River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Startup Cap Table Template

Every row cites the document the holder signed and the board consent that authorised it, because a row with no consent is not stock.

Free download  ·  No account needed

Instrument Register

[Company], Inc. — reconciled as at [date]

One row per instrument, not per holder. Built from the documents before any cap table is opened.

IDHolderInstrumentClassShares per documentInstrument refAuthorising consentShares per consentAuthorisationConsideration received
IN-01
IN-02
IN-03
IN-04

Two source references, and they are different questions

The instrumentWhat the holder signed. A purchase agreement, a grant notice, a warrant, a convertible
The authorisationWhat the company did. A board consent setting the number, the recipient and the consideration
When they disagreeThe authorisation decides, and the register records both numbers rather than picking one

Filling in the second column from the first is the mistake this register exists to prevent. If no consent has been located the field stays empty and the row is flagged. It is never inferred from a signed agreement.

A cap table is not a spreadsheet. It is a claim about what a folder of signed documents says, and the spreadsheet is a cache of that claim which nobody has ever invalidated. So every cleanup that starts from the most recent version inherits whatever was already wrong with it. This pack reconciles the other way round: the register is built from the paper first, and every row carries both the instrument the holder signed and the board consent that authorised it.

Rillmore Labs is an invented Delaware seed company with two founders and eleven holders. Four spreadsheet versions were in circulation and they gave the same founder four different answers: 44.444 percent, 43.716, 40.404 and 40.000. Reconciled to the paper it is 41.026 percent on valid stock, which is 103 basis points above the current file, or about $287,000 of that founder's own position at a $28,000,000 post-money round. The most recent version was the furthest from the documents.

One row was the whole finding. An advisor held 150,000 shares under a signed purchase agreement, with the consideration paid and a certificate issued, and no board consent anywhere. Delaware fixes a stock issuance in a resolution of the board, so until somebody ratifies it the company has two share counts. No spreadsheet can show that. Separately a $450,000 SAFE worth 5.00 percent sat in the folder and in no version, which the conversion model prices.

Every sheet in the pack, worked through on one company

The register built from the paper, four versions bridged to it, the findings by kind, and eight denominators.

Instrument Register

Rillmore Labs, Inc. Built from 3 purchase agreements, 13 grant notices, the plan and one signed amendment, 3 convertibles, 11 board consents and the charter. No spreadsheet was open while it was written.

IDHolderInstrumentShares or amountAuthorising consentAuthorisationNote
IN-01Ines OkonkwoRestricted stock purchase agreement4,000,0002022-03-14Authorised83(b) filed on day 7 of 30
IN-02Teodor VanceRestricted stock purchase agreement4,000,0002022-03-14Authorised83(b) filed on day 41 of 30
IN-03Halvard ReyesRestricted stock purchase agreement150,000None locatedPutativeSigned, paid, certificate issued, never authorised
IN-04Nkem AdeyemiOption exercise, paid 2024-06-1160,0002023-02-08AuthorisedStill sitting in the option line in every version
IN-062022 planReserve, as last signed1,750,0002023-11-02AuthorisedCumulative across the plan and amendment one
IN-072022 planFurther reserve increase100,000None locatedNot authorisedUnsigned draft. The spreadsheet carries it as 1,850,000
IN-08Nine holders13 option grant notices1,318,5007 consentsAuthorised5 notices, 18,500 shares, never entered anywhere
IN-09Three leaversForfeitures under plan terms(90,000)Not requiredAuthorisedStill shown as outstanding in every version
IN-12Marisol FerrandSAFE, $9,000,000 post-money cap$450,0002024-02-09AuthorisedIn no spreadsheet. Converts to 5.000%
IN-13The companyCharter as amended10,000,0002023-11-02AuthorisedIssued and reserved 9,900,000, so headroom is 100,000
TOTALSCommon 8,210,000 claimed, 150,000 of it putative1,168,500options outstandingFully diluted 9,900,000 claimed, 9,750,000 valid

The option arithmetic has to close: 1,318,500 granted less 60,000 exercised less 90,000 forfeited is 1,168,500 outstanding, and adding the 521,500 reserve remaining and the 60,000 exercised returns the 1,750,000 authorised reserve exactly. When it does not close, a grant, an exercise or a forfeiture is missing, and finding which one is more useful than plugging the difference.

