still be important aspects that have gone unnoticed.An example can be seen in the payment frequency dimension, which is weakly backed in the existing literature and occurred only four times in our data (see concept matrix in Figure 2).However, recent work, such as the platform ontology of Derave et al. (2022), emphasizes the importance of frequency and shows that research on digital platforms and their business models is still evolving.
Consequently, we may have overlooked other aspects in our taxonomy that need to be further elaborated in the future.
Future work on the proposed taxonomy should include the study of different "objects", i.e., platform revenue models, to refine or extend the existing dimensions and characteristics, as suggested by Nickerson (2013) as an empirical-to-conceptual process.Our initial contribution of applying the taxonomy to the SLR platform is a first step.Now, the taxonomy needs to be tested on more real-life objects.The overall goal of this research is to provide this taxonomy as a design tool for practitioners to systematically design revenue models, as proposed by Bartels & Gordijn (2022), who called this a "business model construction kit".

Figure 2 :
Figure 2: Concept matrix of search results

Figure 3 :
Figure 3: UML metamodel of the proposed taxonomy

Figure 4 :
Figure 4: Taxonomy for revenue models of platform business models

3 Research Design: Taxonomy Development Process
Täuscher & Laudien (2018)resented a taxonomy for business models of data marketplaces with 17 business model dimensions and 59 business model characteristics.Springer & Petrik (2021)showed a taxonomy for platform pricing of digital platforms in the context of the Industrial Internet of Things (IIoT) with 13 impact factors and 38 characteristics.Staub et al. (2021)elaborated a taxonomy for digital platforms with 16 design dimensions and 44 characteristics.A similar taxonomy for digital platforms was elaborated byFreichel et al. (2021a)with 16 dimensions and 40 characteristics.Täuscher & Laudien (2018)presented a taxonomy for marketplace business models with 14 business model attributes and 43 specifications.They applied their taxonomy to a sample of 100 digital marketplaces and showed that there are

Table 1 : Taxonomy applied to the SLR platform Description of the SLR platform revenue model DB1
The SLR platform operator generates revenue through an access model and monetizes the providing software companies.
DB6 Access fees are fixed at 500€ and are not changeable.DB7 There is no price discrimination.DP1 The software companies generate revenue through the SLR platform by offering digital solutions based on a pay-per-use model and monetizing the counties.

DP2
Revenues are generated through a usage fee for the digital solutions.DP3 Counties that request solutions from the SLR platform are monetized.DP4 Usage fees are incurred each time a digital

solution is operated for a county. DP5 Usage fees are set by the providing software company. DP6 Usage fees are variable. DP7 Usage fees depend on the functionality of
the digital solution and vary.