Behavior-anchored referral questions outperform generic quality assessments in hiring and investment

Generic referral questions are a remarkably durable habit in hiring, and a remarkably poor one. “Who’s the best engineer you know?” is the kind of question that feels productive and almost never is. The person asked searches their memory for a name that fits an undefined category, offers someone safe, and the conversation moves on without yielding anything that predicts performance. A different approach is beginning to surface, one that replaces quality assessments with behavior-anchored questions, and the logic behind it is worth examining.

Adam Ward lays out the mechanic directly. Rather than asking for the best person, Ward argues for questions that target specific, observable behaviors. His framing: “Who stands out to you as the person who’s most collaborative with designers? Or who can translate a framework into a product better than anyone you’ve ever seen?” The shift is meaningful. A question like that gives the respondent a concrete cognitive task. They are no longer ranking an abstract concept of quality; they are scanning their actual experience for a person they watched do a specific thing well. The resulting name is grounded in something real.

This matters because generic questions do not just produce imprecise answers. They produce answers shaped by proximity and recency bias: the last person the respondent worked with who seemed impressive, rather than the best match for a defined need. The behavioral framing is an attempt to address that by giving respondents a more tractable question, one tied to a specific skill or working style rather than an undifferentiated sense of general excellence.

Nothing beats you and I sitting down every other week for 6 weeks and diving into people. Adam Ward

Ward adds a structural point that is easy to underestimate. The best extraction of candidate referrals happens through repeated, in-person sessions, not through automated outreach or one-off requests. His view: nothing replaces sitting down with someone every other week for six weeks and working through the question together. That cadence is not incidental. Memory for specific behavioral episodes takes time to surface, and the relationship built through repeated contact makes respondents more willing to surface real names rather than safe ones. Automation, in his framing, is a poor substitute for that dynamic.

The pattern is not confined to hiring. Julien Bek, speaking from an investor’s perspective, describes having learned to focus more on references than almost anything else when evaluating a founder. That is a notable claim in a domain where pitch quality and presentation often dominate the conversation. Bek’s emphasis on references suggests the behavioral-signal logic is traveling beyond recruiting departments and into the early-stage investment process, where the candidate, so to speak, is the founder and the reference check is doing work that the pitch deck cannot.

What ties these observations together is a shared skepticism about surface-level quality signals. A polished pitch, a strong impression in an interview, a reputation as “one of the best” in a given field: all of these can be constructed or amplified in ways that behavioral history cannot. A question about who specifically was most collaborative with designers is harder to game because the respondent is drawing on memory of actual events, not summarizing a general impression.

The practical implication is narrow but consequential. Hiring managers and investors who build referral processes around open-ended quality questions are generating noise they then have to interpret. Switching to behavior-anchored questions, tied to skills the role actually requires, is a low-cost change that targets the retrieval process more precisely. The cadence Ward recommends requires real time investment, and that is likely why it remains underused. The easier path is to send a referral form and wait. The harder path, the one that produces the signal, is the one that looks more like a relationship than a survey.