The Startup Equity Grant, From Offer Letter to an Expected-Value Number

You have a startup offer with stock options or RSUs, and the recruiter quoted a number - say "$146K in options." Before you weigh that against a cash offer or a second bid, you need to know what the equity is actually worth: net of the strike price you pay to exercise, the dilution to come, the discount between common and preferred stock, and taxes. This guide carries one grant through every step of that arithmetic to a defensible dollar range, including the wrong turns, so you can reproduce the math on your own offer instead of trusting a black-box calculator.

Every other offers guide stops at decoding clauses or comparing cash take-home. None of them carries a grant through the full chain to a single comparable figure. This one does.

Why the headline number is the least reliable figure in the offer

The dollar figure a recruiter quotes is almost always priced off the last preferred share price, but the systems that actually pay you - taxes and exit proceeds - value common stock, which is worth far less. That gap is the single biggest source of overstatement in a startup offer.

Preferred shares are what investors buy. They carry rights common stockholders never get: a liquidation preference that pays them back first, anti-dilution protection, board seats, and conversion rights. Those rights hold real economic value, and it is stripped out of your common stock. A 409A valuation, the independent appraisal of common stock the company commissions, is typically 25 to 60 percent below the last preferred price. At seed stage the discount widens to 70 to 90 percent. It narrows to 15 to 30 percent as a company approaches an IPO and exit outcomes become more certain.

So when you multiply your share count by the last preferred price, you are valuing your common stock as if it were preferred. It is not. Recompute using the 409A common FMV, and a "$146K" grant at seed may be worth a fraction of that on paper before you have touched dilution or taxes.

The formula chain, from share count to a net number

Ownership percentage is your shares divided by the fully diluted share count; notional value is that percentage times the valuation; the real number is what remains after strike cost, dilution, the liquidation stack, and taxes. Each link in that chain subtracts, and skipping one produces a number that is confidently wrong.

Start with the base case the sources use. If you hold 1,000 options in a company with 100 million fully diluted shares, your stake is 0.001 percent. Multiply by a $1 billion valuation and your options are theoretically worth $10,000, minus the cost to exercise. That is the entire skeleton. Everything else in this guide adjusts that base for reality.

The critical word is "fully diluted." The denominator must be the total count across all classes: founder stock, the employee option pool, and the preferred shares issued to investors, plus any warrants outstanding. Ask the company for the fully diluted count, not the pool size. If they hand you the pool, your percentage looks several times larger than it is. One caveat the count itself hides: SAFEs and convertible notes are excluded from stated fully diluted figures, so you must estimate their conversion and add it in, or your ownership will shrink the moment they convert.

The dilution model: today's percentage is not exit's percentage

Every priced round after your grant issues new shares, so your percentage shrinks even as the company grows. Median dilution is around 20 percent per round in Carta's 2024 data, and modeling it is the difference between a fantasy and a forecast.

The compounding is what people miss. Ten thousand options when the company has 1 million shares is 1 percent today. Carry that through several rounds of roughly 20 percent dilution each and the same stake can land at 0.25 percent or less at exit. Option-pool top-ups add dilution on top of the priced rounds, and over 95 percent of term sheets specify the new pool comes out of the pre-money valuation, which means existing holders absorb it.

Dilution and valuation growth pull in opposite directions, and only the net matters. Your percentage falls, but if the valuation climbs faster, your dollars can still rise. The point is to model both, not to assume the percentage is flat. Decide how many priced rounds you realistically expect between your grant and an exit, then dilute for each.

The liquidation stack, and why small exits pay common zero

Preferred investors are paid back before common stockholders see anything, so a modest acquisition can return zero to you even when the company "sold for money." This makes the value of common non-linear across exit scenarios, not a single multiplier you can wave at every outcome.

The standard structure is a 1x liquidation preference: preferred receives its invested capital back first, and only the remainder flows to common. When the exit valuation is below the total preferred investment, common shareholders receive nothing. Above a threshold, preferred faces a choice. If it would receive more by converting to common and sharing pro-rata, it ignores the preference and splits by ownership. It takes either the preference or the pro-rata split, but never both. Participating preferred and higher multiples like 2x or 3x shift more of the proceeds away from common, so read your term sheet for those.

The practical consequence: for each exit valuation you model, subtract the total stack of preferences first. If the number goes negative, common gets $0 in that scenario. This is why the downside case for a startup grant is often not "a little less" but "nothing."

Grant type decides the tax, and ISOs can bill you with no cash

Your grant type determines when you owe tax and at what rate, and ISOs in particular can generate a cash bill on gains you have not sold. Get this wrong and your "net" figure counts money the tax authority will take.

NSOs, or non-qualified options, are taxed at exercise: the spread between FMV and strike is ordinary income plus payroll taxes, reported on your W-2 in Box 12 with code "V." RSUs are taxed when they vest and shares are delivered, with the FMV counted as ordinary income and any later gain or loss on sale treated as capital gain. ISOs, or incentive stock options, owe no regular income tax at exercise, but the spread is an AMT preference item. The Alternative Minimum Tax is a parallel tax that can catch that paper spread. To earn the favorable long-term capital gains rate, ISO shares must be held at least two years from grant and one year from exercise.

The AMT trap is concrete. A single engineer earning $200,000 exercises 10,000 ISOs at a $5 strike while the 409A is $20. That is a $150,000 bargain element and roughly $31,100 of AMT triggered - a real cash cost on gains never received. The 2026 AMT exemption is $90,100 for single filers and $140,200 for married filing jointly, with rates of 26 percent and then 28 percent above $244,500 of AMT base. Long-term capital gains for 2026 run 0, 15, or 20 percent federal plus a 3.8 percent net investment income tax, against ordinary rates up to 37 percent.

