Startup Offer Diligence: How Senior Engineers Vet the Company Before Signing (2026)

The offer lands as a two-page PDF. Base salary, a share count, and a sentence from the recruiter describing the grant as "about 0.4% of the company, worth roughly $600,000 at our last round." You have three days to decide, you are already down-leveled from your last title, and the number in the email is the only piece of financial information anyone has given you.

That sentence is not a valuation. It is a marketing claim built on a share price only preferred stockholders paid, a fully diluted count that may or may not include the option pool, and an exit that has not happened. Senior engineers routinely spend two weeks preparing for a system design loop and forty-five minutes evaluating the instrument that is half the package.

This guide is the diligence pass: the numbers to request, the questions that separate a funded company from one that is quietly out of money, the deal terms that can zero out your common stock in a real acquisition, and the tax mechanics that turn a resignation into a five-figure bill. All of it fits in the week between verbal offer and signature.

Why 2026 Made Offer Diligence Non-Optional

The venture market in 2026 looks spectacular in aggregate and grim in the median. The Q1 2026 PitchBook-NVCA Venture Monitor recorded $267.2 billion in quarterly deal value and $347.3 billion in exit value, both near-records. It also notes that if you exclude the five largest deals and the five largest exits, those figures fall by 73.2% and 86.6% respectively. Nearly all of the market's apparent health belongs to a handful of companies.

Downstream, the attrition is visible. CB Insights analyzed 431 venture-backed companies that shut down since 2023 and found "ran out of capital" cited in 70% of post-mortems, poor product-market fit in 43%. The Bureau of Labor Statistics Business Employment Dynamics series puts five-year survival for new US establishments near 50% -- and venture-backed software startups are a higher-variance subset of that population, not a safer one.

Employment risk moved the same way. Layoffs.fyi tracked more than 122,000 tech layoffs in the first seven months of 2026, already past the full-year 2025 total, and Crunchbase News' tracker shows the cuts hitting private startups as well as public companies. The question is no longer just "is this a good product" -- it is "does this company have the money to keep me employed through my one-year cliff."

The Four Numbers That Decide Whether the Equity Is Real

A share count alone is meaningless. Four numbers convert it into something you can reason about, and every reputable company will give you all four before you sign.

If a company will tell you the percentage but not the denominator, you are being sold a ratio without its terms. That refusal is itself a data point.

Ownership percentage is your grant divided by fully diluted shares: issued common, all preferred as converted, outstanding options and warrants, and the unallocated pool. The Holloway Guide to Equity Compensation is blunt about this -- a percentage quoted against issued shares can overstate your position by 15% to 25%. Ask explicitly: "Is that fully diluted, including the unissued pool?"

Then benchmark the grant. Index Ventures' Rewarding Talent dataset covers more than 20,000 option grants across 1,650-plus startups, with stage-by-stage ranges for senior engineering hires. A grant below the benchmark for your stage is negotiable exactly the way base salary is -- but only if you know the benchmark before the call.

Runway and Burn: The Questions That Reveal Everything

Runway is the single highest-signal number in the entire package, because every other risk resolves through it. Kruze Consulting, which handles accounting for hundreds of venture-backed startups, publishes median net burn of roughly $87,000 per month at seed and $380,000 per month at Series A, with payroll typically exceeding 75% of operating expense. Those two numbers plus cash on hand give you a date.

Compare the answer against how long rounds actually take. The Venture Monitor's median gap between rounds for non-AI companies sits at 1.8 years, and Carta's private-markets data has shown bridge financing absorbing a rising share of all capital raised as companies stretch between priced rounds. If the company has 11 months of cash and its peer group takes 21 months to raise, the gap is not a rounding error -- it is the layoff you will be part of.

Three follow-ups worth asking, in order:

• "Was the last round fully subscribed, or is it still open?" A round still accepting money months after announcement did not close at the headline number.
• "Was it priced, or an extension of the prior round?" Extensions and bridges usually carry the prior round's terms plus new sweeteners for insiders.
• "What happened to headcount over the last twelve months?" Net shrinkage with growing ARR is discipline. Net shrinkage with flat ARR is contraction.

The Preference Stack: How Common Stock Goes to Zero

Investors buy preferred stock. You get common. In any liquidity event, preferred holders are paid their liquidation preference before common holders see a dollar. This is the mechanism by which an engineer at a company that sold for $80 million walks away with nothing.

The good news is that standard terms are genuinely standard right now. Cooley's Q1 2026 Venture Financing Report, covering 165 financings, found 98.2% of deals carried a 1x liquidation preference and 96.4% used non-participating preferred -- the founder- and employee-friendly configuration in which investors choose between taking their money back or converting to common, not both.

