Trade-Offs in Leveraging External Data Capabilities: Evidence from a Field Experiment in an Online Search Market

Firms increasingly leverage external entities’ data capabilities to unlock improvements in their offerings, but measuring the impact of such capabilities is challenging. Collaborating with the search team at a technology company, we analyzed a large-scale field experiment in which we randomized access to an external, leading search engine’s autocomplete application programming interface (API) for more than two million users over 108 days. We measure the causal effects of removing API access on two performance metrics of the focal company’s search product: (a) clickthrough rate (CTR) on search suggestions and (b) CTR on the search engine results page. We find that, on average, compared with the baseline with API access, removing API access reduces the search suggestion CTR by 4.6%. Further, exploiting the experimental variation, we use an instrumental variables approach to establish that a 10% increase (decrease) in CTR on search suggestions leads to a 1.85% increase (decrease) in CTR on top-slot search results. However, the negative effect of removing API access becomes less negative over time with the effect magnitude in the longer term being half what we would have obtained with a short-term experiment. We provide suggestive mechanism evidence of the longer term effect: the focal company’s reliance on the leading search engine’s data capability tapers off the accumulation of internal data and then limits the improvement of its autocomplete predictions. This research informs managers of a critical trade-off in leveraging external data capabilities and sheds light on regulations, such as the Digital Markets Act, that mandate data sharing by large digital platforms. This paper was accepted by DJ Wu, information systems. Funding: This work was supported by the National Natural Science Foundation of China [Grants 72442015, 72171132, and 72171146] and the China Scholarship Council [Grant 202206230110]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01834 .