Newsletter and Subscription Sign Up

Four is the Magic Number in Reference Checking

Published Friday Oct 11, 2019

Four is the Magic Number in Reference Checking

Four is the magic number when it comes to checking the references of new hires, according to new Checkster research. In an analysis of more than 51,000 references, Checkster determined that four is the ideal number for quality hires and to prevent turnover.

Data analysis of the total number of references was correlated with first year turnover. Candidates who had three or fewer references providing feedback showed significantly higher rates of turnover and no shows on first day of employment. The biggest difference between obtaining two and four references was an increase in involuntary turnover and no show by as much as 80 percent.

While companies tend to call an average of 2.4 references when using traditional methods, digital automation tools allow hiring managers to access many more references than would be feasible with phone calls. Because of the confidentiality of the digital process, references feel more comfortable giving candid feedback about candidates.

“It’s a myth that reference checking is a waste a time,” said Yves Lermusi, CEO, Checkster. “Research has shown that digital reference checking is one of the best selection processes and checking four references ensures that candidates aren’t cherry-picking the few references who will say good things about them.”

Unlike traditional phone reference checks, automated tools allow employers to gather more references in a fraction of the time compared to manual methods, increasing the quality. Digital reference checks collect information from a variety of sources, allowing employers to see the rater’s relationship with the candidate and to inform them whether they’re getting information from a supervisor, peer, or subordinate.

Methodology: This research study was conducted across large organizations that use Checkster in the hiring process, with a total number of hires equaling 11,549 and over 51,000 references responding. The data analysis involved statistical tests including point-biserial correlations and t-tests.

To download a free copy of the report, visit

All Stories