Data Evaluation: Mechanisms and Implications

Unlike traditional products, data have unique traits that make their trading a novel retail activity with distinct operational dynamics. Data evaluation is a situation in which a data buyer can ascertain the value of data from a seller. However, as indicated by Arrow’s information paradox, the intangible nature of data presents significant challenges to data evaluation, particularly in data trading between competitors, a scenario that is becoming increasingly common and relevant. Although offering free samples is widely recommended as an effective approach for data evaluation, its actual impacts remain unclear. We develop a theoretical model to examine the mechanics of data evaluation when a seller sells data to a buyer and the two then compete in product selling. In this model, the low-type seller (offering data of low value) has an incentive to pretend to be the high-type seller, leading to signaling challenges in data trading. The results show that data evaluation can be achieved through a signaling approach in which the high-type seller sells a portion of the data to reveal his or her type, whereas the low-type seller sells all the data. Moreover, we show that offering free samples is also an efficient way to achieve data evaluation; in this case, the high-type seller has to offer a certain volume of free samples, whereas the low-type seller does not. The intuition hinges on the effects of data monetization, which induces the seller to sell more data, and an information advantage over the buyer in product selling, which hinders data selling. In addition, offering free samples helps increase data transactions (i.e., selling volume and price) between the seller and the buyer but undermines the profit of the high-type seller. Finally, facilitating data evaluation is not always beneficial for sellers; maintaining buyer uncertainty regarding the value of data may be profitable for both types of sellers.

This paper was accepted by Jeannette Song, operations management.

Funding: This research is supported in part by National Natural Science Foundation of China [Grant nos. 72572015, 72325013, 72532002, 72202017, 72272013, and 72321002], the Fundamental Research Funds for the Central Universities.

Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2025.01256.