Version Reconciliation

Four versions were in circulation. Each is bridged to the register rather than to the version after it, and each is kept, because which number went to which investor is itself a fact.

VersionFilename as foundEditedWent toFully dilutedFounder oneWhy it is wrong
V1captable.xlsx2022-03-18Nobody9,000,00044.444%Correct on the day. Everything since happened after it
V2cap table (shared with Ashgrove).xlsx2023-05-02An investor, in diligence9,150,00043.716%Where the 150,000 unauthorised shares entered and were never questioned again
V3Cap_Table_v6.xlsx2024-11-19A lead that did not close9,900,00040.404%Right total by accident. Two errors offset each other exactly
V4cap table final_v7 USE THIS.xlsx2025-08-07Nobody10,000,00040.000%The newest version and the furthest from the paper
P1Reconciled, counting the putative shares9,900,00040.404%Defensible. One of two answers a board has not chosen between
P2Reconciled, valid stock only9,750,00041.026%The published figure, with its convention named beside it

The bridge from V4 to the paper, in three lines. Common moves up 60,000, because a paid option exercise was never taken out of the option line. Options move down 71,500, being 90,000 forfeited never removed, 60,000 exercised never moved, and 18,500 granted never entered. The reserve moves down 100,000 to the last signed amendment.

Every one of these versions was made in good faith by opening the previous one and editing the rows somebody remembered. That is the failure mode, and it is why the register is built from the documents before any spreadsheet is opened.

Discrepancy Register

Organised by kind rather than by holder, because the four kinds need four different answers. Findings are never netted against each other.

IDKindLineShare deltaWho decidesWhat it changes
DR-01Row with no authorisationAdvisor common, 150,0000Board, with counselMoves no shares and decides the share count. 40.404% or 41.026%
DR-05Row with no authorisationPlan reserve-100,000BoardThe whole gap between 40.000% and 40.404%
DR-02Document with no rowExercised common+60,000Nobody. Data entryFully diluted unchanged, composition wrong
DR-03Document with no rowOption grants+18,500Nobody. Data entryUnderstates dilution, which is the direction counsel finds
DR-04Number disagreesOption grants-90,000Nobody. Data entryOverstates overhang. Never netted against DR-03
DR-06Document with no rowSAFE, $450,0005.000%Nobody. It is signedCosts the founder 2.051 points on conversion
DR-07Document defect83(b) election0Holder, with a tax adviserFiled on day 41 against a 30 day deadline
DR-08Document defectEquity incentive plan0Board, with counselNo recorded stockholder approval of the plan or either increase
DR-09Number disagreesAuthorised common100,000Board and stockholdersHeadroom is 100,000 shares, not zero. Changes sequencing
9 findings2 unauthorised rows, 3 documents with no row, 2 numbers disagreeing, 2 document defects4 are data entry. 5 need somebody to decide something

Read the last row twice. The largest share movements here are data entry, and the items that need a decision move almost no shares at all. That is exactly how a cap table stays comfortable for three years and then falls over in a fortnight of diligence.

Fully Diluted View

Eight defensible denominators for the same holder on the same day. The pack publishes one and labels it, and keeps the rest so a different figure can be traced to a convention rather than argued about.

ConventionWhat is in the denominatorDenominatorFounder onevs narrowestWhen it is right
FD-AValid common only8,060,00049.628%A stockholder vote
FD-BCommon including the putative shares8,210,00048.721%-0.907Nothing, until the board decides one row
FD-CValid common plus options outstanding9,228,50043.344%-6.284Treasury method work
FD-DPlus the whole remaining pool9,750,00041.026%-8.602The published figure. Term sheet default
FD-EFD-D counting the putative shares9,900,00040.404%-9.224What version three said, by accident
FD-FFD-D with the unsigned reserve increase9,850,00040.609%-9.019Never. A draft dilutes on paper and nowhere else
FD-GFD-D with all three SAFEs converting12,810,21931.225%-18.403Before signing a term sheet
FD-HFD-D with only the two known SAFEs12,020,54833.276%-16.351Never. Shown to price the missing one

An 18.403 point spread, and every figure in it is correct. Seven of the eight are wrong for the question being asked, which is why a percentage published without its convention is not an answer. The three post-money caps take 10.000, 8.889 and 5.000 percent, so 23.889 together, and the pre-money base of 9,750,000 shares grosses up to 12,810,219 on conversion.

