The Investment Memo, Standardised: What Belongs in a Pre-Seed IC Memo
An investment memo is a written argument for a decision, made before the decision. That sequence is the entire point. A memo written afterwards is a justification, and justifications teach you nothing.
For a pre-seed cheque, one page is enough. This is the structure, what each block owes the reader, and the two rules that make a memo worth writing at all.
Key takeaways
- Write it before you decide. A memo written after the fact is a rationalisation with a template.
- One page for pre-seed. Two for seed. The discipline of compression is most of the value.
- Every thesis must be falsifiable: state what would have to be true, so you can check later whether it was.
- Record the decision and the reasoning, including for passes. The passes are the more valuable archive.
- Standardise the structure across every memo, or you cannot compare across your own portfolio.
Why bother at a $50K cheque
Three reasons, only one of which is governance.
It forces the thesis to exist. Most investment reasoning, unwritten, is a feeling with supporting detail attached. Writing "this works if X" reveals, sometimes uncomfortably, that you do not know what X is.
It creates the only real feedback loop you have. Outcomes are slow, noisy and heavily influenced by luck. Comparing what you predicted against what happened is the one mechanism that teaches you anything about your own judgement. Without written predictions there is nothing to compare.
It makes the portfolio comparable. Thirty memos in one structure is a dataset. Thirty memos in thirty structures is thirty documents.
The nine blocks
1. The one-liner
What the company does, in a sentence, in your words rather than theirs. If you cannot write it without the founder's phrasing, you do not understand it yet.
2. The deal
Amount, instrument, valuation or cap, your cheque, resulting ownership, who else is in, who is leading.
3. The four-dimension read
Team, product, market, deal, each scored one to five with one line of reasoning. This is your screening score carried forward, so a memo is never written from scratch.
4. Why now
The timing argument. Why is this possible now and not three years ago, and why will it not be obvious to everyone in two years? "Why now" that reduces to "AI" is not a why-now.
5. The thesis, stated falsifiably
The core block. Not "great team in a big market" - that is a description. A thesis is a claim that could turn out to be wrong:
"We believe mid-market logistics operators will pay $2K/month for automated customs classification because the manual process currently costs them 1.5 FTE, and that this team can reach 50 such customers within 18 months through direct sales."
Every clause there is checkable. That is the test.
6. What has to be true
Three to five conditions. These are your review criteria for every subsequent update:
- They reach $40K MRR within 12 months
- Gross margin stays above 70% as they add customers
- The regulatory position holds in the EU
- They hire a sales lead by Q2
7. Risks, and which one would kill it
Every company has ten risks. Name the two that are actually fatal and say why you accept them. A memo listing ten risks with no ranking has done no thinking.
8. Diligence performed
What you actually verified, not what you read. Customer calls made, references taken, cap table reconciled, IP assignment confirmed. Distinguishing between "the founder told me" and "I checked" is most of what makes a memo trustworthy a year later.
9. The decision
Invest, pass, or hold, with the amount and the reason in one line. For a pass, the category and the reason. This is the block people skip and the block that matters most in aggregate.
The two rules
Rule 1: Written before, not after. A memo drafted after the wire is a summary. The value is entirely in the sequencing - having to state a falsifiable claim while you can still change your mind.
Rule 2: Same structure every time. Same nine blocks, same order, same scoring scale. This is what lets you ask questions across your portfolio: which dimension do my losses share, how often was my "what has to be true" wrong on timing rather than substance, which risk category do I systematically under-weight.
Variable structure destroys all of that for a small gain in per-memo elegance.
What AI can and cannot do here
AI drafts a good memo skeleton from a structured company record: the one-liner, the deal terms, the market summary, a first-pass four-dimension read, the obvious risks. On a standard field set, that is a genuinely reliable first draft and it removes the blank-page cost that stops most investors writing memos at all.
What it cannot produce is blocks 5, 6 and 9 - the thesis, the conditions and the decision. Those are the investor's judgement and the accountability for them cannot be delegated. An AI-written thesis reads plausibly and commits to nothing, which is the opposite of what a thesis is for.
RaiseTalks generates a validated memo from an approved data room - team, product, market and deal scored, transaction terms, market sizing, named competitors and risks, exportable as a PDF. Treat it as the skeleton. You write the thesis.
Memos for passes
The pass memo can be three lines: company, category of pass, one-sentence reason.
The reason this archive is more valuable than your investment memos: your investments are a biased sample of companies you liked. Your passes are the full distribution. If a pattern in your passes is systematically wrong - a sector you keep dismissing that keeps working, a founder profile you consistently under-rate - the pass archive is the only place that pattern is visible.
Almost nobody keeps one. It is the cheapest edge available in early-stage investing.
Frequently Asked Questions
One page for pre-seed, two for seed. Compression is a feature.
Before the decision. A memo written afterwards is a justification.
In practice, little at early stage. An IC memo is written for a committee and may carry more procedural detail. The reasoning blocks are the same.
Yes. Angels have the least structured feedback on their own judgement, so the written record is worth relatively more, not less.
It can write the descriptive blocks well from structured data. It cannot write the thesis or the decision, and should not be asked to.
Read next
The Modern Dealflow Stack: From Inbox Chaos to a Scored Pipeline
Pre-Screening at Scale: Scoring Startups Before the First Call
Proportionate Diligence: A Pre-Seed Checklist That Is Not a Series A Checklist
Angels and Micro-Funds: Running Real Diligence Without an Analyst TeamRaiseTalks turns an approved data room into a structured investment memo in one click, scored across four dimensions and exportable as a PDF. You bring the thesis. Start your 30-day trial.