Run it: the eight-step procedure

Here is the full procedure, in order, from the moment you have the offer to a single number you can set against cash. Doing the common-versus-preferred correction early, at step 3, prevents you from anchoring on an inflated headline; some sources fold it into the exit-scenario step instead, and that is also defensible.

Collecting inputs is the only step gated on someone else. Expect one to two days for the company to return the fully diluted count and preferred price. If they refuse the fully diluted count, treat that as a signal in itself. Refolk can draft the exact information request from your offer letter and history so you ask for every field at once instead of trading emails for a week.

The table below shows the common-versus-preferred discount you apply in step 3, by stage. Read it as the haircut you take off the headline before anything else happens.

A worked case: "$146K in options" carried to a number

Take the recruiter's headline of "$146K in options" and walk it through the chain. The exact figures depend on your grant, but the shape holds for every seed-stage offer.

The $146K was priced off the last preferred price. Step 3 says recompute on common. At seed, common is 10 to 30 percent of preferred, so the honest current notional value is closer to $15K to $44K, not $146K. That is the first and largest correction, and it happens before dilution or taxes touch the number.

Now step 4. Suppose two priced rounds remain before a plausible exit. At roughly 20 percent dilution each, your projected exit ownership is 0.8 times 0.8, or about 64 percent of today's percentage. A 0.5 percent stake becomes about 0.32 percent at exit, before any pool top-up.

Step 5 applies exit scenarios against the liquidation stack. In a downside acquisition below total invested capital, common receives $0 - the number is binary, not scaled. In a base case that clears preferences, credit your projected 0.32 percent of the proceeds available to common. In an upside where preferred converts, the same percentage applies to a larger pool.

Step 6 subtracts the strike cost and taxes for each surviving scenario. If your grant is ISOs and you exercise, model the AMT on the bargain element as a cash outflow even before a sale. Step 7 weights the scenarios. The base rate is unforgiving: roughly 75 percent of venture-backed companies never return capital, and 30 to 40 percent lose everything. AngelList estimates a 1-in-40, or 2.5 percent, chance a seed-stage startup reaches unicorn status. An unweighted base case ignores all of that.

The base rate argues for a steep risk discount, not optimism: three in four venture-backed companies never return a dollar.

Scenario     | Exit value | Clears prefs? | Common % at exit | Gross to you | Less strike+tax | Net    | Probability
Failure      | $0         | No            | 0.32%            | $0           | $0              | $0     | 0.55
Downside     | small M&A  | No            | 0.32%            | $0           | $0              | $0     | 0.20
Base         | mid M&A    | Yes           | 0.32%            | (fill)       | (fill)          | (fill) | 0.20
Upside       | large exit | Yes, converts | 0.32%            | (fill)       | (fill)          | (fill) | 0.05
Expected value = sum of (Net x Probability), then divide by vesting years to annualize.

How this goes wrong: the failure modes that inflate the number

Almost every overstatement comes from one of eight recurring errors, and each has a false positive you can catch by rerunning one calculation. This is the section to reread before you trust your own model.

• Multiplying shares by the last preferred price. The false positive is an inflated "worth." Recompute on the 409A common FMV; at seed, common may be only 10 to 30 percent of preferred.
• Using the option-pool percentage as fully diluted ownership. The false positive is a percentage several times too high. Confirm the denominator includes preferred, warrants, and an estimate for SAFEs and notes.
• Ignoring dilution to exit. The false positive is today's 1 percent treated as 1 percent at exit. Apply about 20 percent per remaining round; 1 percent can become 0.25 percent.
• Ignoring the liquidation stack. The false positive is crediting common with a pro-rata share of a small acquisition. Subtract total preferences first; below that threshold, common gets $0.
• Forgetting exercise cost and phantom AMT on ISOs. The false positive is a "net" figure that ignores a cash tax bill on gains you never received.
• Assuming refresh grants offset dilution. The false positive is modeling ownership as flat. Refreshes are discretionary; model only what is granted.
• Ignoring the vesting cliff and unvested fraction. The false positive is valuing the full grant on day one. Apply the vested fraction for the horizon you actually expect to stay.
• Treating the base-case exit as certain. The false positive is no risk discount. Around 75 percent of VC-backed companies never return capital.

Note that the comparable public guidance you inherit is heavily US-shaped. In Refolk's index of professional profiles, 347,874 people are currently titled Software Engineer in the United States against 43,070 in the United Kingdom, roughly an 8.1x larger pool. US-benchmarked equity norms dominate the guides a UK-based reader will otherwise absorb, so sanity-check any benchmark against your own market.

Set it against cash and keep the model current

The whole point of the arithmetic is a single like-for-like comparison: annualize your expected-value range over the vesting period and place it beside the cash you are giving up for this offer. If the annualized equity expected value does not cover the cash gap, the risk discount has done its job and the cash offer wins on the number, whatever the upside story.

Before you call the model done, run this check.

The inputs decay. Your 409A resets after each priced round, dilution changes your denominator, and tax exemptions shift by year, so revisit the model after any funding announcement rather than trusting a stale figure. When you want a reality check from people who lived through it, the fastest route is to ask someone who held common stock through a real exit and can tell you what actually landed in their account.

Talk to a startup finance lead about your specific 409A and liquidation stack before you sign, and treat any offer where the company will not share the fully diluted count as a grant you cannot value - which is a finding in itself.