Turning an Equity Grant Into a Number You Can Compare

A job offer with equity in it is really two offers stapled together: a guaranteed salary you can compare against anything, and a lottery ticket denominated in a currency nobody agrees how to price. This guide is for candidates holding or expecting an offer who need to put those two things on the same axis. It gives you a framework - the six dimensions that set the discount, how to score each one, and what the score means - so you finish with a single risk-adjusted annual dollar figure and a memo that says exactly why.

Most ranking guides stop at "apply a 30 to 60% haircut to private equity" and move on, which leaves you guessing which end of the range applies to your grant. That guess is where two people looking at the same offer end up thousands of dollars apart. The fix is to stop treating the haircut as one number and start treating it as a stack of separate judgements, each with its own evidence.

What a defensible equity number actually is

A defensible equity number is a net-of-tax expected value, annualised over the vesting schedule, with every haircut you applied written down and attributed. It is not the face value on the offer letter, and it is not the headline valuation times your ownership percentage.

The honest approach computes two numbers in order. First the conditional payout: what your equity is worth if the company actually exits at a given value. Second the expected value: that conditional payout weighted by how likely the exit really is. Equidam frames it plainly - always compute both, because the first tells you the size of the prize and the second tells you whether to count on it.

The full conditional-payout formula looks like this:

Conditional payout =
  ownership %
  x exit value
  x (1 - future dilution)
  x (1 - preference haircut)
  - (strike x shares)
  - tax

Expected value = conditional payout x realistic exit probability

Annual equity value = net expected value / vesting years

Every term in that formula is a dimension you score separately. The rest of this guide is how to fill each one without fooling yourself.

The six dimensions that set the haircut

Six dimensions, scored one at a time, turn a gut feeling into a number two people can reproduce. Each one proves something specific, and each one has a tell when it lies to you.

Public RSUs sit at the shallow end of this: they carry only price and tax risk, no illiquidity haircut. Private options sit at the deep end because they stack illiquidity, the preference gap, the strike cost, and failure risk all at once. That is the whole reason a blanket percentage fails - it treats a public RSU and a private option as if they lived on the same scale.

Dilution: the haircut that compounds before anyone exits

Dilution quietly erodes an early stake by nearly half before any exit, because each priced round issues new shares and shrinks your percentage. You apply it by multiplying your ownership by (1 minus per-round dilution) for every round you expect between now and the exit.

Use published medians rather than guessing. Two datasets converge closely, which is itself reassuring:

Applying the Carta medians multiplicatively from seed through Series C erodes roughly 45% of a seed-era stake. That matches the pattern where founders typically retain about 36% after Series A and about 15% at IPO. If you join at seed and the company raises three more rounds, model those three cuts explicitly. If you join at Series C and expect one more round plus an exit, you apply far less. The stage you join at is the single biggest lever on this line.

Probability: use the stage-specific rate, not the 90% meme

Weight the conditional payout by a survival rate that matches the company's stage, not a blanket failure statistic. The "90% of startups fail" meme flattens the exact differences that should move your number.

The honest range is wide and worth internalising. Across a 2,000-company dataset, 75% of venture-backed startups never return cash to investors, and investors lose their full stake in 30 to 40% of cases. General business survival is gentler: BLS data shows 20.4% of new US businesses fail in year one and 49.4% by year five. Venture-backed firms specifically show about 32% five-year survival, versus roughly 58% for bootstrapped companies.

The practical move is to pick a probability that reflects both the stage and the specific company's traction, and to write it down as an assumption you can defend. A Series C with a clear line to profitability and named secondary activity deserves a materially higher weight than a Series A burning fast with no revenue. Do not write off the Series C at seed-stage odds, and do not credit the Series A with Series C odds.

The failure rate you plug in is a judgement, so make it a documented judgement rather than a reflex.

Liquidity: near-cash or lottery ticket

The strongest signal that private equity is close to cash is an active, sanctioned secondary or a repeated tender, not the funding stage. Stage tells you about survival; liquidity tells you about timing, and they are different questions.

Named secondary activity is hard evidence. NPM tender-offer volume doubled to $6bn in 2024 from about $3bn the year before, with $10bn projected the following year. A company appearing on a platform's approved list is itself a signal - one platform lists investment opportunities in over 200 leading private companies including SpaceX, OpenAI, Anthropic, Databricks, Figma, and Stripe. The counter-signal is duration: average time from founding to IPO has stretched past a decade for many venture-backed companies, so a locked grant can stay locked for years.

