Institutional Knowledge and the Transition Problem

The new provost is six weeks into the job when the reaffirmation letter arrives. The regional accreditor wants a narrative on how the institution arrived at its current general-education model — the rationale, the dissenting departments that were brought along, the compromise that closed the 2019 faculty-senate fight. None of it is in a file. The dean who ran that process retired in March. The provost calls him at home. He is gracious, and his memory is good, but he is reconstructing from recollection a set of decisions that took two years and a dozen people to make. Some of it he gets right. Some of it he has quietly merged with a different reaccreditation cycle. Nobody in the room can tell which is which.

This is not a filing problem. The institution has filing cabinets, shared drives, a document-management system the size of a small library. The accreditation self-study is filed. What is not filed — what was never a document in the first place — is the reasoning behind it: why the institution chose the path it chose, what it weighed, who it had to persuade, and what it would do differently. That layer lived in the dean's head. When he left, it left with him.

Every leadership transition is a knowledge-loss event. Higher education in 2026 is running more of them, faster, than at any point in living memory — and stacking two other forces on top that attack the same asset from different directions. The institutions that survive the transitions they didn't see coming will be the ones that figured out how to make what they know about themselves outlast the people who happen to know it.

Three forces, one casualty

Institutional memory has always leaked. Tenure-track faculty retire, long-serving administrators move on, the person who knew how the budget model actually worked takes a job at a peer institution. What is different now is the rate, and the fact that three distinct pressures are converging on the same vulnerability at once. Each is well-documented; together they compound.

None of these is a new story on its own. The new story is the overlap. Deloitte's 2026 Higher Education Trends frames the year as one of structural pressure: institutions making consequential decisions — about programs, partnerships, and operating models — under conditions of demographic and financial strain. Tyton Partners' outlook for 2026 describes a sector reorganizing itself in real time. Reorganization is precisely the condition under which institutional memory is most valuable and most fragile: the people who hold the context are the same people whose roles are being eliminated, merged, or vacated.

The casualty in all three cases is the same. It is not data — the institution keeps its data. It is the layer of reasoning, relationship, and rationale that sits on top of the data and explains what the data means and why it looks the way it does. That layer is what walks out.

What actually gets lost

It is worth being specific about what disappears in a transition, because the loss is rarely visible until someone needs the thing that is gone. The institution does not lose its records. It loses the connective tissue between the records — the answers to “why” that no one wrote down because, at the time, everyone in the room already knew.

The cost is paid in re-litigation. A new dean reopens a question that was settled and closed five years ago, because the file shows the decision but not the reasoning, and without the reasoning the decision looks arbitrary. A development officer rebuilds a donor relationship from the dollar amount in the database, having lost the eight years of context about what the donor cares about, who they trust, and the one topic never to raise. A procurement team signs with a vendor the institution already knows is a bad fit, because the person who learned that lesson the hard way has retired. None of these failures show up as a missing file. They show up as an institution that keeps paying tuition to learn things it already knew.

Documents are not memory

The instinct, faced with this problem, is to write more down. Better documentation. Mandatory transition memos. A “knowledge capture” checklist for departing staff. These help at the margins, and they are worth doing. But they run into a structural wall that every knowledge-management initiative eventually hits: the most valuable knowledge is the knowledge that is most expensive to document, and people document it least precisely when they are busiest — which is to say, always.

There is a name for this in the knowledge-management literature: the distinction between explicit knowledge, which can be written down, and tacit knowledge, which lives in judgment, relationship, and pattern recognition. McKinsey's State of Organizations research locates one of the central organizational challenges of the era exactly here: in workforce churn and the retention of knowledge that organizations have never successfully made portable. You cannot solve a tacit-knowledge problem with an explicit-knowledge tool. A departing dean asked to “document everything important” will produce a memo that captures what he can articulate under deadline pressure — which is a fraction of what he knows, and rarely the load-bearing fraction.

Worse, the transition memo is a snapshot taken at the worst possible moment: on the way out, after the institutional attention has already moved to the successor. The knowledge that mattered was being created continuously, over years, in the course of ordinary work — in the emails, the meeting decisions, the corrections, the “actually, the reason we don't do it that way is” asides. By the time you ask someone to capture it, most of it has already become invisible to them. It is just how things are done.

A transition, watched in slow motion

Play a single transition forward and the leak becomes visible. Each stage is ordinary — nothing goes obviously wrong — and the institution loses a layer of itself at every step.

The pattern is the same whether the trigger is a presidential succession, a program merger forced by enrollment, or a budget-driven reorganization. Run several of these at once, across a stressed institution, and the cumulative effect is an organization that increasingly does not know why it is the way it is. That is the transition problem. It is not an HR problem or a records problem. It is a memory problem, and it is structural.

Persistence as a buffer

The structural fix is not to document harder. It is to change where the knowledge lives, so that capturing it does not depend on a busy person remembering, at the moment of departure, to write down what has become invisible to them. The knowledge has to be captured continuously, as a byproduct of the work itself, into a system that belongs to the institution rather than to any individual.

Consider what that means concretely. When a committee makes a decision and someone explains the reasoning, that reasoning should be captured as it is spoken, not reconstructed years later. When a development officer notes that a donor responded badly to a particular ask, that observation should attach to the institution's memory of the donor, not to the officer's. When a new piece of information contradicts what the institution thought it knew — the budget assumption that turned out wrong, the policy that was quietly reversed — the contradiction should surface rather than sit silently in two incompatible files. And all of it should carry its own timeline, so that “what we decided” is distinguishable from “what we used to think.”

This is the inversion that generative AI makes newly affordable. For most of the history of knowledge work, capturing institutional memory in a structured, retrievable form was more expensive than reconstructing it on demand — so institutions reconstructed, transition after transition, paying the tuition again each time. That cost-benefit calculation has flipped. The capture can now happen automatically, in the course of ordinary use, which is the only way it ever gets done at all.

The other half of the buffer is the retrieval layer. Capture only matters if the next person can find what was captured, in the moment they need it, without knowing in advance that it exists. The new provost facing the accreditation narrative should be able to ask, in plain language, why the institution structured its gen-ed model the way it did — and get an answer grounded in the institution's own record, with citations pointing back to the source material, rather than a phone call to a retiree's recollection. That is the difference between an archive and a memory: an archive holds documents; a memory answers questions.

The transition you didn't see coming

The uncomfortable truth about transitions is that the costly ones are rarely the planned ones. A retirement announced two years out can be managed. The transition that hurts is the one nobody scheduled: the provost recruited away in a single news cycle, the reorganization ordered by a new board, the program merger that consolidates three offices into one over a summer. You cannot prepare for the specific transition, because you do not know which one is coming. You can only prepare for the category — by building an institution whose memory does not depend on which particular people happen to be holding it this year.

That reframes the work for higher-ed leaders facing 2026. The goal is not to slow the churn, which is largely outside any one administrator's control, and not to extract heroic documentation from departing staff, which does not scale and never has. The goal is to make the institution's knowledge accrue to the institution — captured as a byproduct of the work, retrievable on demand, carrying its own confidence and its own timeline — so that each transition is a change of personnel rather than a loss of self.

An institution that remembers can absorb a transition. An institution that relies on the people who happen to remember is one departure away from forgetting why it does what it does. In a sector running more transitions than it ever has, under more pressure than it has faced in a generation, that distinction is no longer a nicety of good administration. It is the difference between an institution that survives the transition it didn't see coming and one that quietly pays, again, to relearn what it already knew.