Skip to main content
Blog

Decision-making

What an investment committee memo should hold

The thesis, what it stands on, the issues and how each one closed, what's still open, the returns on your own cheque, and a judgment a person signs. The ten sections we use, and why two of them are never written by Aventro.

Dor Tagger

Co-founder & CTO, Aventro

November 20, 2026

4 min read

Short answer: an IC memo should let a partner who hasn't seen the deal decide whether to back it, without surprises in the meeting. That means the thesis, the evidence it stands on, the issues you found and how each one closed, the fit with your mandate, what's still open, and the economics on your own cheque. Then the judgment and the recommendation, written by a person. Anything the memo doesn't know should look missing.

Most memos I've read are long and still skip one of those. Usually the open questions. The memo reads like the deal is done, and the partner finds out in the meeting that the cap table never arrived.

The ten sections we use

These are the ten sections in Aventro's standard memo. A fund can use its own template instead: reorder them, drop most of them, add its own. Two are required in every template.

  1. Thesis and company summary. What the company does and why it could matter to the fund, in a paragraph.
  2. Score, maturity and evidence coverage. The number, and how much of it was checked.
  3. Critical diligence issues and resolutions. What you found and how it closed. "Deck revenue didn't match the bank export. The gap is a signed contract that starts in March. We have the contract."
  4. Fund fit. Where it sits against your mandate.
  5. Mandate status and exceptions. If you're going outside the mandate, say so here, with the reason.
  6. Key evidence and verification highlights. The few documents and checks the case rests on.
  7. Unresolved questions and open requests. What's still out, who owes it, by when.
  8. Economics and returns. On your cheque and your fund size, including whether it can return the fund.
  9. Human judgment.
  10. Recommendation and decision considerations.

The last two belong to a person

Sections 9 and 10 are where someone puts their name to the deal.

In Aventro the first eight are assembled from what the deal already holds: the facts, the score, the checks, the fit, the open requests. No model writes or rephrases them. The last two are never assembled at all. The database refuses machine text there, and a memo can't be finalized until a person has written both.

The same rule covers your own committee forms. When Aventro fills a fund's IC template, any box someone signs is left for them: a recommendation, a decision, "signed off by". A wrong answer in a box someone signs is worse than an empty one, because the signature vouches for it.

Missing should look missing

A memo with every field filled should make you suspicious. If the company never stated its burn, the memo should say so in plain sight.

Aventro composes the memo from what the record holds and nothing else. A fact it doesn't hold shows as missing.

Compute the numbers

Ownership, returns, the exit it takes to return the fund. That's arithmetic, and nobody should eyeball it, person or model.

Aventro's returns model is calculated in code and priced with your fund's own size and cheque. If the fund never set a return bar, the memo doesn't grade the deal against one. Grading a company against a 3x nobody chose would be Aventro inventing your strategy.

Positions before discussion

The first partner to speak sets the anchor. If everyone reads everyone else's view first, you lose the disagreement the committee is there to surface.

In Aventro each person records a position before seeing anyone else's: leaning yes, leaning no, undecided or abstaining, with their strongest reason and what worries them most. You see the others once you've given your own. Everyone can see the count, "2 of 4 have answered", but never who answered or which way.

After the vote, keep the reasons

The memo is the input. The decision record is what the fund keeps.

Write down the decision and its rationale, the version of the mandate it was judged against, the terms, and where each person stood. If you approve past an issue the team marked as blocking, write why. If it's a rejection, write what would bring the company back: "when ARR passes $1M", "after they raise a seed".

Aventro won't record a decision without a rationale, and asks for a reason when an approval goes past a blocking issue. It freezes the fit, the Fund DNA version, the terms and every position with the decision. A rejection's condition gets read back later: when it names a figure or an event, Aventro checks it against what the fund now holds for the company.

I think that last part is the one most funds lose. The memo, the returns and the decision record are on the decision page, and the whole workflow is on AI due diligence for venture capital. To see it on your own pipeline, book a demo.