What Goes in the Finance Section of a Series A Data Room

All right, here is the direct answer. A Series A data room needs nine finance files: monthly accrual profit and loss and balance sheet for twenty-four months, a tying bank reconciliation, a fully diluted cap table, the current 409A report, a written revenue recognition policy, the bottom up model, revenue by customer, aged receivables and payables, and your top contracts.

Now here is the part nobody tells you, and it is why the finance folder is the weakest room in most houses. Nobody is going to send you that list. The diligence request list you eventually get from the investor's counsel will be long, detailed, and almost entirely about legal records. Your financial statements will appear once, near the bottom, under a heading like Miscellaneous. That is not an oversight. It is a description of whose job this is.

So the finance section is the one part you design yourself, against a specification that exists but is written somewhere else. It is written in the agreement you have not signed yet. Every reporting obligation you will carry for the life of the investment sits in the investors' rights agreement, and it tells you what shape your finance function is expected to be in. Read it backwards and you have your folder structure.

So let me walk you through it in the order it matters. What the lawyers' list really asks for. The nine folders, written so you can lift them. Where the real specification is hiding. Why this is two rooms rather than one. The version control problem that costs founders most. A cautionary tale. And when to start.

What the lawyers' list actually asks you for, which is almost nothing

Start with Y Combinator's Series A diligence checklist, because it is the most widely read document on the subject and it is honest about its own scope. Jason Kwon, then general counsel of YC Continuity, assembled it after being involved in hundreds of financings. It runs seven sections and dozens of items. The section headed Business Plan and Financials contains exactly two: "Current business plan and any financial projections." and "Most recent financial statements."¹ That is the entire financial ask.

The published law firm lists say the same in more words. Cooley's sample venture diligence request list runs nine lettered sections, and has no financial section, no accounting section and no tax section at all.¹⁷¹⁸ Financial statements sit at item I.2 under Miscellaneous, asking for the most recent audited statements and the latest unaudited monthly statements. DLA Piper's published checklist has the identical architecture and, at the same item number, the same sentence.¹⁹ Two firms, independently published, both filing your books under Miscellaneous.

It would be easy to read that as the investor not caring, and the opposite is true. Legal and financial diligence are different workstreams on different clocks. Counsel confirms the company is what it says it is and that the shares can be issued cleanly. The investor's own team does the numbers, and they do not send you a list because they assume the numbers already exist in a form somebody can read. That assumption is the whole game.

Two things to clear out of the way, because both turn up in most articles on this topic. The widely repeated statistic that sixty-eight percent of failed deals cite disorganized documentation, credited to CB Insights, is not in the CB Insights research it is hung on. I read all of it. Their current work covers 431 venture backed shutdowns since 2023, and documentation quality is not among the reasons.³² Nor is the numbered folder convention attributed to YC, the one running 01_Financials, 02_Legal, 03_GTM, anywhere on the YC page. I read the whole thing in a browser this morning and the word folder never appears. Number your folders if you like the discipline, and I do, but number them because it fixes the sort order.

The nine folders, and why the order is the argument

Here is the list, written so you can lift it into your own structure. This is my own framing rather than a published standard, assembled from what actually gets asked for across the sources below. Cooley's own guidance on running a data room is about as prescriptive as anyone gets: "You can create folders for specific types of documents and then upload them each time a new document comes into being."³ The structure is yours. Choose one that answers questions in the order they get asked.

1. Monthly profit and loss and balance sheet, accrual basis, trailing twenty-four months. By month, in one workbook, not restated partway through. This gets opened first and everything else has to agree with it.
2. A bank reconciliation that ties, for every month in that period. Somebody will tie your revenue to your deposits in the first hour. Do it before they do.
3. The fully diluted capitalization table, agreeing to the signed instruments. Every safe, every note, every option grant, reconciled to the board consents that authorized them.
4. The current 409A valuation report, with its date visible on the cover. Its age is the first thing anyone checks, for a reason we will come to.
5. Your revenue recognition policy, written down, one page. When the obligation is satisfied, how multi element contracts are split, how usage overages and annual prepayments are handled.
6. The financial model, built bottom up, with the actuals tab still in it. The assumptions get tested, not the output. Leaving the actuals in is what proves the assumptions were ever true.
7. Revenue by customer and by cohort. Concentration and retention live here, and both get priced straight into the valuation.
8. Aged receivables and aged payables, as of the most recent close. Cheap to produce, and their absence reads as a company that does not run a close.
9. Your top customer and vendor contracts, redacted. Redacted deliberately, for reasons that get their own section below.