If you want to test whether your grant is genuinely near-cash, look at the actual secondary channels and their friction:

Those fees are a real haircut on any value you might actually realise, and the minimums tell you whether your position is even large enough to sell. A grant that clears a $100k seller minimum on an active platform is a different asset from one that does not.

The procedure, step by step

Run these eight steps in order. Each has an owner and a rough time, and each ends with a concrete artefact so you know when to move on.

The tax step deserves its own attention because it is where identical face values diverge most. NSOs trigger ordinary income tax on the spread between strike and fair market value at exercise, treated as regular compensation. ISOs pay no regular income tax at exercise if you hold, but the spread may trigger Alternative Minimum Tax, and favourable capital gains treatment requires holding at least one year after exercise and two years after grant - sell earlier and profits convert to ordinary income. ISOs also carry a $100,000 cap on grant value that can become exercisable in a calendar year. RSUs are taxed as ordinary income at vesting with no exercise cost. Two grants with the same face value can land 10 to 30% apart net once you run these through.

How this goes wrong

The failure modes below are where careful people still produce a wrong number. Each one has a false positive - a way the mistake makes the figure look more credible than it is - and a specific check.

• Double-counting illiquidity. Applying a 20% illiquidity haircut and a low exit probability that already prices illiquidity produces a "conservative" number that is actually punitive. Pick one convention. Equidam warns illiquidity is often already inside the exit probability; Blind stacks them multiplicatively. Either is defensible if you document it. Doing both silently is not.
• Valuing common at the preferred price. Multiplying shares by the last round's price per share gives a headline number 10 to 30% too high, because that price reflects preferred shares with preferences your common lacks. Use the 409A common FMV.
• Ignoring the strike and AMT drag. Treating an option grant like an RSU makes it look identical to a same-face-value RSU. Subtract the exercise cost and model AMT on the bargain element before you compare.
• Year-one figure ignoring the cliff. Dividing the grant by four when nothing vests before month 12 overstates near-term realised value. Confirm the cliff date and the forfeiture terms.
• Reading stage as liquidity. Assuming late-stage means cashable is wrong; a Series E with no tender is still locked. Check for a named sanctioned tender or an approved-list secondary.
• Trusting the 90% failure meme both ways. A blanket rate flattens stage differences and can write off a Series C at seed-stage odds. Use stage-specific survival - about 32% five-year for VC-backed firms - not a headline number.
• Anchoring on old Black-Scholes. Historically option values ran 33 to 34% of the underlying stock price at grant, but that figure moves with volatility and term. Recompute for your grant rather than reaching for the old anchor.

Who can pin down the numbers you can't

The specialists who price equity for a living are scarce and geographically lumpy, so most candidates outside the US will have to self-model or source a reviewer deliberately. In Refolk's index of professional profiles there are 3,220 US compensation and total-rewards specialists versus 56 in the UK, a roughly 57x gap, and only 166 US stock-plan administrators - about 19x fewer than comp analysts.

Two inputs are worth outsourcing even when you model the rest yourself: whether a company has actually run a sanctioned tender, and how the ISO AMT math lands for your specific position. A stock-plan administrator who worked at the company, or a former employee who sold on a secondary, can confirm the liquidity history that no headline will tell you. This is where naming the right person by role and stage saves hours, and where Refolk turns a vague "I should ask someone" into a specific list.

If you are pricing an ISO grant, a tax-side reviewer is the other high-leverage contact, since AMT on the bargain element is where self-modelling most often goes wrong.

Before you call the number final

Run this checklist before you drop the figure into your offer comparison. If any item is unchecked, the number is not yet defensible, and a number you cannot defend is worse than no number.

Keeping the number current

An equity valuation is a snapshot, not a fact, because three of its inputs move. The 409A FMV is re-set periodically by appraisal, funding rounds change your dilution path and can reset the preference stack, and liquidity conditions shift as tender activity opens or closes. Re-run the framework whenever any of those three changes, and whenever you are within a few months of a vesting cliff, since the cliff is the moment the year-one figure stops being hypothetical.

The mechanism to re-check is simple: ask the company for the current 409A FMV and the latest cap table position at each new round, and watch for announced tenders or secondary programs. If the company runs a tender, that single event can move your grant from the illiquid quadrant toward near-cash and justify thinning the illiquidity haircut. The framework stays the same. Only the inputs age, and the memo you wrote tells you exactly which line to update.