On this page
The distinction the category blurs
Every equity tool on the market sells you a cap table and calls it the truth. Carta, Pulley, Mantle, Cake: the promise is a single, clean, always-current picture of who owns your company. It is a good picture. It is not your corporate record, and the difference is the one that surfaces at your worst possible moment, in the middle of a financing or an acquisition, when someone asks you to prove that the picture is true.
A cap table is a summary of ownership. The corporate record is the evidence that the ownership is real. One is a spreadsheet-shaped view; the other is a set of signed, dated legal documents that authorize every number the view displays. When they agree, nobody notices the difference. When they drift apart, and they drift by default, the cap table is the version that looks right and the record is the version that counts.
This piece is about that distinction: what each one is, why they separate, how to tell whether yours still agree, and what it costs when they do not.
What a cap table actually is
A cap table is a derived view. It is the answer to a single question, rendered as a table: who owns what, and what would they get in a given scenario. It is genuinely useful for the things it is built for. Modeling a round. Sizing an option pool. Showing an investor the post-money split. Answering, in ten seconds, a question that used to take counsel a day.
But a cap table is a report, not a record. It carries no inherent authority. A row that says an investor holds 400,000 Series A preferred shares is a claim, and the claim is only as good as the documents behind it: the board resolution that authorized the issuance, the certificate or book-entry that evidences it, the subscription agreement the investor signed, the register entry that logs it. Delete all of those documents and the cap table row does not change. It still shows 400,000 shares. It just can no longer prove them.
That is the tell. You can edit a cap table into almost any shape you like, and nothing stops you, because it sits downstream of the things that are actually true. It reflects what you told it, not what you can prove. We wrote about what belongs on it, and what does not, in cap table basics for founders.
What the corporate record actually is
The corporate record is the controlling account of the company's existence and ownership. It is the minute book and everything the law expects to find in it: the articles and bylaws, the register of directors and officers, the securities register (the share ledger), the certificates or book-entry records, and the resolutions and written consents that authorize every action the company has ever taken. We covered the full contents in what a corporate minute book is.
The distinction that matters is authority. The record is not a description of what happened. It is the thing that made it happen and the evidence that it did. A share exists because a board resolution authorized its issuance and the securities register logs it, not because a cap table row asserts it. The record is signed, dated, and hard to change after the fact, which is exactly what makes it evidence.
A cap table is easy to change, which is exactly what makes it a view and not a record. The share ledger, not the cap table, is the controlling list of who owns the company. We covered what belongs in it, and why it is the source rather than the summary, in what goes in a stock ledger.
Where the two drift apart
If the record is the source and the cap table is the view, the two should always agree. In practice they diverge, and they diverge quietly, because the view and the record are kept by different people, in different systems, on different schedules.
The founder keeps the cap table. It updates the moment a deal is agreed, because the founder wants to see the new picture. Counsel keeps the record. It updates when the paperwork is executed, which might be days or weeks later, or, in a busy quarter, not until someone remembers. The gap between "the deal is agreed" and "the paperwork is signed and logged" is where drift lives.
A few concrete ways it happens. An option grant goes on the cap table the day it is promised, and the board consent approving it is signed a month later, or never. A share transfer is recorded as a clean new line, and the transfer instrument the parties were supposed to sign sits half-finished in an inbox. A SAFE converts and the cap table shows the resulting shares, but the conversion was never papered as an issuance the board approved. A founder issues advisor shares from a pool the plan does not yet authorize. Each of these is invisible on the cap table, which renders whatever it is given. Each is a hole in the record.
We catalogued the gaps diligence keeps finding in the records findings that delay closings, and the single most common, the issuance nobody can prove the board approved, in who authorized this issuance. Drift is not a sign of a careless founder. It is the default behavior of a system where the view is edited directly and the record is maintained separately. We traced the same mechanism from the records side in how share transfers quietly corrupt ownership records, and the timeline problem underneath it in why ownership breaks without a timeline.
The reconciliation test
Here is the useful part. You do not need a diligence process to find out whether your cap table and your record agree. You can run the check yourself. Take any line on your cap table and try to trace it back to the four things that have to exist for it to be real:
- Authorization. Is there a board resolution or written consent approving this issuance, grant, or transfer, dated on or before the date the cap table shows?
- Instrument. Is there a signed document that effects it: a subscription agreement, an option grant under an active plan, a transfer form signed by both parties?
- Register entry. Does the securities register record the same holder, class, number of shares, and date the cap table shows?
- Certificate or book-entry. Is there a certificate or book-entry record evidencing the holding, and does it match the rest?
If all four exist and agree, that line reconciles. If any one is missing or says something different, you have found drift, and you have found it on your own schedule instead of in a data room with a term sheet on the clock. Run the test on every line, and the lines that fail are your actual to-do list. This is, more or less, the order a buyer's counsel works in, which we walked through in what a diligence lawyer actually reads in your minute book.
Most founders who run this for the first time are surprised by two things: how many lines reconcile cleanly, and how the ones that fail are almost always the same few, the option grants without consents and the transfers without instruments.
What it costs when they don't agree
A cap table that does not reconcile to the record is not a bookkeeping annoyance. It is a priced risk that lands at the worst time.
In a financing, the company gives representations about its capitalization, and counsel has to be able to stand behind them. If the record does not support the cap table, the reps cannot be given cleanly, and the deal stops until the gap is closed. Closing a gap under deadline means reconstructing authorization after the fact: getting consents signed for issuances that already happened, chasing signatures on transfers from people who have since moved on, sometimes ratifying actions that were never properly approved. That work is expensive, it is done by lawyers billing by the hour, and it is done while the round waits. We wrote about why this is structurally a closing problem rather than a founder problem in cap table hygiene is a closing problem.
In an acquisition it is worse, because the buyer's counsel is looking for exactly these gaps and pricing them into the deal, into the indemnities, into the holdback, or into a delay that drains momentum. The cap table looked fine the whole time. The record is what got read.
How to keep them reconciled
The reconciliation problem has a structural fix, and it is not "reconcile more often." Manual reconciliation is the work the software was supposed to remove, and it is the first thing to lapse when the quarter gets busy.
The fix is to stop maintaining the cap table as a separate artifact and start generating it from the record. In that model there is one system. Authorization, instrument, register entry, and certificate are the workflow, and the cap table is the view that comes out of it. A number on the cap table cannot change unless the underlying record changes, which requires the same authorization the record has always required. The view is read-only by construction, because it is derived, not maintained.
Founders still get the dashboard, the modeling, the investor-ready export. They just cannot edit ownership into existence, because ownership lives in the record and the cap table only reflects it. That is the idea behind our digital corporate records and cap tables and financing: the register, the resolutions, and the certificates are the source, and the cap table is generated from them, so it reconciles by construction instead of by quarterly effort.
The bottom line
Your cap table is a view of your ownership. Your corporate record is the ownership. When the two agree, the cap table is the fastest way to see what you own. When they drift, and they drift on their own, the cap table becomes a confident-looking claim the record may not support, and diligence is where you find out which.
The category sells the view and calls it the truth. The truth is the record. The best thing you can do for your next round is to stop treating the cap table as the source and start treating it as what it is: the report that has to reconcile to the documents underneath it. Run the test on your own schedule, and the data room stops being the place you find out.
