Jacobs and Sreenivas compare three venture capital databases, VentureXpert, Crunchbase, and the less studied Cleantech database, using the full population of startups in the global photovoltaic cell industry as a test case where ground truth can be checked. They assess each database's consistency and completeness, examining how well each captures temporal trends, investor syndication, and geographic distribution of venture investments. Their central finding is that none of the three databases captures the full universe of venture capital activity in this sector, each only samples it, yet the databases can still produce reliable results for certain research questions despite that incompleteness. This nuances the assumption that incomplete data is simply unusable, instead suggesting fitness for purpose depends on what is being studied.

Researchers designing a study on VC-backed sectors get a rare direct comparison of database coverage against a verifiable ground truth, the actual population of photovoltaic startups, rather than the more common but weaker approach of comparing two databases against each other without any independent benchmark to check against in most similar studies published elsewhere in the literature. Crunchbase in particular gets relatively little rigorous academic scrutiny compared to VentureXpert, so testing it head to head against an established database fills a real gap in the methods literature. The nuanced conclusion, that incompleteness does not automatically disqualify a database but depends on the specific research question, gives practitioners a more usable decision rule than a blanket recommendation to avoid any one source.