Investment committee memo: first draft and fact-check

You are helping a deal team prepare the investment committee memo for [deal] for the [fund] committee meeting on [date]. A human reviewer checks everything you produce. You do not write the investment thesis, the ranking of risks or the recommendation, and you do not change a figure to make it agree. The committee decides; the deal team writes the case.

I attach:
- Model: [the deal model, with its file name and version]
- Diligence reports: [financial or quality-of-earnings, commercial, legal, tax, and any others]
- Management materials: [information memorandum, management presentation, business plan]
- Terms: [term sheet or offer letter, and the sources and uses]
- Template: [the firm's IC memo template, with its metric definitions]
- Draft, if one exists: [the current memo draft]

Work through the six steps below. Show the result of each step before you move on.

Two outputs. First, if no draft exists, draft only the descriptive sections (business overview, financial history, market, transaction structure) as a separate document in the template, give every figure and factual statement a claim ID in square brackets, and leave the thesis, risks and recommendation sections as headed placeholders for the deal team. Under the risks placeholder, list each diligence red flag as the report states it, with its report and page, for the deal team to address; do not assess or rank them. Second, the fact-check workbook below, which covers every claim ID in the draft, whoever wrote it.

STEP 1. MAP THE SOURCES
For each file, record: who produced it; the as-at date it states (not the date it was sent); the currency; what one row or section represents; the fields that identify a metric, a period and a section; the value fields and how each is labelled; and any subtotals, hidden rows, merged cells or footnotes that change how it should be read. Where the files differ in as-at date, period or currency, list each difference and say how you will handle it: convert at an FX rate from the Inputs tab with its source, or compare like periods only. Stop and ask me only if a difference means two figures cannot validly be compared.

STEP 2. TRACE THE PREPARATION
For each headline figure, say how it was built: the earnings measure (reported, adjusted or pro forma EBITDA, with the diligence adjustments listed), the period (last twelve months, financial year, run-rate), the net-debt definition, the entry multiple and the price it rests on, and each returns case (base, downside, upside) with the model cells that produce it. Note where the management materials and the diligence reports state the same metric differently, and which one the model uses.
Label each statement "stated in the file" or "my inference".

STEP 3. MATCH AND RECONCILE
3a. Keys. List the identifiers in each source: claim IDs and section numbers in the draft, sheet and cell references in the model, report section and page numbers. Say which claims carry a source reference already and how complete each is.
3b. Names. Where no shared reference exists, describe how the draft names each metric and how the model and the reports name it (for example "Adj. EBITDA", "EBITDA pre-IFRS 16", "normalised EBITDA").
3c. Standardise. Write numbered rules that map each metric name to the definition in the template, and apply them in new columns, keeping the original text.
3d. Map. Build a mapping table: claim ID, source file, sheet and cell or page, match method ("shared reference", "standardised name" or "suggested, needs review"), reason. A suggested match is never confirmed.

STEP 4. BUILD THE LINKED WORKBOOK
Build an .xlsx file with these tabs, in this order:
- Index: a link to each tab, one line on what it holds, the as-at date, and the steps to refresh the workbook for the next draft.
- Inputs: deal, committee date, base currency, FX rates with source and date, the model file name and version the memo must use, the metric definitions the template sets, one per standardised metric name, the list of scenarios or cases the model runs (for example base, downside, upside, plus "none" for a figure outside any case), the categories of claim ("figure", "factual statement", "forward-looking statement"), the rounding tolerance per unit shown (for example a figure stated in millions to one decimal place); plus the as-at date of each source file, with any differences between them and how each is handled; the tolerance for agreement with documents (0.00 wherever exact agreement is required). Each value carries its source beside it, and every formula that needs one points here.
- _Source tabs: each file exactly as received, one tab per file. Never edit these tabs.
- Map: the mapping table from step 3d. Each row carries its match method ("shared reference", "standardised name" or "suggested, needs review"), the reason, and the person who confirmed each suggested match and the date.
- Claims: one row per claim ID in the draft. The claim ID, section, the text as written, the category, the value and unit as stated, the period, the scenario or case from the list on the Inputs tab, the metric name as written, the standardised metric name from step 3c, a forward-looking flag (TRUE where the category is "forward-looking statement"), the source file and page from the Trace tab, and the author of the text ("AI draft" for text you write, "deal team" for text a person writes), which you set when the claim is created and never change afterwards.
- Trace: one row per claim ID. The scenario or case from the Claims tab, source file, the model version where the source is the model, sheet and cell or page, source value by formula from the _Source tab, source value rounded to the unit the memo shows, difference and status are formulas. Status reads "Agrees", "Check", "Superseded version" or "No source". Each row carries its source document and the page, cell or row reference for each figure.
- Metrics: one row per standardised metric, period and scenario or case, so a base-case and a downside IRR for the same period sit on separate rows. The definition from the Inputs tab, the count of claims stating it, and the lowest and highest value stated, by formula from the Claims tab. Status reads "One value" when the lowest equals the highest and "Check" otherwise.
- Totals: one row per total the draft states (sources and uses, the capital structure, a sum of segments), each with a total ID. The claim IDs of its parts, the sum of those parts by formula from the Claims tab, the total as stated, difference and status. For the sources and uses, each part carries its side ("source" or "use"): one row holds total sources and one holds total uses, each summed by formula from the parts on its own side, and a third row holds total sources less total uses by formula. Status reads "Agrees" within the rounding tolerance on the Inputs tab, otherwise "Check".
- RedFlags: one row per finding a diligence report rates as a red flag or high risk. Report, page and rating as stated, the claim ID where the draft covers it, and status by formula: "In draft", "On deal-team list" (in the list under the risks placeholder) or "Missing".
- Exceptions: a formula-driven list of every "Check", "Superseded version" and "No source" on the Trace, Metrics and Totals tabs, every unmapped claim, every RedFlags row reading "Missing", and every placeholder section still empty, with a type column ("Check", "Superseded version", "No source", "unmapped claim", "Missing red flag", "empty placeholder", "untraced or disagrees", "failed control", "query" or "timing difference"). It also lists every failed control and open query, each with its owner, status ("open" or "closed"), expected clearing date and the approver's sign-off (name and date).
- Checks: the controls in step 5.
- KPIs: the measures in step 6.
- _Log: the audit trail in step 6, with the names of the preparer and the approver.
Every derived number is a formula that leads back to a _Source tab or the Inputs tab. Nothing is typed in as a value. Use Excel tables and XLOOKUP or INDEX/MATCH. Wrap lookups in IFERROR and show "Not found", never zero. Dates as yyyy-mmm-dd, negatives in brackets, units in every value header, header rows frozen, no decorative formatting.

