Cap table: what it tracks, and the four events that quietly change it

What it actually records

A capitalisation table is the authoritative record of who owns a company: every share class, option, warrant and convertible instrument, held as of today and as it would convert under a future financing or exit.

The common mental model — a pie chart of percentages — is the output, and treating it as the input is where the errors start. Percentages are derived from share counts under a stated set of assumptions. Change the assumption and the percentage changes without a single share moving.

That is why "we own 60%" is an incomplete sentence. Sixty percent of issued shares today, or of fully diluted, or after the outstanding SAFEs convert, or after the pool expansion the term sheet requires? Those are four different numbers about the same company, and at seed stage they can differ by fifteen points.

The four rows

Issued shares, by class

Common for founders and employees who have exercised; preferred for investors, usually one class per round with its own rights.

The class matters more than the count. Preferred shares carry the liquidation preference, and in a modest exit that preference decides who gets paid before the percentages ever apply.

The option pool, split

Not one number. Granted options belong to identified people on vesting schedules; unallocated pool is reserved and unassigned.

Keeping them apart is the whole point. Unallocated pool dilutes you already — it exists in the fully diluted count — while doing nothing for you until it is granted. A large unallocated pool that never gets used is ownership given away for hiring that did not happen.

Convertible instruments

SAFEs, convertible notes, warrants: each with its cap, discount, and whether it is pre- or post-money.

These are not shares yet, which is exactly what makes them dangerous. They sit outside the share count, feel like debt, and convert all at once at the priced round.

The preference stack

Which class gets paid first, at what multiple, and whether it also participates in the remainder.

This is the row most often missing from a founder's spreadsheet and the one that decides outcomes below the very large. A cap table without it describes ownership in a good exit and says nothing about a mediocre one — which is the more likely one.

The four events that change ownership without a new round

1. Option pool expansion

An investor requires the pool go from 10% to 15% before closing. Created out of the pre-money, it is funded entirely by the existing shareholders — at seed, the founders.

The arithmetic is worth doing once by hand, because it is invisible in the headline. On a $2M raise at $8M pre with founders at 80%: the round alone takes them to 64%. Add a pre-money pool expansion of five points and they land near 60% instead. The valuation did not change. The ownership moved by four points, and nothing in the term sheet's headline says so.

The negotiable part is not usually whether there is a pool but where it comes from and how big. A pool sized to an actual hiring plan for the next eighteen months is defensible; a round number chosen because it is conventional is worth arguing about.

2. SAFE conversion

Instruments raised over a year at different caps convert simultaneously at the priced round.

Each looked small. A $250K SAFE at a $6M cap feels like nothing. Four of them, at caps from $6M to $12M, converting alongside a $3M Series A and a pool expansion, is a different conversation — and the first time most founders model it is when the Series A term sheet arrives, which is too late for it to influence anything.

Model the cap table at conversion before signing the next instrument, not after. This is the single highest-value hour in early-stage finance and it is almost never spent.

3. Anti-dilution adjustment

If a later round prices below an earlier one, anti-dilution reprices the earlier investor's stake.

Broad-based weighted average — the common, reasonable form — adjusts proportionally to how much was raised at the lower price. Full ratchet reprices the entire earlier stake as though it had been bought at the new lower price, and in a meaningful down round it can move a double-digit share of the company from founders and employees to one investor.

The term is invisible while things go well. It exists precisely for the case where they do not.

4. The preference stack paying out

Not a change in percentages, but a change in who gets money — which is what percentages were a proxy for.

Sell for $20M with $12M of 1x non-participating preferred outstanding: preferred holders take the better of $12M or their converted share, and common splits what is left. Make that preference 2x participating and $24M comes off the top before anyone else sees a dollar — on a $20M exit, common gets nothing.

This is why a $10M valuation with a 2x participating preference can be worth less to a founder than $7M on clean terms, and why negotiating the preference stack before the headline number is the correct order.

A worked cap table through two rounds

Two founders, 4,000,000 common each, 8,000,000 issued.

Pre-seed. $500K on a post-money SAFE at a $5M cap — 10% post-conversion, fixed. Nothing appears on the cap table yet. Founders still read 100% of issued shares, which is true and misleading.

Seed. $2M at $8M pre-money, and the investor wants a 12% post-close option pool, created pre-money.

Work it in order, because the order is what does the damage:

Founders hold 63.4% where a naive reading of "we sold 20%" would have said 80%. Nothing here is unusual or predatory — every one of those steps is a market-standard term. The gap is entirely the compounding of three things agreed at different times.

The pool matters twice over: created pre-money it costs the founders alone, and because it was sized at 12% rather than to a hiring plan, roughly half of it will still be unallocated a year later — ownership surrendered for hires that were never made.

Fully diluted, and why the other number flatters

Issued shares exist today. Fully diluted adds everything that could become a share: unexercised options, the unallocated pool, warrants, convertibles at their conversion terms.

Investors quote ownership fully diluted, always. Founders quote issued, usually without noticing. The difference at seed stage is routinely ten to fifteen points, and it is the reason two people can leave the same meeting with different beliefs about who owns what.

Quote fully diluted, and state what is included. "63% fully diluted, including the unallocated pool and both outstanding SAFEs at their caps" is a sentence nobody has to interpret.

What breaks a cap table

Kept in a spreadsheet nobody reconciles. The version in the deck, the version in the data room, and the one the lawyer maintains diverge within a year. Only one of them is the legal record, and it is not usually the one being shown.

Percentages entered directly. The moment a cell holds 20% rather than a share count, the table stops being derivable and starts being asserted. Errors become invisible because nothing is inconsistent with anything.

Convertibles left off until they convert. They are not shares, so they get omitted, so their effect arrives as a surprise at exactly the moment when nothing can be done about it.

Vesting ignored. A founder who leaves at eighteen months with a four-year schedule and a one-year cliff has earned less than they hold. A cap table showing their full stake describes a company that does not exist.

No scenario view. A single table describes today. What matters is what happens at a $20M exit, a $200M exit, and a down round — three views the same data produces and the flat table hides.

The habits that keep it honest

One source of truth, and it is the legal one. Whatever the lawyer or the platform maintains is the cap table; everything else is a copy that will drift.

Model before signing, not after. Every instrument, at the moment it is proposed, against the round it will convert into. An afternoon spent here has changed more founder outcomes than any negotiation over headline valuation.

Size the pool to a plan. Name the roles for the next eighteen months. A pool derived from that is defensible in the negotiation and does not silently gift away what you do not use.

Read the preference stack first. Before the valuation. It determines the outcome in every scenario except the one everybody is imagining.

Keep the scenario view current. Three exit values, one down round. If the numbers surprise you when you build it, they would have surprised you far more expensively later.