Almost every mentoring program I've watched get launched has followed the same arc. Enthusiasm at rollout, careful pairing, a kickoff, calendars set. Six months later half the pairs have quietly stopped meeting. Twelve months later nobody mentions it.

Then it gets relaunched two years on by someone new, and the cycle repeats.

The failure isn't the idea. Developing people through relationship is genuinely how it works. The failure is assuming that development is a scheduling problem, when it's actually a content problem.

Why the pairs stop meeting

Ask people why the meetings stopped and you'll get "we got busy." That's the polite answer. The real one, in my experience, is that they ran out of things to talk about.

A mentoring conversation with no specific work at its center becomes a career chat. Career chats are pleasant and there are maybe three of them available before it's just catching up. Neither person wants to say it's not useful, so the calendar invite gets moved, then moved again, then dropped.

The pairs that survive are almost always the ones where the mentor has some actual connection to the work the mentee is doing — where the conversation can be about a real decision the mentee is facing this week.

That's the whole insight, and it has a practical consequence: proximity to the work beats seniority in the pairing. Companies tend to pair junior people with the most senior person available, which maximizes the status of the pairing and minimizes its usefulness.

What actually develops people

Looking at the people I've watched grow fastest — including the ones I had a hand in — almost none of it came from a program. It came from a few specific things.

Being given a decision that was genuinely theirs. Not a recommendation to be approved. An actual call, with real stakes, that they owned including the outcome. Nothing else comes close for compressing development time. It's also the thing leaders are most reluctant to do, because it means accepting a worse decision in the short term to get a better leader in the long term.

Being in the room for conversations above their level. Most executive judgment is pattern recognition, and you can't build the patterns from summaries. Bringing a promising person into the board prep, the hard customer call, the layoff discussion — as an observer with permission to ask questions afterward — does more than a year of structured meetings.

Getting specific feedback within days, not quarters. "You lost that room when you led with the data instead of the conclusion" is worth more than any amount of general encouragement, and it's only useful while the memory is fresh.

Watching someone senior handle something badly and then talk about it honestly. This one is underrated. People learn enormous amounts from seeing a leader misjudge something and then narrate their own error without defensiveness. It teaches both the lesson and the norm.

None of those require a program. All of them require leaders willing to give up control and be publicly imperfect, which is why they're rarer than programs.

The cheapest version that works

If you want to build this deliberately without the machinery, three things get most of the value.

Make the "why" explicit when you decide something. When you make a call, take an extra sixty seconds to say what you weighed and what you were worried about. Most of what people need to learn is the reasoning, and leaders skip it because it's obvious to them. It isn't obvious to anyone else.

Invite one level down into one hard conversation a month. Not to participate — to watch. Then spend ten minutes afterward on what they noticed.

Give feedback in the moment, small and specific. The instinct is to save it for a formal setting, where it will arrive too late, carry too much weight, and be about a situation nobody remembers clearly.

That's it. No pairing spreadsheet, no program manager, no relaunch in two years.

If you're going to run a formal program anyway

Sometimes there's a real reason — scale, or a specific population you're trying to develop. If so, a few things meaningfully raise the survival rate:

  • Pair on the work, not the org chart. Proximity to what the person actually does beats title.
  • Give it a defined end. Six months with a clear finish, renewable by mutual choice. Open-ended relationships die awkwardly; defined ones end cleanly and often get renewed.
  • Give the mentor something specific to do. "Meet monthly" is not a job. "Help them get ready to run their first cross-functional project by June" is.
  • Measure whether it's still happening, not whether people liked it. Satisfaction surveys will be positive right up until the pairs stop meeting.

The part that isn't about technique

The best mentors I've had weren't the most accomplished people available to me. They were the ones willing to tell me something true that I didn't want to hear, and who had enough relationship with me that I could actually hear it.

That's not a program feature. It's a character trait, and it's the reason mentoring cultures tend to reflect whether the senior people in a company are genuinely interested in other people succeeding — or just interested in being seen developing talent.

You can tell the difference from about two levels down, and so can everyone else.