Notice how much of that is reconciliation rather than production. Eight of the nine already exist somewhere in your systems. The work is making them agree with each other, and that is exactly the work founders defer, because it is unglamorous and nobody is asking yet. Then somebody asks in the same week they ask for eleven other things, the reconciliation gets done at speed by whoever is free, and that is how errors enter a data room.

Where the real specification is hiding

If the lawyers will not give you a finance specification, the financing documents will. In Icosavax's investors' rights agreement, filed with the Securities and Exchange Commission, the annual covenant requires a balance sheet, income and cash flow statements, a comparison of actuals against both prior year and the board approved budget, and a capitalization table as of year end.¹⁰ The quarterly covenant adds a capitalization statement detailed enough "as to permit the Major Investors to calculate their respective percentage equity ownership". Read that slowly. Your cap table is a quarterly deliverable, forever.

The basis of accounting is specified too, and this is where cash basis stops being a preference. Samsara's agreement requires the annual package within ninety days, "all of which shall be unaudited and prepared in accordance with GAAP", with audited statements required only on board approval.¹¹ Note what that does and does not say. Most Series A companies are not signing up for an audit. They are signing up for generally accepted accounting principles, which is the more immediate commitment because it is retrospective. Convert after the round and you are restating periods the investor already read. There is a regulatory floor underneath it too, for any Regulation D offering that includes a purchaser who is not accredited.¹²

What GAAP buys you in practice is the deferred revenue line. Under Topic 606, cash received before you have delivered is presented as a contract liability, the obligation to transfer goods or services for consideration already received.²⁵ On cash basis that line does not exist, because as the AICPA's Center for Plain English Accounting puts it, advance customer receipts are recognized immediately instead of being deferred.²⁷ So an annual contract banked in January produces a January that never happened, and your best month and your riskiest month look identical on the statements.

Three shorter specifications outlast the round. A statement of cash flows is required of any enterprise reporting financial position and results of operations.³⁰ Delaware now defines what a stockholder may demand as a closed list including "Annual financial statements of the corporation for the 3 years preceding the date of the demand under subsection (b) of this section."¹⁶ And under Topic 718 equity awards are measured at grant date fair value.²⁶ That last one is your 409A doing structural work in the financial statements, not just the option paperwork.

This is two rooms, not one, and the door matters

Here is the mistake I see most often, and it is a security mistake dressed as a generosity mistake. A founder builds one room, puts everything in it, and sends the link to everyone who asks. Then a partner who passed in week two still holds a live link to the customer list, the pricing and the churn.

The timing of real diligence gives you the seam. Gunderson Dettmer's primer on the financing process says that while a firm may begin business diligence before proposing terms, "it usually will hold off on commencing full legal due diligence until a term sheet is signed."⁵ So: room one, open during the process, holding the monthly profit and loss and balance sheet, the model, revenue by cohort with customers anonymized, and the reconciliation summary. Room two, opened when a term sheet is signed, holding everything named.

Do not expect a signed confidentiality agreement to do this work for you. Cooley's Mike Lincoln is blunt: "professional investors nearly always refuse to sign NDAs, and in many cases are very open and public about their refusal to do so".² Morgan Lewis, writing for the other side of the table, carries a section headed simply No NDA Before Initial Meeting, and ties any willingness to sign to later stage companies and more specific information.²⁰ The protection you actually have at Series A is what you chose to put in the room, and when.

That matters most for the contracts in folder nine, because those hold other people's data as well as yours. California's privacy statute carves out transfers where "The business transfers to a third party the personal information of a consumer as an asset that is part of a merger", but the clause goes on to require that the third party assumes control of all or part of the business.¹⁴ A prospective investor reading your data room has assumed control of nothing, so the carve-out speaks to the closing, not to diligence. The European position arrives in the same place: an investor pursuing its own purposes is not your processor, so what matters is your lawful basis for a new purpose, with pseudonymisation named among the safeguards.³¹ In plain terms, redact names and individual pricing until a term sheet is signed.

The folder that quietly loses you the round

Version control. I know how that reads in an article about fundraising, and it is here anyway, because it does more damage than any single missing file. The failure mode is not that a document is absent. It is that two documents are present, they disagree, and the investor finds the disagreement before you can explain it.

No standard for this applies to a startup, so let me borrow one that is clear about the principle. The Public Company Accounting Oversight Board tells auditors that "Audit documentation must not be deleted or discarded after the documentation completion date, however, information may be added", and that anything added must show the date, the preparer and the reason. The securities rules for broker dealer records say the same in engineering terms, requiring a time stamped audit trail of every modification and deletion.⁶ Neither binds a company raising a Series A. Both describe the standard a reader unconsciously applies: I should be able to see what changed, when, and at whose hand.

