Sampling Bias in Entrepreneurial Experiments

Abstract

Using data from a prominent online platform for launching new digital products, we document that “sampling bias”—defined as the difference between a startup’s target customer base and the actual sample on which early “beta tests” are conducted—has a systematic and persistent impact on the venture’s success. Specifically, we show that products with a female-focused target market launching on a typical day, when nine in 10 users on this platform are men, experience 45% less growth a year after launch than those for whom the target market is more male-focused. By isolating exogenous variation in the composition of beta testers unrelated to the characteristics of launched products on that day, we find that on days when there are unexpectedly more women beta testers on the platform—reducing the amount of sampling bias for female-focused products—the gender performance gap shrinks toward zero. Our results highlight how sampling bias can lead to fewer successfully commercialized innovations for consumers who are underrepresented among early users. This paper was accepted by Alfonso Gambardella, business strategy. Funding: This work was supported by the Harvard Business School, Ewing Marion Kauffman Foundation [Grant 201708-2801]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2021.01740 .

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 20 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.