Picture a logistics firm subscribing to weather forecasts, a mobile provider offering data packages, or individuals participating in webinars and streaming movies online. Information goods, such as data and digital products, permeate various aspects of modern life, from business operations to entertainment consumption, illustrating their omnipresence and interdisciplinary nature.
Data as a Commodity: From Byproduct to Asset
In today’s economic landscape, where numerous products and services are exclusively delivered in digital formats, a novel addition to traditional production factors labor, capital, and land has emerged: data. No longer perceived as a mere byproduct, data has transformed into a valuable commodity in its own right. This shift has catalyzed the trend of data commoditization, where data owners aim to monetize their assets while buyers seek high-quality data for various purposes, notably for training advanced machine learning models, which necessitate top-tier training data.
The term “commodity” simplifies data’s complexity. Unlike gasoline, where two liters typically yield similar mileage in comparable vehicles, data operates on a different premise. Foremost among the characteristics of data assets is their non-rivalrous nature. Simply put, selling data to one customer does not prohibit its sale to another. Those trading data aim to make it exclusive, meaning that only paying customers can access it, thereby transforming it into a club good. This endeavor represents a fundamental challenge in fostering a thriving data economy.
How to Put a Price Sticker on Information Goods?
This aspect is also crucial. Information goods, including digital products and data products, revolutionize traditional economic models by significantly reducing costs across various dimensions such as search, production, replication, transportation, and tracking. Because data can be replicated at negligible (often zero) marginal cost, their prices in marketplaces tend to be very low. While this makes them economically appealing, it also poses a risk as it allows competitors to easily enter the market.
This trend has spurred a surge in demand for platforms facilitating the exchange of data assets and services. Data marketplaces serve as intermediaries connecting data sellers with potential buyers. Like brokers, these marketplaces often delegate the price-setting process to sellers or facilitate negotiations between sellers and buyers. Sellers face challenges in pricing datasets efficiently due to competition and buyers’ willingness to pay, which is often addressed by versioning – linking prices of different versions to values placed by different customer groups. FAME is revolutionizing price setting by giving advice on the intrinsic value of data assets or digital products and considering the volumes and prices of previous trades. Another challenge lies in the fair distribution of generated revenue to ensure sellers are compensated for their marginal contributions fairly.
On the other hand, buyers need to assess the usefulness of datasets and how accuracy can lead to financial benefits. This brings us to another specific feature of datasets – customers cannot test the data before the purchase. As they cannot evaluate the quality of data beforehand, their ability to select the best data source for their needs is limited. Furthermore, the significance of prediction tasks and the value associated with enhancing prediction accuracy differ significantly among various firms. How should the data be evaluated, then? If it is done by data owners, they could potentially manipulate information on data quality to maximize profit. On the other hand, if the buyer conducts data valuation, they must access the data, potentially leading to a situation where they no longer find it necessary to purchase the data.
Furthermore, the value of data to a firm is inherently linked to its combinatorial nature, meaning that the value of a specific dataset depends on the availability of other potentially correlated datasets. The potential of data is constrained, as a single data source may not fully meet the buyer’s needs, while multiple sources from different sellers could prove more effective instead. Therefore, determining prices for a collection or bundles of datasets with correlated signals poses challenges.
Innovative Solutions: Bundling and Testing Strategies
To establish pricing for a wide range of non-rival information goods with zero marginal costs, bundling is frequently employed. This entails selling multiple products together at a unified price. Given that numerous information goods can be bundled without a significant rise in cost, it may be economically advantageous to bundle thousands of digital products to accommodate diverse and independent customer preferences.
How can the data buyer see or test it before transacting? Very often, buyers rely on methods like free samples, trial periods, sandbox environments, live demos, or reputation mechanisms provided by platforms to gauge data reliability and usefulness before making a purchase.
Due to these specific characteristics, commercial data markets are still immature, and it is the mission of the FAME platform to enable accessible, efficient, and mutually beneficial interactions.
References:
[1] Agarwal, A., Dahleh, M., & Sarkar, T. (2019, June). A marketplace for data: An algorithmic solution. In Proceedings of the 2019 ACM Conference on Economics and Computation (pp. 701-726).
[2] Azcoitia, S. A., & Laoutaris, N. (2022). A survey of data marketplaces and their business models. ACM SIGMOD Record, 51(3), 18-29.
[3] AZCOITIA, Santiago Andrés; IORDANOU, Costas; LAOUTARIS, Nikolaos. What is the price of data? A measurement study of commercial data marketplaces. arXiv preprint arXiv:2111.04427, 2021.
[4] Firmani, D., Mathew, J. G., Santoro, D., Simonini, G., & Zecchini, L. (2023). Bridging the Gap between Buyers and Sellers in Data Marketplaces with Personalized Datasets. In CEUR Workshop Proceedings (Vol. 3478, pp. 525-534).
[5] Pei, J. (2020). A survey on data pricing: from economics to data science. IEEE Transactions on knowledge and Data Engineering, 34(10), 4586-4608.
[6] Tian, Z., Liu, J., Li, J., Cao, X., Jia, R., Kong, J., … & Ren, K. (2022). Private data valuation and fair payment in data marketplaces. arXiv preprint arXiv:2210.08723.
More information:
EUBA
