Nobody slams a door. That is what makes it hard to spot.
The first meeting goes well. The second is moved to next week, then to the week after. The third happens on a video call, ten minutes late, with your mentor reading something else on a second screen. By the end of the year you have a mentor in name, a calendar full of rescheduled invitations and a quiet suspicion that the problem must be you.
It may not be you. What happens next is yours to decide all the same, because a mentor who has drifted is rarely the one who raises it. This piece is about spotting the fade while it is still small, and about what a mentee can do besides wait.
How a mentor goes missing
Bruno Andrade has mentored medical scientists and physicians for over two decades, in several countries. In an opinion article published in Frontiers in Medicine in December 2025, he sets out the ways he has watched mentoring go wrong. His central observation is that the harm is rarely a single blow-up. It accumulates through small lapses that nobody stops.
One of his composite sketches is a junior infectious diseases fellow paired with a renowned tuberculosis researcher. Meetings slip, then stop. Deadlines are flexible for the mentor and rigid for the fellow. Feedback arrives as late-night line edits, with no conversation about direction. No single episode feels serious. A year on, the fellow has a thin CV, a narrow set of skills and a growing sense of dread.
Andrade also names the recognizable types. The Absentee Sponsor offers prestige without presence. The Extractor treats a mentee as labor. The Gatekeeper rations introductions and visibility. The Unsafe Teacher uses embarrassment as a teaching method. The Boundary Crosser blurs the personal and the professional.
A fair caveat. This is an essay built on experience and earlier research, set in academic medicine, and it reports no new data. Swap "authorship" for "credit on the project" and most of it travels to an office without difficulty.
Why it is rarely about you
The standard advice to a neglected mentee is to be patient. Senior people are busy, their time is a gift, and it would be ungracious to count the cancellations.
Andrade's argument cuts against that. Bad mentoring, he says, is a predictable product of how the work is organized. Time pressure turns mentoring into an optional extra, so it is the first thing dropped. One person is often supervisor, evaluator and sponsor at once, which makes an honest conversation harder for both sides. And the rules about who gets credit and opportunities are rarely written down.
The part that matters most for a mentee is what happens next. Trainees, he observes, gradually reinterpret these structural problems as personal failure. They decide they are "not ready," even when their work says otherwise. He lists dread before sending a routine email as a warning sign in its own right. If you have started rehearsing a two-line message to your mentor, treat that as information about the relationship.
What to do differently
Count, don't feel. Andrade advises looking at patterns over months, not single episodes. One canceled meeting is life. A run of them is a signal. Keep brief notes of what you agreed, when, and what happened next. The record helps you see a pattern you would otherwise talk yourself out of.
Raise it once, plainly, with a proposal. His first step is repair: a short written plan with a time limit, covering what you are working toward, how often you will meet and who does what. A mentor who drifted through overload may be relieved that somebody has supplied the structure.
Widen before you walk. If the plan produces only cosmetic change, add a second mentor, a sponsor outside your reporting line or a network your current mentor does not control. Leaving then becomes a smaller step. Andrade treats exit as managing a risk, with no betrayal involved.
Know what is not a fade. Harassment, discrimination, retaliation or someone taking your work belong in a different category. The aim there is protection. Write dated, specific notes and seek confidential advice from someone you trust in the organization.
Ask about presence next time. Prestige is easy to see and presence is hard, so ask a prospective mentor how often they meet the people they mentor now, and whether you could speak to one of them.
The awkward conclusion
A fade takes two people. Your mentor stopped turning up. You stopped mentioning it.
Chapter 10 of The Accidental Mentee describes the slow fade as the way most mentoring relationships die, and observes that nobody ever decides it. The flat session becomes the canceled one, then the unanswered message, and in time neither of you can say when it ended. The chapter is frank about the mentee's share. A mentor's hour is unpaid, which makes raising a problem feel ungrateful, so most mentees say nothing. One of its exercises asks where your own fingerprints are on the problem, and the cancellations you accepted without comment are on the list.
The book's remedy starts with a check you run every third session or so, whether or not anything feels wrong, the way a pilot scans the instruments on a schedule. You score the relationship from 1 to 5 on six points, one of which is reliability: do sessions happen as planned? A low score becomes an agenda item, and it hands you a sentence that is easy to say: "I have noticed our last three sessions slipped." Then comes a decision the book calls Repair, Reset or Release. A pattern you have never raised gets a conversation before it gets an exit. A pattern that improved and then slipped again means the arrangement needs redesigning. A pattern that survives being raised means it is time to end things, briefly and warmly, with a thank-you note inside the week. The one option the tool leaves out is to carry on waiting.
Your mentor may be overloaded, distracted or plain bad at this. One short message will tell you which. The worst a ghost can do is fail to reply, and you already know what that feels like.
Source
Bruno B. Andrade, "When Mentorship Goes Wrong: From Personal Vignettes to a System-Level Playbook for Safer, Fairer Academic Training," Frontiers in Medicine 12 (2025): 1714113, https://doi.org/10.3389/fmed.2025.1714113.
