July 31, 2026 · Operations, Emerging Managers
What "institutional memory" means for a 3-person fund
By Emily Buckley, founder of SamCIO
Institutional memory sounds like a large-firm concern, which is why most small funds skip it. The phrase suggests archives and knowledge bases and a person whose job is to maintain them. At a three-person fund that framing is correct and useless.
The working definition is narrower and much more practical: institutional memory is the ability to answer why you decided something, years after you decided it, when the people who were in the room have changed or simply forgotten. By that definition a three-person fund needs it more than a thirty-person firm, not less, because at three people the entire decision record lives in two or three heads and there is no redundancy at all.
The four questions that define it
You have institutional memory if you can answer these about any deal from the past three years, in minutes, without calling anyone:
Why did we do this? Not the outcome. The case at the time, including what you thought the risks were.
Why did we pass on that? Harder, and more valuable. Passes are the larger dataset and the one nobody records.
Who disagreed, and what did they say? The dissent, in the dissenter's words, not as remembered by the person who won the argument.
What did we say would happen? The falsifiable part of the thesis. Without it you cannot tell a good decision from a lucky one.
Most small funds can answer the first question and none of the other three. That is the actual gap, and it is not a tooling gap so much as a habit of writing down the reasoning at the moment it is cheapest to write.
Decisions, not documents
The instinct is to solve this with storage: a folder structure, a wiki, a Notion database with a page per company. Funds build these and they decay within two quarters, reliably, for a structural reason. The knowledge base is a second system that requires effort separate from doing the work, and separate-effort systems lose to deadlines every time.
The version that survives is the one where the record is a byproduct. If the memo is what the committee decides on, the memo exists without extra effort. If the vote is cast in the system that tracks the deal, the vote record exists without extra effort. If the pass reason is a required field on moving a deal to passed, the pass rationale exists without extra effort.
Everything that requires a separate act of documentation after the decision will be skipped, and it will be skipped worst during the busy stretches, which are exactly the periods you will most want to reconstruct later.
What to capture, specifically
Four things, and only four, are worth being strict about.
The memo, in a consistent structure. Consistency is doing more work than length here. Two memos with the same sections are comparable; two thorough memos with different shapes are two documents. Comparability across deals and across vintages is what makes the archive queryable by a human rather than just searchable.
The vote, per person, with rationale. A recorded outcome tells you what happened. A recorded distribution tells you how the partnership reasons. When SamCIO logs a yes, no, or abstain per partner with an optional rationale, the useful artifact is not the tally, it is that the one person who was uneasy said why, in writing, before anyone knew the answer.
Changes of mind, not just end states. A vote that flipped after a reference call is one of the more informative records a fund can have, and a system that shows only where everyone landed hides it. Two things get you most of the way: a time-stamped activity trail, so the sequence of positions on a deal is visible, and versioned memos, so you can see what the case looked like before the revision. A current-state-only record is not memory.
The pass rationale. The single highest-leverage habit on this list, and the one most funds skip entirely. Passes outnumber investments by a large multiple, and they are the only way to measure your own false negatives. When a company you passed on raises again at four times the price, the question worth answering is not whether you were wrong, it is whether your reasoning was wrong, and those are different. You cannot tell without the original rationale.
Scope by vintage from the start
The specific failure mode at a growing fund is that Fund II starts investing while Fund I is still active, and suddenly the record needs to distinguish which fund made a decision, held a position, and reported which numbers. Retrofitting that distinction onto a flat record is a migration project, and it arrives at exactly the moment the team has the least capacity for one.
Building with fund scoping in place before you need it is close to free, and skipping it is expensive later. This is the same argument as writing the thesis down before the first hot deal: the cost is an afternoon now and a project later.
The compounding part
Here is the thing that makes this worth doing rather than merely virtuous. A record of past decisions is not just an archive, it is a comparison set. Once you have scored eighty deals against the same framework with the same weights, a new deal can be positioned against your own history rather than against a general sense of what a good company looks like. "This scores like the three fintech deals we passed on in 2024, and two of them went on to raise strong follow-ons" is a genuinely useful sentence, and it is only available to a fund that kept comparable records.
That is the difference between a knowledge base, which is inert, and institutional memory, which does work. The first stores what happened. The second changes what you do next.
The two-person version
None of this requires headcount. A two-person fund can hold the whole practice in three habits: every deal that gets partner time gets a memo in the same structure, every decision gets a recorded vote with a sentence of reasoning, every pass gets a written reason before it moves to passed.
That is perhaps twenty minutes per deal, and it is the difference between a fund that has been investing for four years and a fund that knows what it learned in four years. The funds that skip it are not moving faster. They are accumulating experience without accumulating knowledge, and at the next raise the distinction becomes visible to everyone in the room.
For how this looks in practice at a fund building process on Fund I economics, how emerging managers build an institutional process covers the sequencing. The AI Chief Investment Officer page covers how the record gets used against new deals rather than just stored.