Every closing has two speeds. There's the deal speed (term sheet, negotiation, handshake) and then there's diligence speed: the weeks in which the investor's lawyers read everything you've ever signed and check it against everything you've ever claimed. Founders control the second speed far more than they think. A company whose records reconcile closes in the time the wire takes; a company with three unexplained discrepancies closes a month later, if at all.
This is the startup due diligence checklist from the founder's side of the table: what the other side is actually looking for, the 12 documents that appear on essentially every request list, how to organize them so counsel finds things without asking, and the specific failures that stall closings. It assumes you've read the closing checklist. This lesson is about being ready before the term sheet exists.
What investors actually check
Diligence requests look long, but investor counsel is really running five verifications. Understand these and the document list stops feeling arbitrary.
- The incorporation chain. Does the company legally exist, in the form claimed, with the share capital claimed? Certificate of incorporation, constitutional documents, every amendment, and, outside the US, the commercial license and statutory registers. A missing charter amendment means the share class you sold someone may never have been validly authorized.
- Cap table vs. documents. The single biggest exercise: counsel rebuilds your cap table from the signed paper and checks it against the one you presented. Every share on the table needs an executed document behind it; every executed document needs to appear on the table. Both directions fail in practice.
- IP assignments. Does the company actually own what it's selling? Every founder, employee, and contractor who touched the product needs a signed assignment, including work done before incorporation, which founders almost always forget to assign in.
- Grant approvals. Options aren't valid because you promised them; they're valid because the board approved them. Counsel matches your grant ledger against board consents. If the ledger shows 14 grants and the consents cover 11, the 3 unapproved grants (say 145,000 options) need ratifying before close. (Background: granting options the right way.)
- Signatures. The dumbest and most common failure. A document labeled "execution version" with blank signature blocks is not a contract. Counsel checks every signature page, on every document, and one missing countersignature can hold a closing hostage.
The 12-item startup due diligence checklist
Request lists vary in length, but the core is remarkably stable across jurisdictions and stages. If these 12 items exist, are signed, and are current, you've covered the questions that actually block closings:
- Certificate of incorporation: plus every amendment since, in order.
- Constitutional documents: articles or memorandum of association, or bylaws, as your jurisdiction has them.
- Commercial license: where applicable (GCC, much of Asia), and current, not the expired one.
- Statutory registers: register of members, directors, and beneficial owners, up to date.
- Board and shareholder resolutions: every issuance, plan adoption, officer appointment, and charter change should trace to one.
- Cap table: a current, dated snapshot that reconciles to the documents below (see how to read one).
- Investment instruments: every SAFE, convertible note, and share purchase or subscription agreement, fully executed, with side letters.
- Equity incentive plan: the plan document, the current authorized pool size, and every grant agreement under it.
- Founder and employment agreements: founder agreements with vesting terms, plus employment or offer agreements for key people.
- IP assignments: from founders (including pre-incorporation work), employees, and every contractor who wrote code or created assets.
- Founder KYC: passports or national IDs for founders and significant holders; funds can't clear compliance without them.
- Financial statements: management accounts at minimum; audited statements if you have them.
Notice what the list is: not a pitch, but proof. Every item either establishes that the company exists as claimed, that ownership is what the cap table says, or that the assets are actually the company's.
Reconciling the cap table against the documents
Do the investor's biggest check yourself, before they do. Sum the shares in your executed documents and compare the total to your cap table, to the share, in both directions.
Worked example: the 500,000-share gap
The cap table shows 11,150,000 shares issued. The signed documents say:
Founder A stock purchase: 4,500,000
Founder B stock purchase: 4,500,000
Advisor restricted stock: 250,000
Seed SAFE conversions (closing set): 1,400,000
Documented total: 10,650,000
Gap: 11,150,000 − 10,650,000 = 500,000 shares on the cap table with no signed document behind them. Digging in: an angel invested in 2022, the 500,000 shares were recorded on the table, the subscription agreement was drafted and sent, and never countersigned by the company. Legally, those 500,000 shares are in limbo: money received, issuance never properly papered. The fix (locating the draft, executing it, and a board resolution ratifying the issuance) took nine days. Nine days added to the closing because a five-minute signature was skipped three years earlier.