STEP 5. CONTROLS (EVERY ONE MUST PASS BEFORE SIGN-OFF)
On the Checks tab, write each control as a formula that returns TRUE or FALSE, with the count or difference beside it. A FALSE is not a score to improve: it blocks sign-off until it is resolved, by correcting the figure with a note or by getting the source corrected. An explanation alone does not clear it. The only item that may stay open at sign-off is a timing difference with an expected clearing date: it is listed by name on the Exceptions tab as a reconciling item and re-checked next period. A document tie-out failure is never a timing difference and can never be carried. Show every FALSE at the top of the Index tab under "Checks failing".
a. Complete: every figure and factual statement in the draft has a claim ID and a row on the Claims tab, and every claim ID has a row on the Trace tab. Target: 0 unexplained rows.
b. Totals tie: every row on the Totals tab shows a difference within the rounding tolerance on the Inputs tab, and the row for total sources less total uses on the Totals tab reads 0.00. Targets: 0 Totals rows outside the rounding tolerance, and a sources-less-uses difference of 0.00.
c. Formula-linked: every calculated cell is a formula (test with ISFORMULA), and there are no #N/A or #REF errors. Target: 100%.
d. Tied to documents: every claim on the Trace tab cites its file and its cell or page. Every figure must be traced (it cites its document and page) and must agree (it equals the figure in that document), within the tolerance on the Inputs tab and no wider. Any figure that is untraced or disagrees goes to the Exceptions tab at once. If the document itself looks wrong, for example a diligence report that disagrees with the model on the same period, raise it as a query; never change the workbook to make it agree. Target: 100%.
e. Matches confirmed: count the rows matched by shared reference, by standardised name and by suggestion. Target: 0 suggested matches left unconfirmed at sign-off.
f. FX and dates: every FX rate and every as-at date difference has its source on the Inputs tab. Target: 100%.
g. One model version: every claim sourced to the model comes from the model version on the Inputs tab. Target: 0 "Superseded version" rows.
h. One definition per metric: every standardised metric on the Metrics tab has one definition set on the Inputs tab, and every row reads "One value", so every claim of the same metric for the same period and the same scenario or case states the same figure. Target: 0 rows reading "Check".
i. Returns and multiples: every IRR, multiple of money, entry multiple and net debt to EBITDA claim reads "Agrees" on the Trace tab against the model output cell for that case. Target: 0 exceptions.
j. Red flags carried: every row on the RedFlags tab reads "In draft" or "On deal-team list". The check passes once each flag is in one or the other; deciding how the memo treats it stays with the deal team. Target: 0 rows reading "Missing".
k. Judgement left to the deal team: the thesis, risks and recommendation sections contain no Claims rows whose author is "AI draft", apart from the red-flag list under the risks placeholder, which quotes each finding and its page without assessing it; every forward-looking statement in the draft is marked as such on the Claims tab with its source.
l. Open exceptions: count the Exceptions rows that are not closed, leaving out only named timing differences with an expected clearing date. Target: 0.
m. Four-eyes: the _Log names the preparer and the approver, who must be different people, and the approver signs off every Exceptions line.

STEP 6. REUSABLE, AUDITABLE AND MEASURED
- Reusable: for the next draft or the next model turn, a person replaces the _Source tabs and updates the Inputs tab, and every other tab recalculates. Keep the Map tab so confirmed matches carry forward.
- Auditable: on the _Log tab, list each source file (name, producer, as-at date, date received, row count, control total), every rule and match you applied with the reason, every assumption, and who prepared and approved. Never overwrite source data. A correction goes in a separate column with a note.
- Measured: add a KPIs tab that the team fills in for each memo: back-test recall (run on a past memo with known errors; share of known differences found; target 100%, and do not adopt the workbook below that); claims agreed first time with no manual fix; claims flagged that turned out fine; corrections made by the reviewer before sign-off; corrections found after sign-off (target 0); manual edits outside the Inputs and _Source tabs at refresh (target 0); and cycle time in hours from receiving the files to sign-off, against the manual baseline.

If your tool cannot create an .xlsx file, give me each tab as a table plus the exact formula for the first row of every calculated column.

At the end, give me the number of claims checked, the number that agree to source, the number with no source and the number on Exceptions, the result of every check, and a list of anything you were unsure about.