Capitalization Table

Ownership gets promised out loud long before it gets written down. Three founders agree over coffee to split everything into equal thirds. Two months later someone joins and is told they will get one percent once things settle. The first investor signs an actual contract. Three promises about the same company, and each one survives in a different form — one in memory, one in a chat thread, one in a legal document.

The trouble shows up afterward. Once the company has grown enough that people start asking what their stake is worth, nobody can answer. The founder still thinks in thirds, but two rounds have closed since that conversation. The early engineer was promised one percent, but nobody ever wrote down one percent of what, measured when. In a company that keeps issuing shares, one percent later is a smaller piece than one percent at the start, and the gap between those two readings is where arguments begin.

The state of things usually becomes visible during an investment review. An investor asks for a one-page ownership breakdown, and the company spends three days producing it. Someone digs through five contracts, a folder of board minutes, and chat logs from years ago, adds up the percentages, and gets 103 instead of 100. Somewhere along the way the same slice was promised twice. Until it becomes clear whose three percent that is, the investment conversation sits still. Nobody in that chain lied. Each document is accurate on its own, and the thing missing is a single place where all of them are read against each other.

A capitalization table — a cap table in everyday use — is the document that collects those scattered promises into one sheet and forces them to sum to exactly 100 percent. Each row carries a holder's name, the class and number of shares they hold, and the share of the whole that represents. Shares that have not gone to anyone yet, and promises that will convert into shares later, belong on the same sheet; leave them off and every percentage on it is wrong. That last point carries a phrase people use constantly: the fully diluted count, meaning the share total as it would stand if every reserved share and every convertible promise turned into stock today. A percentage quoted on any other basis flatters whoever is quoting it. The moment a company needs this table is not the day it raises money. It is the day it first splits ownership at all.

Within finance, the job this table does is specific. Valuation settles what the company is worth; the cap table settles who holds how much of that worth. That is why an investor reviewing a deal asks for it near the top of the list, every time.

Reading one cap table takes four things: what kinds of shares appear on it (classes), how unassigned ownership is set aside in advance (reserves), how everyone's percentage moves when new money arrives (change), and what promises live in the contracts rather than in the percentage column (rights).

Classes exist because two people can hold stock in the same company and hold very different things. That is why rows split by share class rather than by person.

1. Common stock — the basic form, held by founders and employees. It carries one vote per share at shareholder meetings and stands last in line whenever the company distributes money. It comes with no special protection, and in exchange it captures the full upside if the company does well.
2. Preferred stock — the form investors receive. It carries rights that rank ahead of common stock: dividends paid first, or money recovered first if the company winds down. Investors insist on it because their downside is sharper than a founder's. If the company fails, the founder loses years; the investor loses cash.

Preferred stock usually arrives with more than voting rights attached — how many board seats the holder may appoint, which decisions need their consent. Taken together those terms decide who supervises the company and what it cannot decide alone, and that structure has its own name: governance.

Reserves cover ownership that has no owner yet. An option pool is the block of shares set aside for people the company plans to hire and for those already on the team. Reserve ten percent and that ten percent belongs to nobody in particular, yet it occupies a row on the table from day one. Writing down an empty slice is what keeps the remaining percentages honest. A founder stake of 60 percent calculated without the pool drops to 54 the moment hiring starts. Shares leaving that pool reach employees as equity compensation, and on the table, how much of the pool is left maps directly onto how many more people the company can hire before it has to create another one. Investors read the same row for the opposite reason. A pool running low signals another round of dilution arriving before the next senior hire does.

Change covers the moments the numbers move. Dilution is what happens when new shares are issued and existing holders end up with a smaller percentage. Nobody takes their shares away — the denominator simply grows. Hold 20 shares out of 100 and you hold 20 percent; issue 25 new shares so the total reaches 125, and those same 20 shares are now 16 percent. Dilution is not a loss by itself. If the company was worth 4 billion won before the round and 9 billion after, the smaller slice points at a larger amount of money than the bigger slice did. What matters is the direction the two numbers move together. A founder holding 30 percent of a company worth 9 billion won is ahead of one holding 45 percent of a company worth 4 billion, and rounds get agreed to on exactly that trade.