Run the same reconciliation on options (ledger vs. board consents vs. plan capacity) and on SAFEs (your instrument inventory vs. the signed PDFs). Every mismatch you find now is a mismatch investor counsel doesn't find later.
The trust multiplier: investors don't price discrepancies one at a time. The first unexplained gap counsel finds converts diligence from verification into forensics: every subsequent document gets read with suspicion, and requests multiply. One clean reconciliation buys more speed than any polished deck; one 500,000-share hole costs more than the hole itself.
Organizing the data room by section
A data room's job is to let a stranger find any document in under a minute without emailing you. Structure beats volume: eight sections, mirroring how counsel divides the work, cover everything.
- Corporate structure: incorporation, constitutional documents, licenses, registers.
- Governance: board and shareholder resolutions, consents, minutes.
- Equity: cap table snapshots, SAFEs, notes, purchase agreements, the option plan and grants.
- Team: founder, employment, and consultancy agreements.
- Intellectual property: assignments and IP-related agreements.
- Compliance & KYC: identity documents, declarations, regulatory filings.
- Financials: statements, management accounts, audits.
- Material agreements: key customer, supplier, and partnership contracts.
Name files so they sort themselves: date first (2024-03-15), then type, then party. "2024-03-15 SAFE - Alrai Ventures - 250k.pdf" answers three questions before it's opened. And maintain the room continuously, not the week a term sheet lands: file each document the day it's signed. This is the one place tooling helps as much as discipline: Vquity's Data Room ships this structure built in, with around 120 document types across eight sections, a 12-item readiness score tracking the checklist above, and a one-click zip when an investor asks for everything.
Version hygiene
The second way clean documents create dirty diligence: the room contains a real document, just not the current one. Amendments, renewals, and restatements supersede earlier versions, and a data room that presents a superseded version as current makes true numbers look false.
Worked example: the stale option plan
The data room contains the equity incentive plan, version 1, authorizing a pool of 1,000,000 shares. At the seed close, the board approved an amendment raising it to 1,500,000, but only v1 was ever uploaded. The grant register shows 1,320,000 options granted.
Investor counsel checks the register against the plan in the room: 1,320,000 granted against 1,000,000 authorized = 320,000 options that appear to exceed plan capacity, a genuine red flag, because unauthorized grants can be void. Cue a formal query, a request for all board minutes since incorporation, and a week of back-and-forth before the amendment surfaces and 1,320,000 ÷ 1,500,000 = 88% utilization turns out to be fine. Nothing was wrong except the version.
Three habits prevent this. Chain versions: when a document is amended or renewed, file the new version with the old one, marked as superseding it. Counsel wants the history, not just the endpoint. Never let a superseded version sit unmarked where it reads as current. And when any number lives in multiple documents (pool size in the plan, the board consent, and the cap table), check they agree every time one changes.
The failures that delay closings
Across diligence processes, the same handful of problems cause most of the delay. All are cheap to fix in month 3 and expensive in the closing week:
- Missing signature pages. Execution copies without signatures, agreements signed by one party, missing exhibit initials. Audit every signature block in the room before sharing it.
- Pre-incorporation IP. The founder wrote the core codebase two months before the company existed and never assigned it in. A confirmatory IP assignment fixes it, but only if signed before counsel asks.
- Departed contributors. The contractor who built v1 and left in year one, with no assignment on file. Chasing a signature from someone with no ongoing relationship, and possibly leverage, is the single slowest fix on this list.
- Unapproved grants. Options in the ledger with no board consent behind them. Ratification works, but it's a board action on the critical path of your closing.
- Cap table drift. The table and the documents disagree. See the reconciliation above, and the mistakes that surface in diligence.
- Lapsed formalities. Expired commercial licenses, unfiled annual returns, registers that stop in 2023. Individually trivial; collectively they read as a company that doesn't keep its records.
- The KYC scramble. Compliance can't clear funds without founder identity documents, and passports have a habit of being in another country the week of the wire. File them early.
The pattern behind every item: diligence never creates problems, it only reveals them later, with a deal on the line, at maximum cost. The 12-item checklist plus a reconciled cap table plus honest versioning is the whole game. Build the room as you go, and the next request list is a filing exercise, not an excavation.
One thing this lesson deliberately ignored: which documents those 12 items are depends on where you incorporated: articles vs. bylaws, 409A vs. FMV assessment, license or no license. That's the next lesson.