Get rich or die trying… finding revenue model fit using machine learning and multiple cases

• Author: Kathleen M. Eisenhardt,Ron Tidhar
• Published date: 01 July 2020
• DOI: http://doi.org/10.1002/smj.3142
• Date: 01 July 2020

RESEARCH ARTICLE

Get rich or die trying…finding revenue model

fit using machine learning and multiple cases

Ron Tidhar | Kathleen M. Eisenhardt

Department of Management Science and Engineering, Stanford University, Stanford, California, USA

Correspondence

Ron Tidhar, Department of Management

Science and Engineering, Stanford

University, Stanford, CA 94305.

Email: rtidhar@stanford.edu

Abstract

Research Summary: While revenue models are stra-

tegically important, research is incomplete. Thus, we

ask: “What is the optimal choice of revenue model?”

Using a novel theory-building method combining

machine learning and multi-case theory building, we

unpack optimal revenue model choice for a wide range

of products on the App Store. Our primary theoretical

contribution is a framework of high-performing revenue

model-activity system configurations. Our core insight is

the fit between value capture (revenue models) and

value creation (activities) at the heart of successful busi-

ness models. Contrastingly, low-performing products

avoid complex value capture (i.e., freemium) and mis-

understand value creation (e.g., overweight effort).

Overall, we contribute a theoretically accurate and

empirically grounded view of successful business

models using a pioneering method for theory building

using large, quantitative data sets.

Managerial Summary: Revenue models are critical

for product performance. Yet, the high-performing

choice is often unclear. We combine machine learning

with multiple-case deep-dives to unpack optimal reve-

nue model choice for a wide range of products on the

App Store, a significant setting in the digital economy.

Our primary insight is that high-performing products

fit value capture (revenue models) and value creation

Glueck Best Paper Award, STR Division, AOM, 2019

Received: 31 January 2018 Revised: 21 August 2019 Accepted: 12 December 2019 Published on: 5 March 2020

DOI: 10.1002/smj.3142

Strat Mgmt J. 2020;41:1245–1273. wileyonlinelibrary.com/journal/smj © 2020 John Wiley & Sons, Ltd. 1245

(activity systems) to form coherent business models.

Contrastingly, low-performing products avoid complex

value capture (i.e., freemium) and misunderstand value

creation (e.g., overweight effort and price). We also

identify the importance of user resources,marketing,off-

line brand, and product complexity for specific revenue

models. Overall, we contribute a framework for the

optimal choice of revenue model and spotlight the reve-

nue model-activity system configurations of successful

business models.

KEYWORDS

business models, machine learning, mobile application products

(apps), multi-case theory building, revenue models

1|INTRODUCTION

In 2011, music streaming service, Spotify, launched its product to U.S. listeners. Although simi-

lar to other music streaming services, Spotify was notably different in its revenue model. While

Pandora used an advertising revenue model and Apple Music chose paid, Spotify selected

freemium. That is, it offered its basic service for free, and a paid premium version that provided

more content, greater functionality, and a better user experience. Given these similar products,

a puzzle emerges: Why do they have different revenue models? As other examples like Netflix

and Hulu, Stitch Fix and Rent the Runway, and 23andMe and Ancestry.com suggest, rivals

across many industries may choose surprisingly different revenue models. Yet, resolving this

puzzle and the related optimal choice of revenue model remain elusive.

Consistent with others (Casadesus-Masanell & Zhu, 2010), we define a revenue model as the

monetization approach by which a firm derives sales from its products. Thus, revenue models

are the means by which firms capture value, and so earn the revenue essential for superior

financial performance. Revenue models are part of the broader concept of business models. By

business model, we mean the system of interconnected activities performed by a focal firm (and

often by users and partners) to create value, with part of that value captured by the firm

(i.e., revenue model; Massa et al., 2017; McDonald & Eisenhardt, 2019).

Research suggests that revenue models are a novel source of innovation (Casadesus-

Masanell & Zhu, 2013; Snihur & Zott, 2019), and that revenue models (and broadly business

models) are early and critical strategic choices (McDonald & Eisenhardt, 2019). Yet choosing a

revenue model can be challenging. Falling communication costs and the internet have enabled

new ways to create value (beyond the activities of producer-firms) such as with users and part-

ners in settings like marketplaces and ecosystems (Gambardella & McGahan, 2010; Hannah &

Eisenhardt, 2018; Ott & Eisenhardt, 2020; Zott, Amit, & Massa, 2011). These changes add flexi-

bility in how to create value, but also complicate how to capture it since more actors (e.g., users,

partners) are involved. In addition, many goods have low marginal costs or are experience

goods that further complicate pricing (Massa et al., 2017; Shapiro & Varian, 1998).

1246 TIDHAR AND EISENHARDT