Measuring Fair Competition on Digital Platforms

Five-step approach to test for unfair recommendations on digital platforms

Do you use Google, Amazon, Airbnb, Booking, Spotify, or an App store on your smartphone? Then you will be familiar with their ubiquitous recommendations — search result rankings on Google and Amazon, Spotify-curated playlists, or a featured apartment on Airbnb’s homepage. Those recommendations heavily influence your and other users’ clicking or purchasing decisions and play a major role in how competition works on such platforms.

Platforms could often have incentives to design unfair competition, especially if they act in a dual role, such as Amazon acting as an e-commerce platform and selling products on its own platform — directly competing with third-party sellers. In this setting, it is easy to imagine why Amazon could unfairly favor its own products through “self-preferencing” by recommending them more prominently. However, detecting such unfair behavior presents a challenge. Does Amazon recommend its products because they are better than competing products or simply because they are from Amazon? This question is challenging to answer.

The research project “Measuring Fair Competition on Digital Platforms” by Lukas Jürgensmeier and Bernd Skiera proposes a five-step approach to test for fair competition through recommendations. Combining extensive data from Amazon’s search results with high-frequency Amazon product data, the authors illustrate how this approach can test for potentially anti-competitive recommendations on digital platforms, including self-preferencing. Testing for such conduct is important for consumers and sellers but for the platforms: new laws, such as the European Digital Markets Act, explicitly prohibit self-preferencing, and regulators could impose hefty fines for violations.

What is your gut feeling — is competition on platforms fair? Does Amazon unfairly rank its own products higher in its search results than similar competitors? Read the (perhaps surprising) results and the full working paper on SSRN: This research is part of the first author’s doctoral dissertation, whose proposal won the 2023 ISMS Doctoral Dissertation Award.