The phone call that ended a candidate’s chances, the one I remember best, lasted about four minutes, and the candidate never knew it happened.
A hiring manager at another company reached me because we shared a contact. They were about to make an offer to someone who had spent two years delivering projects into my operations group at a large telecom. The candidate’s listed references were glowing: a former director, a vendor partner, a peer from a different department. None of them had ever been on a bridge call at two in the morning when one of this person’s releases took down a provisioning system. I had been on several. I didn’t trash anyone. I answered the questions honestly, said I would not hire them into a role that owned production changes, and explained why. The offer went to someone else.
I’ve thought about that call for years, mostly from the other side. Every product manager has a version of me somewhere in their history: a person they never would have listed, who watched their work from the receiving end, and who will answer the phone.
The reference list is the least informative part of the check
Formal reference checks still matter. When Robert Half’s staffing arm surveyed more than 2,800 senior managers in the US, respondents said they removed 34 percent of candidates from consideration after a reference check. The Canadian version of the survey, covering more than 600 managers, landed at 32 percent. Roughly one candidate in three gets cut at the stage most people treat as paperwork.
But the references you pick are a weak signal, and hiring managers know it. You chose people who like you. Research on referee behavior keeps finding that liking drives rating, and a study of more than 1,700 managers in the US and UK by Kathryn Jacobson, Ryan Jacobson and Carl-Johan Holmberg found managers rated former employees more positively in a reference check than they did in private. The authors’ explanation is uncomfortable and believable: a manager praising a former report is partly praising their own management.
So the people doing the hiring go around the list. For senior roles in tech this is close to standard practice. Bain Capital Ventures’ field guide for its portfolio companies, published in November 2025, describes backchannels as informal, off-the-record conversations with trusted contacts who have worked with the candidate, and treats them as the way to fill gaps the interviews and formal references leave open. Its main etiquette rule is to never call anyone at the candidate’s current company. Everyone else is fair game.
Product managers leave more witnesses than almost anyone
An engineer’s work is seen mostly by other engineers. A product manager’s work lands on everybody. In a normal year a PM makes decisions that an engineering team builds, a design team defends, a support team absorbs, a sales team sells, and an operations team keeps running after launch. Each of those groups formed an opinion of you, and very few of them are on your reference list.
That is why the backchannel check for a PM tends to surface different things than the formal one. Your listed director will talk about strategy and outcomes. The engineer you worked with for eighteen months will talk about whether your requirements changed after the sprint started, whether you took the blame in the incident review or let the team take it, and whether you said “the business wants this” when you meant “I want this.” The support lead will remember whether you ever read the tickets.
The question that does the most damage is a simple one. Keith Rabois, describing how he checks executives in an interview with Elad Gil, says he always asks: “If this person joined my company, would you join?” He listens for the spark. A polite yes with no energy behind it is the answer that gets probed. Nobody prepares their backchannel for that question, because nobody knows who their backchannel is.
You can’t manage the call. You can manage what the caller remembers.
The practical advice in most career articles is to prep your references, and that is fine as far as it goes. It doesn’t touch the problem. You can’t prep people you didn’t choose, and trying to would be strange. What you can do is treat the backchannel as a running account that you add to every week, mostly through behavior that has nothing to do with job searching.
A few practices that change what people say about you when you are not in the room:
Find out what your receiving end thinks, while you can still act on it. Once a quarter, ask one engineer and one person from a downstream team a version of the Rabois question about yourself: what would make them hesitate to work with you again? Most people won’t answer that straight the first time. Ask again in a month. The pattern in what they avoid saying is the information. This connects directly to why senior PMs stop getting honest feedback; the backchannel is the feedback you never received, delivered to a stranger.
Own the incident review in public. The single thing operations people remember about a product manager is what happened after a bad launch. If you stood up and said the call was yours, that story travels. If you were mysteriously unavailable, that story travels faster.
Make your decisions visible enough to be credited correctly. A lot of PM work is invisible, and invisible work gets attributed to whoever was loudest. Written decision records, short and dated, mean the engineer who gets the backchannel call can say “they made the call to cut the reporting module, and they were right” instead of “I think that was their project?” The same habit protects you from the glue work trap, where you did the coordinating and someone else got the credit.
Leave well, especially when leaving badly would feel good. The people at the job you are leaving become backchannel sources the day after your last day, and the last month is the period they remember most clearly. Finish the handover document. Write the context nobody asked for. Don’t use the exit interview to settle scores.
When you are the person who picks up the phone
Sooner or later you will get the call yourself, if you haven’t already. A former colleague’s future employer will find you through a mutual connection and ask what it was like to work with them.
The Jacobson study had a useful finding here too: when referees were asked to take the hiring manager’s perspective, their ratings got less lenient and more accurate. That’s a fair standard to hold yourself to. Answer the way you would want someone to answer you if you were the one about to hire. Stick to what you directly saw. Separate what someone did from how you felt about them; Bain’s guide makes the same point from the caller’s side, warning that a review from a close friend or from someone the candidate fired may say more about the relationship than the person. If you only worked with them for three months on one project, say so, because the caller will weigh you accordingly.
And decline when you should. If you can’t be fair, say you don’t know them well enough to help. That is a complete answer.
I’ve been on the receiving end of maybe a dozen of these calls over my career and made more than that. The pattern that held across almost all of them: nobody’s career was decided by one dramatic story. It was decided by whether the people who worked beside them, unprompted and unprepared, sounded like they would do it again. That’s built in ordinary weeks, on ordinary projects, long before anyone is looking. If you’re about to look for your next role, it’s also worth doing the reverse, and running your own backchannel on the manager you’d be working for.