Does Recurring Revenue Really Drive Financial Outcomes in Search Fund-Acquired Businesses?
Daniel Lazier, Jacob Thomas, A. J. Wasserstein. (2025)
Yale Case, no number printed, September 23, 2025
Sample. "we relied on a valuable data set provided by a single generous investor with decades of search fund experience and results." (p2). "the data set consists of 59 observations (Exhibit 1), all of which are traditional search fund projects that have completed the full cycle of raising a fund, acquiring a company, and ultimately exiting." (p2). The investor coded recurring revenue share from each company's confidential information memorandum at acquisition. Returns are masked (indexed 0 to 100 and ranked) at the investor's request. Bins: none 29, low 11, moderate 9, high 10. Included and excluded. Exited traditional search deals from one investor only; no unrealized companies, broken searches or self-funded deals. Recurring revenue measured only at entry. Headline numbers. Indexed MOIC on recurring revenue share: "the t-stat of 1.10 and p-value of 0.27 indicate there is no reliable statistical relationship here." (p12). Indexed IRR: "the low t-stat of 1.32 and high p-value of 0.19 reveal that the slope is not significant." (p12). High versus none medians: "With p-values of 0.40 and 0.92 for indexed MOIC and indexed IRR respectively, we find no significant difference" (p14). Entry price: "a regression line at 1.61, a correlation of 0.36, a t-stat of 2.83, and a low p-value of 0.006." (p16), that is, higher recurring revenue goes with higher entry EBITDA multiples. Limits. Authors: "All our data comes from a single investor, which could introduce sampling bias." and "We are not performing multivariate analyses" (p4); the recurring revenue coding has "some subjectivity" and does not separate contractual from repeat revenue. Reader's notes: the ranked-MOIC relationship in the authors' own table (figure, p14: correlation 0.26, p 0.04) is significant at 5%, yet the text says the ranked results confirm no significance; group sizes of 9 to 10 give little power, so "no significant difference" should not be read as "no effect". The N of 59 matches the value-attribution note but that note describes six investors and this one a single investor; neither says they are the same companies.