FD-G and FD-H differ only by the SAFE that was in the folder and in no spreadsheet. The 2.051 point gap between them is what that omission was costing the founder, and it is the number that explains why this work belongs three weeks before a round rather than during one.

What's in the pack

01

Instrument Register

One row per instrument built from the documents, each citing what the holder signed and the consent that authorised it.

02

Version Reconciliation

Every spreadsheet version kept and bridged to the reconciled position, with what it said and who received it.

03

Discrepancy Register

Findings by kind rather than by holder, with the share delta, the document that settles it, and who decides.

04

Consolidated Cap Table

Derived from the register rather than typed, with any unauthorised row shown below the outstanding common subtotal.

05

Fully Diluted View

Eight denominators computed side by side, so the published percentage carries the convention that produced it.

06

How a Share Becomes Valid

The four-minute method behind the rest: two source references, four kinds of discrepancy, and why order matters.

07

Discrepancy Notes

One note per finding, ordered by what it costs to resolve rather than by how many shares it moves.

08

Reconciliation Note

The two-page document that goes to an investor or to counsel: found, fixed, open, and out of scope.

How to use it

  1. 1

    Open in River, or download it

    Take the blank Word and CSV files away with no account, or install the pack in River and have the agent build the register from your own documents.

  2. 2

    Send the documents before the spreadsheets

    Charter and amendments, purchase agreements, the plan and every grant notice, board and stockholder consents, convertibles, exercise and termination records.

  3. 3

    Check each issuance against its consent

    Authorised, not located, or contradicted. A missing consent is never inferred from a signed agreement, and the flagged rows are the finding.

  4. 4

    Then reconcile the versions to the register

    Every version, including the ones that went to investors, bridged to the reconciled position rather than to the version after it.

Frequently asked questions

Is this template free?

Yes. Five sheets and three documents download as Word and CSV files with no signup and no credit card. "Edit with AI" is the optional path where the agent reads your own instruments and consents and builds the register from them. Other packs sit in the template library.

What format are the downloaded files?

Word documents (.docx) for the method, notes and reconciliation note, and CSV (.csv) for the Instrument Register, Version Reconciliation, Discrepancy Register, Consolidated Cap Table and Fully Diluted View, zipped together. They open natively in Word, Pages, Google Docs, Excel, Numbers and Sheets.

How is this different from a cap table spreadsheet template?

A blank grid assumes the numbers going into it are right, and the numbers are the problem. This one forces two fields no template carries: the instrument the holder signed, and the board consent that authorised it. A row with only the first is the most common serious defect there is.

What is putative stock?

Delaware's ratification statute defines it as shares that would be valid but for a failure of authorisation, or that the board cannot determine to be validly issued. In practice: a signed agreement, paid consideration, a certificate, and no board consent. From inside a spreadsheet it looks exactly like the row above it.

Can it fix a missing board consent?

No, and nothing should. Curing a failure of authorisation is a board action taken with counsel, and Delaware provides routes for ratification and for a determination by its Court of Chancery. The pack's job is to find it, size it, and name who decides. It is not legal or tax advice.

Does it handle SAFEs and convertible notes?

It registers them and prices their conversion in the Fully Diluted View, including any instrument the spreadsheet never knew about. For scenario work across several instruments with different caps and discounts, use the SAFE and note conversion model alongside it.

When should a founder do this?

Three weeks before diligence rather than during it, because everything here gets found eventually and the only question is who finds it first. It pairs with the term sheet explainer and the consents your board meeting pack produces. Where the founder rows are themselves unsettled, the founder equity split model comes first.

Find out which of your four cap tables is closest to the paper

Download the blank pack as Word and CSV files, or open it in River and have your own instruments and board consents read first, before any spreadsheet is opened.

Edit with AI