investment_memo_writing v1.0.0

---

name: investment_memo_writing description: Write investment memos that support a real committee decision: a falsifiable thesis, bottom-up market sizing, team assessment, scenario returns with shown arithmetic, risks stated plainly with residual exposure, deal terms, and a recommendation that follows from the evidence. Use when drafting, reviewing, or fixing a VC, PE, growth, or public-market investment memo, an IC memo, a deal summary, or a follow-on memo.

---

Investment Memo Writing

Purpose

An investment memo is a decision record, not a pitch. Its job is to let a small group of experienced readers vote yes or no on the same evidence, and to let the firm learn later whether the reasoning was right. A memo that only argues one side has failed, even if the writing is good.

This skill covers the full memo for venture, growth, and buyout deals, plus shorter forms: a 1 to 2 page deal summary, a one-page early gate memo, and a follow-on memo for an existing position. The standards below apply to all forms. Short forms cut sections, they do not cut honesty about evidence.

Workflow

1. Fix the decision before drafting

Record at the top, before any analysis:

• Company or asset, round or transaction, date, deal lead, memo author
• The exact decision asked for: amount, instrument, valuation, and the vote or gate (pass, track, diligence, term sheet, invest, conditional invest)
• The evidence cutoff date. Anything learned after it goes in an addendum, not silently into the body
• The firm criteria the deal must meet, if any: stage, sector, ownership target, reserve plan, return bar

If the decision is not clear, stop and get it clear. A memo without a decision drifts into a research report.

2. Inventory evidence and tag it

List every material fact you plan to use and tag each one:

• [CONFIRMED source, date] - verified in a primary document, data room file, filing, signed contract, or direct test
• [MGMT source, date] - supplied by management or the seller, not independently checked
• [ASSUMPTION basis] - your estimate, with the basis stated in one clause
• [UNKNOWN] - needed, not yet known. It stays visible as an open question

Build the evidence matrix as you go (see references/evidence-and-risks.md). Every number in the memo must appear in the matrix with its tag. A number with no source becomes [UNKNOWN] and moves to open questions. Never carry a management forecast as if it were a fact.

Watch for confirmation diligence: if the team has already decided, say so in the draft header and give the bear case extra weight in review. Diligence that cannot kill a deal is paperwork.

3. Write the thesis as falsifiable pillars

Write exactly three pillars unless the deal genuinely needs four. Each pillar has four parts:

1. Claim, stated as a sentence that can be wrong
2. Evidence for it, with tags
3. Observable metric and expected path (for example: net revenue retention stays above 110 percent through the next four quarters)
4. Failure condition (for example: retention falls below 95 percent in any two consecutive quarters)

Add the variant view in one line: what you believe that the price or the consensus does not. If there is no variant view, the edge claim in step 8 will be weak, and the memo should admit that.

Add kill criteria: 3 to 6 concrete conditions that, if found true, flip the recommendation to pass. Write them before you finish the analysis, not after.

4. Size the market bottom-up

Do this in references/market-and-returns.md order:

1. Define the customer unit precisely: buyer, segment, geography, budget line
2. Bottom-up size: customers times revenue per customer, each input sourced and dated. Show TAM (all such customers), SAM (those reachable with the current product and channel), and SOM (share plausibly won by a stated year)
3. Put any top-down figure from the deck beside yours and explain the gap in one paragraph. Never present a percentage of a large market as a size
4. Back-solve the required market: required exit value, required revenue at exit, and required market. If the required share is implausible, the thesis has a math problem no narrative fixes

5. Assess the team on evidence, not adjectives

For each key person: relevant past outcome, domain time, and what references actually said. Use independent references where possible: former colleagues, churned customers, lost prospects, not only names management supplied. Name the gaps plainly: the hire not yet made, the function with no owner, the key-person dependency. State what would have to be true for the team to handle the plan's hardest step.

6. Build financials and scenario returns with shown arithmetic

• Historicals: at least the periods available, with any adjustments listed one by one and justified. Management numbers and adjusted numbers sit in separate columns
• Projections: list each assumption on its own line with its basis. A growth rate with no basis is a wish
• Scenarios: base, upside, downside, each with entry terms, ownership at entry and at exit after expected dilution, exit value, MOIC, and IRR. Show the arithmetic line by line (formulas in references/market-and-returns.md, calculator in bin/returns_calc.py)
• Sensitivity: vary the two assumptions the outcome depends on most. State which single assumption, if wrong, breaks the return

Base case must be reachable without heroic assumptions. Never present the upside case as the base case.

