Building design and product organizations at Honor, Lyft, Grammarly and Pitch, from first hire to dozens later, I spent more hours on reference calls than on portfolio reviews. It wasn't close. Somewhere in there the calls stopped being a formality and became the most informative half hour in the entire process.
The reference call is the interview
Most teams treat references as the last box to tick before an offer. Confirm the dates, confirm they weren't fired, hang up. That's a waste of the one conversation in your process with someone who has actually watched this person work for two years, under pressure, without an audience.
Once I started treating the call as a source of evidence rather than a rubber stamp, two things changed. I ran them earlier, sometimes before the final loop. And I stopped asking questions that had a socially correct answer.
"Would you hire them again?" has one answer. Everybody says yes. That question does no work. Though I have sat at the other end of it. Writing a reference at Honor, the only line I was willing to stand behind was whether I would hire them again. That is the shape of the thing from the inside: one sentence you would repeat in a room, not a paragraph of private detail. Which is exactly why asking for it gets you nothing. You are requesting the summary someone has already decided to give.
Ask what the person is like to manage, not whether they are good. Everyone is good. Almost nobody is easy to place.
Four questions that actually move
These are the ones I kept, after discarding a lot.
"What were they hired to do?" Not what they did, but what the mandate was on day one. All the interesting information lives in the gap between the mandate and the outcome. Sometimes you learn the person rewrote a broken remit on their own initiative and saved the team. Sometimes you learn they were hired to build a design system and spent two years doing marketing collateral because they couldn't say no.
"What would their manager say were their strengths and weaknesses at that time?" The distance is what does the work. Ask someone to speak for a third party and you get candour a direct question never produces, because they are not the one on the hook for the verdict. This single reframing produced more usable signal than any other change I made.
"In what context did you work with them?" Asked first, always. A peer on the same project, a skip-level who saw them twice a quarter, and a direct report hold three conflicting views. What weight you give the rest of the call depends entirely on which one you're talking to. This matters most with senior candidates, where the warm reference is usually a peer. I've watched hiring committees treat a warm reference from someone who barely worked with the candidate as equivalent to a measured one from their closest collaborator.
"Why did they leave?" Then stop talking. A pause is the instrument. First answer is the official one. Second answer, if it comes, is the real one. It arrives about four seconds into a silence most interviewers rush to fill.
What the calls kept telling me
The strongest predictor, I think, is whether someone fits a particular kind of mess. Designers who thrive at seed stage often stall at scale, and the reverse is more common than anyone admits. Someone brilliant inside a functioning system can be helpless without one. A reference call is where you find out which environment produced the work in the portfolio. Portfolios are silent on this. A case study never shows you whether there was a PM, a researcher and a working design system, or whether the person did all of it at 11pm.
Glue work, the holding-it-together kind that leaves no artefact, is real and undervalued. You can hear it on a reference call. Some people show up in every reference as the one who kept the team together, who built the relationships, who made the cross-functional thing work. It rarely appears in their own portfolio because it doesn't screenshot. If you're hiring into a team that's fragmenting, that person is worth more than a better visual designer, and you will only ever hear about it on a call.
Conflict-avoidant managers produce a specific kind of damaged designer. When a reference says someone's previous manager was "a great communicator, well-liked, avoided difficult conversations," take it seriously. People who spent three years without honest feedback often arrive with a distorted sense of their own level, through no fault of their own. That's something you can coach, but only if you know to do it.
Almost nobody is a bad hire in the abstract. After enough calls the question stops being is this person good and becomes what does this person need to be good, and can we give them that? That is a far more answerable question, and it puts the responsibility where it belongs: on the hiring team, not the candidate.
The obligation
One thing I want to be direct about. These conversations exist because people speak candidly about former colleagues on the understanding that it goes no further. That understanding is the whole mechanism. If it breaks, you get nothing but platitudes forever after, and every hiring team downstream of you is worse off.
So: don't repeat what you're told to the candidate. Don't put verbatim notes in a shared drive. Don't quote a reference in a debrief where twelve people can read it. Write your conclusion, not your evidence.
I'm reasonably sure I got this wrong for the first few years, keeping detailed notes because they felt rigorous. They were also a risk sitting in a folder, and it took me longer than it should have to see that.
If you only change one thing
Move the calls earlier and ask the manager-perspective question. It costs you nothing, it takes half an hour, and it will tell you more about how this person will land in your team than a fourth interview loop ever does.
Then design the role around what you learned. That's the part most teams skip. You do all the work of discovering that someone needs a strong PM partner, or protection from stakeholder churn, or a manager who gives direct feedback. Then you onboard them exactly like everyone else and act surprised at month five.
Who this disadvantages
Weighting the reference call this heavily has a cost, and it lands on predictable people.
A strong reference is partly evidence about the candidate and partly evidence about how they were treated. Somebody who spent three years with a manager who invested in them will produce a former manager who speaks fluently about their growth. Somebody who spent three years being managed badly will produce a reference who is vague, or brief, or slightly cool, and the difference reads as a difference in the candidate.
It gets worse for anyone who left a job because it was unbearable. Best placed to speak for them are often the people they left. Asking a candidate to nominate a reference from a role that damaged them is a request most hiring teams make without thinking about it at all.
It also makes the network problem worse. Backchannel references, which I have used and would use again, run on who you already know. A process that leans on them favors candidates from the same handful of companies as the hiring manager, and nobody ever chose that.
I do not have a clean answer. So I treat a thin reference set as a question rather than a finding. I ask directly whether there is context I should know about who they have and have not nominated. And I put more weight on the working session for anyone whose record suggests they have been badly served. That is a partial correction and I would rather name it than pretend the instrument is neutral.
I've built design and product organizations from first hire to dozens later, at Honor, Lyft, Grammarly and Pitch. One honest limit on all of the above. References are a biased sample. You are talking to people the candidate chose. Backchannel references correct for that and introduce their own problems, which is a separate essay.
© 2026 Renato Valdés-Olmos