The mentorship program launched with a town hall. The CEO talked about the power of peer learning. The CHRO showed a video of last year's mentors describing the "transformative" experience. There were 47 slides about how the program would work. Three different tracking systems. A carefully designed matching algorithm that would pair high-potentials with senior leaders.
On week one, 312 mentee-mentor pairs were activated. On week eight, 143 of them had never met.
Of the 169 pairs that did meet, the average conversation duration was 23 minutes spread across 4 months. The mentor survey asked: "How much did this experience change how you think about developing the next generation?" 61% said "not at all." The mentee survey asked: "Did your mentor help you understand the decision architecture of this organization?" 8% said yes.
The program ran for 18 months. Then it was archived.
Nobody said it failed. Instead, the organization adopted a new framework. More mentorship, but differently structured. A peer-to-peer model. Smaller cohorts. Quarterly check-ins instead of ad hoc. More tracking. New software. A second iteration of the same failure, with better project management.
This is what I see across Fortune 500 L&D functions: Mentorship programs are designed, launched, measured, and archived on a cycle of 18-24 months. Very few of them produce anything resembling actual mentorship. And yet, new ones are built constantly. The investment in mentorship programs is among the highest in corporate L&D. The ROI is among the lowest.
The gap is not between the design and the execution. It's between what the organization says mentorship is and what mentorship actually requires.
The Organization's Story.
Mentorship is positioned as knowledge transfer. A senior person shares frameworks. A junior person absorbs them. The mentor is a repository of experience, and the relationship is the vessel. The program design flows directly from this: match people well, give them meeting prompts, track the outputs, report the metrics.
But this story solves a problem that mentorship doesn't actually solve.
What the organization actually needs from mentorship is judgment transfer. Not what to think, but how to think. Not the playbook, but how to read the room. Not the answer, but how to ask the question. Judgment is not transferable through a 23-minute conversation. Judgment is learned by watching someone think under pressure, repeatedly, until the pattern becomes visible.
Mentorship that produces judgment transfer requires proximity, not once-a-quarter but embedded in real work. It requires stakes: not a safe conversation, but feedback after a failed decision. It requires discomfort, where the mentor tells you what you don't want to hear because the stakes are real. And it requires time, measured in months, not hours.
The corporate mentorship program provides: matching software, meeting prompts, 4 conversations per year, a final survey.
Why the Program Can't Work.
The organization is trying to solve a mentorship problem with a training system. These are incompatible architectures.
Training systems are built on scheduled contact time, predictable content, measurable outputs, scalability across many participants, and closure: start date, end date, completion certificate. Mentorship requires the opposite on almost every dimension. Unscheduled, on-demand access. Emergent content driven by problems the mentee brings, not topics the mentor predicts. Outputs that are unmeasurable in real time. An architecture that cannot be scaled. And no closure, because genuine mentorship doesn't end; it deepens.
When an organization builds a mentorship program using training system architecture, it has already failed before the first meeting is scheduled. The mentors are selected for their availability, not their judgment. The mentees are selected by a matching algorithm, not by seeking. The content is predetermined by the prompts, not emergent from what the mentee actually needs to learn. The success is measured by completion and satisfaction, not by judgment development.
Every step of the design is optimized for the wrong thing.
The Real Cost.
The mentorship program doesn't just fail to produce judgment transfer. It actively disrupts it.
When an organization announces a mentorship program, it sends a signal: mentorship is what happens in this program, in these four scheduled meetings, with this matched partner. Everyone else hears: mentorship is not happening outside the program. If you want to learn from someone, wait for the official structure.
The informal mentoring that was actually producing judgment transfer gets crowded out. The senior engineer who spent 90 minutes with a high-potential, explaining why a proposal would fail in political reality, becomes background noise. The mentee is now waiting for their matched mentor, at the scheduled time.
The organization measures the program. 312 pairs. 143 meetings. 8% reported judgment development. The data feels true, but it masks the real failure: the informal mentorship that died so this program could exist.
The cost of the mentorship program is not what it costs to run. It's what it costs the organization to lose the informal judgment transfer that was happening organically before the program formalized, and therefore killed, it.
What Would Actually Work.
If an organization wanted to produce judgment transfer, the starting point would be to stop trying to scale it. Judgment transfer is not scalable. A senior leader can mentor two or three people seriously. Not twenty. Not with the same program.
It would require stopping the impulse to measure it. You cannot measure judgment transfer in real time. You can measure it three years later, when the mentee is making different decisions under pressure. That timeline is too slow for the program management office, which is part of the problem.
It would require identifying who is actually doing the mentoring. In most organizations, there are eight to twelve senior leaders who are natural mentors: people who attract junior talent and invest real time in developing them. These people are often not the most senior. Not the most visible. When identified, the organization's job is to protect their time. Not to enroll them in a program, because the program will dilute their effectiveness and eventually extract them from the informal relationships where the real work was happening.
The mentorship program is not a mistake of execution. It's a mistake of design. And it's a choice organizations make with full visibility of the alternative.
The most structurally honest thing an organization can do is separate two different investments that get conflated into one word. For 300 high-potentials, run a structured cohort learning experience. Name it accurately. Measure it against cohort learning outcomes. For the top 50 among them, create the conditions for real mentorship: access to senior leaders in live work situations, real stakes, unscheduled conversations, and a three-year development horizon. Then stop calling both of them mentorship, because they are not the same thing and they cannot be built with the same architecture.
SSUNDAR's diagnostic work on L&D architecture has consistently surfaced this in Fortune 500 organizations: the informal judgment transfer that was working gets catalogued, systematized, and scheduled into irrelevance. What remains is a program. What disappears is the development.
The Uncomfortable Truth.
The reason organizations build mentorship programs is not because they want mentorship. It's because they want to show that they care about development, and they want to measure that care. A program is visible. Mentorship is invisible until three years later, when the mentee is ready for a bigger role and no one quite knows how they got there.
The program tells a story the board understands: investment, participants, completion rate, satisfaction score, archived. Mentorship tells a story that resists every template the L&D function has for reporting it.
So the organization runs the program. The numbers look reasonable. The mentors feel recognized. The mentees feel supported. The CHRO can point to it in the annual report. The board nods. And the judgment transfer that was happening quietly in hallways and late conversations has been replaced by a tracking system that measures how many times two people met.
Most organizations choose the program.