7. State risks plainly, with residual exposure

Use the risk table in references/evidence-and-risks.md. For each of the top 4 to 6 risks: what happens, probability, impact, evidence it is real, the mitigant, and the residual exposure after the mitigant. A mitigant that is only a hope ("team will execute") does not count. Flag any deal-breaker risk that must clear before closing, and name the test that clears it.

Run a short pre-mortem: "It is 24 months later and this investment returned near zero. The three most likely reasons are..." Write the reasons from the actual risk table, then check whether the memo already prices or tests each one.

8. Terms and source of edge

Terms table: round or purchase size, pre and post money valuation or entry multiple, instrument, your cheque, ownership, pro rata and board rights, key protective terms, use of funds, reserve plan. Then one short section: why does this opportunity exist for you, and what is your right to win it (access, knowledge, speed, portfolio help). If the honest answer is "it was shopped widely and we are the highest bidder," write that.

9. Recommend, with conditions that can be checked

Pick one: Invest, Pass, or Conditional invest. For venture early gates: Term sheet, Diligence, Track, or Pass.

• The recommendation must follow from the pillars, the scenario math, and the kill criteria. If a kill criterion is already met, the recommendation is Pass
• Conditions are exact: the test, the owner, the date ("invest subject to two independent customer references confirming renewal intent, led by [name], before signing") not "subject to further diligence"
• For Track: name the trigger that reopens the deal and the date you will check it
• List next steps and the timeline to close

10. Review as a hostile reader

Run bin/memo_check.py on the draft, then read once asking only: where would a skeptical partner attack, and is the counterevidence already on the page? Check that open questions appear in the summary, the risk section, and the recommendation, not buried in an appendix. Check every number against the evidence matrix. Check that units and periods are consistent throughout.

Output Contract

Deliver a memo with these sections in order. Length guide: full memo 4 to 8 pages, deal summary 1 to 2 pages, one-page gate memo fits one page.

1. Header and decision box (company, round, date, lead, author, decision asked, recommendation in one line)
2. Executive summary (half a page: what it is, thesis in three bullets, terms in one line, top risk, recommendation)
3. Investment thesis (pillars with failure conditions, variant view, kill criteria)
4. Company and market (what it does, how it makes money, bottom-up TAM SAM SOM, required market, timing)
5. Team (evidence per person, references, gaps)
6. Financials and returns (historicals, assumptions table, scenario table with MOIC and IRR, sensitivity note)
7. Competition and edge (3 to 6 rivals compared on the dimensions buyers care about, moat claim and its test, source of edge)
8. Risks and pre-mortem (risk table with residual exposure, pre-mortem paragraph)
9. Terms and recommendation (terms table, recommendation, exact conditions, next steps)

Appendices, not counted in the page guide: evidence matrix, open questions list, reference call notes.

Every memo also states, in the summary or a footnote: this memo is an internal decision document, not investment advice to any outside party.

Operating Rules

1. Every factual claim is verified or tagged. Unverifiable claims are cut or tagged as assumption or unknown. Never invent numbers, dates, quotes, or events.
2. Write findings as assertions backed by evidence. When evidence is weak, say so in plain words ("No pricing test has been run. Willingness to pay is unproven.") instead of hedging every sentence.
3. Keep management-sourced material visibly separate from independently confirmed material, in the text and in the matrix.
4. Show arithmetic for every return figure. A MOIC or IRR with no visible inputs is not usable.
5. The bear case gets real space in every memo, written by the deal lead, not delegated to a token paragraph.
6. Do not let a later draft quietly rewrite the original thesis. Follow-on memos append a dated update and score the original pillars against what happened.
7. Keep units consistent (state the currency and whether figures are in thousands or millions) and periods labeled (fiscal year, trailing twelve months, quarter).
8. Confidential deal material stays in the memo's intended audience. Widening circulation is a separate decision with its own permission.
9. If evidence does not support a recommendation either way, recommend Diligence or Track and name the exact evidence that would settle it. Do not force a yes or no the record cannot carry.
10. No em dashes in the memo. Use commas, periods, or a new sentence.

Files

• references/memo-templates.md - fill-in templates for the full memo, the one-page gate memo, and the follow-on memo
• references/market-and-returns.md - bottom-up sizing steps, required-market back-solve, MOIC and IRR formulas with a worked calculation
• references/evidence-and-risks.md - evidence tags, evidence matrix format, risk table format, pre-mortem method, review checklist
• references/worked-memos.md - two complete worked memos on fictional companies, one venture seed and one growth buyout, with all tables filled
• bin/returns_calc.py - MOIC and IRR calculator for scenario tables (single cheque, dated cash flows)
• bin/memo_check.py - structural checker: required sections, evidence tags, kill criteria, and return figures present or flagged