Rights cover the promises that never appear in the percentage column. They sit in the contracts, invisible to anyone reading only the ownership split, and they determine what each person actually receives when the company is sold. They are why two companies with identical ownership splits can pay their holders completely differently. Four show up constantly.

1. Liquidation preference — the right of an investor to recover money first when the company is sold or wound down. An investor who put in 20 billion won with a 1x preference takes 20 billion off the top of the sale price before anything is divided by percentage.
2. Anti-dilution — a clause that protects an investor when the next round prices the company lower than the last one did. The investor receives additional shares to offset the drop in value, which means founders and employees absorb that much more dilution.
3. Drag-along right — the right of a defined majority of shareholders to force the remaining holders to sell on the same terms. Without it, one holder with two percent can stall an entire acquisition.
4. Tag-along right — the right of minority holders to join a sale on the same terms when a major shareholder sells. If the previous clause protects the majority, this one protects everyone else. It stops a large holder from exiting at a good price and leaving the rest behind with an unfamiliar owner.

The four connect in sequence. Deciding what class to issue leads into negotiating what rights attach to it; issuing the shares recalculates everyone's percentage as dilution; and that calculation has to count the reserved block that still has no owner. Change one cell and the others move with it.

"Can you run the cap table simulation with this round's terms in it?"

A request to calculate, in advance, what each person's percentage becomes once the proposed terms are applied.

"Let's settle whether the option pool goes in pre-money or post-money first."

Who absorbs the dilution from the pool depends on when it is created, so that question comes before the rest.

"If the table and the contracts disagree, diligence stops there."

Diligence is the review an investor runs over a company before wiring money, and mismatched numbers halt it on the spot.

During the week Pagedeck closed its Series A, Grace spent half of one day inside this table. Entering the investor's proposed terms dropped the two founders' combined stake from 62 percent to 48, which was the expected result. The next line was the problem. The term sheet required a 12 percent option pool measured before the new money went in, and on that basis the pool came entirely out of existing holders, pushing the founders to 43 percent. Same investment amount, same company valuation, five percentage points moved by a single phrase.

Grace built both versions of the table and brought them to the board side by side. Only with the two numbers next to each other did the discussion shift from whether 12 percent felt like a lot to how many percentage points the next two years of hiring actually needed. The pool ended up at nine percent, with part of it measured after the round instead of before. The table did not make that decision. But without it, those five points would have gone into the signed documents with nobody noticing. Terms of that kind are not hidden. They are written in a register where no percentage ever appears, so somebody has to translate them into the table before a board can see what they cost.

The most common error is reading the ownership percentage as the share of proceeds. The table shows percentages and nothing else, so a 20 percent holder in a company selling for 50 billion won expects 10 billion. What actually happens is that preference money leaves first. If an investor put in 20 billion with a 1x liquidation preference, that 20 billion comes off the 50 before the remaining 30 is divided, and the 20 percent holder receives 6 billion rather than 10. The gap widens as preferences stack up across rounds, and in a modest exit they can absorb the whole sale price before common holders see anything at all. Someone who misses this hears about the acquisition, does the simple arithmetic, and discovers the gap only after the deal has closed. It is the reason the percentage column and the rights column have to be read together.

Treating the cap table as paperwork for fundraising is the other one. The first time most companies are asked for it is during an investment review, so it looks like an investor's document. But the events it records happen much earlier: the day founders divided the company between themselves, the day an early hire was promised shares, the day a friend put in a small amount of money. Rows not written then cannot be reconstructed later. Memory differs by person, and chat histories disappear when a company switches tools. Leave those rows blank for a few years and a departed early employee claiming a promised one percent leaves the company with no grounds to refuse and none to agree either. Investment reviews do stall at exactly this point, and untangling it afterward costs far more than recording it would have.