Practically, three habits. One file per artifact, with the period in the name and no version suffix, because Morse is right that you should "only include fully-executed agreements with clear, descriptive file names".⁴ A single source of truth for every number, so the model pulls from the close rather than restating it. And a dated change log at the top of the folder, one line per update. When a number moves mid process, and one will, you want to be the person who already documented why.

The 409A report is the sharpest instance, with an edge that catches founders every year. The Treasury regulation makes a previously calculated value unreasonable where it was "calculated with respect to a date that is more than 12 months earlier than the date for which the valuation is being used", and also where it fails to reflect later information materially affecting value.¹³ A term sheet is exactly such information. YC's checklist flags the consequence in an aside, noting that more people than not believe a term sheet ends your ability to use the current 409A for grants. So clear promised equity grants before you start, not during.

The cautionary tale I would rather you hear from me

Early in my career I joined a specialty pharmacy as it was scaling, and my first real assignment was a three year lookback to restate books that had not been kept to generally accepted accounting principles. Three years, with a fine toothed comb, following the breadcrumbs back through transactions nobody had documented at the time. That company went from about fifty million to about five hundred million in revenue and sold to private equity, so it ended well. But not one hour of that cleanup created a dollar of enterprise value. It produced what should already have existed, late, on somebody else's clock.

A data room punishes this efficiently, because a restatement during a live process is the one event that makes an investor doubt everything else, and doubt is cheap for them and expensive for you. For scale, the PCAOB reports that "From 2005 to 2024, Big R financial restatements occurred at a rate of around 3% per year."⁷ Rare things attract attention when they happen to you. And the likeliest cause at your stage is revenue: in the first three years after the new revenue standard took effect, Audit Analytics found revenue recognition the most commonly cited restatement issue, three years running.²⁸

A founder led finance function is structurally fragile rather than merely busy, and this is not a competence question. COSO's guidance for smaller companies said it plainly two decades ago: in many smaller businesses one person may have complete control of all aspects of a process, and compensating controls are weaker because they happen after the transaction is complete.²⁹ The Commission's 2024 settled action against a microcap issuer illustrates it cleanly: the company "improperly recognized and reported approximately $102,000 in revenue for an order that never left the control of C-Bond", and the individual charged alongside it held both the chief executive and chief financial officer roles.¹⁵ Nobody was there to ask the second question.

When to start, and what to do first

Six to twelve months before the process starts. That is an operating judgement rather than a published benchmark, and it comes from arithmetic rather than caution. Converting to accrual and rebuilding two years of monthly statements is roughly a quarter of work. A revenue recognition policy has to survive a couple of closes before you can defend it. A cap table reconciliation depends on other people finding documents. None of that compresses.

The calendar gives you the room. Carta reports that "The median interval reached 616 days in Q2, or a little more than 20 months." between a seed round and a Series A, more than two months longer than two years earlier.⁸ That is twenty months in which almost every company builds product and almost none builds the finance function the round gets underwritten on. Meanwhile the market has got more selective: Carta's most recent quarterly read opens with "Venture is back, but it is not back for everyone.", with more than sixty percent of capital going to artificial intelligence companies.⁹ Median deal value has been rising across every series except seed.²⁴

The attention you are competing for is finite and measured. A survey of 885 institutional venture capitalists at 681 firms found the average deal takes 83 days to close, with the average firm spending 118 hours on diligence inside that window.²² A 2025 study that timed in person meetings across roughly twenty-one thousand deals found less diligence associated with hotter deals and busier investors, and more volatile outcomes as a result.²³ Those hours are already budgeted. The only question is whether yours get spent getting comfortable with your business or untangling your bookkeeping.

If you want one number to make the case, it is this. In the foundational study of how investors actually read fundraising materials, covering two hundred companies through their seed and Series A rounds, the financials page drew the longest per page attention of any section in the deck, and it appeared in only fifty-eight percent of the successful decks.²¹ The place they look hardest is the place founders leave thinnest. That gap is the whole opportunity, and it has not moved in a decade.

So if you do one thing this week, reconcile your last twelve months of revenue to your last twelve months of deposits and write down every difference. That one exercise surfaces almost everything else: the missing deferred revenue, the contract never billed, the month somebody reclassified without telling anyone. Ultimately you have to do what is best for the business, and you are the captain of the ship. But I have never once seen a founder regret starting this before the term sheet arrived rather than after.