The risk concentrates in companies that raised on tough terms. Cooley's Q1 2026 data shows pay-to-play provisions in 7.3% of deals and redemption rights in 6.1%, both up quarter over quarter, after pay-to-play hit 10.1% in Q3 2025. Down rounds ran 11.4% of Q1 2026 deals. Structure clusters in exactly the companies that need senior engineers to fix things.

The law firm Morse states the consequence plainly in its analysis of liquidation overhang: as an acquisition price falls toward the total preference amount, proceeds to common approach zero, and at or below it common holders receive nothing. Holloway's venture guide and Carta's explainer reach the same conclusion.

So ask the one question that surfaces it: "What is the aggregate liquidation preference across all preferred series, and are any of them participating or above 1x?" The number you want is total dollars, not a multiple. A company that has raised $180 million needs to exit above $180 million before your shares are worth anything at all -- and the Fenwick venture financing terms glossary is a useful reference for decoding whatever the answer contains.

Reading the 409A Against the Preferred Price

Two share prices exist inside every startup. The preferred price is what investors paid in the last round. The 409A valuation is an independent appraisal of common stock fair market value, and it sets your strike price. Recruiters quote your grant using the preferred price, because it is the larger of the two -- often by a factor of three or more.

Published benchmarks put the 409A common price at roughly 10–30% of the preferred price at seed, 20–40% at Series A, and 35–55% at Series B and beyond, with the gap closing as an exit becomes more probable. Qapita's breakdown of the two prices explains why: common carries none of the preferences, anti-dilution protection, or information rights that make preferred valuable, and is discounted again for illiquidity.

Two things follow. First, the "$600,000 grant" in the recruiter's email is a preferred-price number applied to common shares; valued at the current 409A it may be worth a third of that. Second, the ratio is diagnostic. A 409A marked down while the company still quotes the old preferred price to candidates tells you the internal view and the recruiting pitch have diverged.

Ask for the date of the most recent 409A and whether it moved up or down from the previous one. Companies refresh at least annually or after a material event, so a stale 409A on a company that raised six months ago is worth a follow-up question.

Exercise Windows, AMT, and the Bill After You Quit

Vesting is the part everyone reads. Exercise terms decide whether vested options ever become shares.

Most option grants give departing employees 90 days to exercise or forfeit. Carta's data on post-termination exercise periods shows that 91.4% of terminated grants on its platform carry a window of 90 days or less. The reason is tax law rather than malice: under IRS rules for incentive stock options, an ISO loses its favorable treatment if exercised more than three months after employment ends, so companies apply the same window across all grants for consistency.

The tax layer compounds it. Exercising ISOs and holding adds the spread between strike price and current 409A to your alternative minimum tax calculation on IRS Form 6251. Engineers have paid AMT on paper gains for shares that later became worthless. An extended post-termination window changes that risk entirely: it lets you wait for a liquidity event before committing cash.

Extended windows exist and are negotiable. Cooley GO's guide to extending exercise periods lays out the mechanics -- extending past three months converts ISOs to NSOs, which most engineers should accept in exchange for years instead of weeks. Carta's history of the 90-day window documents companies from Pinterest to Coinbase moving to seven-year and ten-year windows. Ask for it in writing before you sign; nobody has ever amended a grant on the way out the door.

Liquidity: Tender Offers and the Exit Clock

Equity you cannot sell is a lottery ticket with a decade-long draw date. The question is not "will this company IPO" but "what has it actually done to give employees liquidity, and when."

The secondary path has widened. Carta administered 71 tender offers in the first half of 2026 -- roughly $3 billion in volume, the highest H1 figures in at least six years, with transaction value up 200% year over year. Nearly 70% were run by Series C and later companies, and the median cap on employee sales was 20% of holdings. Forge Global reports the same from the marketplace side: liquidity is now a recruiting tool.

It is a late-stage benefit, though. A Series A company has no tender-offer history and will not for years, and the Q2 2026 Venture Monitor describes an exit environment still concentrated in a few mega-listings rather than a broad reopening. So ask:

• "Has the company run a tender offer or approved secondary sales -- when, at what price, and with what participation cap?"
• "Does the stock plan or a right of first refusal restrict private sales?" Most do. Assume no sale without board consent unless told otherwise.
• "What is the board's stated timeline for a liquidity event?" The answer is usually vague. How vague is informative.

What the Public Record Tells You Before You Ask

A meaningful share of this diligence happens without talking to anyone, which means you can walk into the offer call already knowing whether the answers are accurate.

Start with SEC EDGAR full-text search. Any US company raising from investors under Regulation D must file a Form D within 15 days of first sale, and as Fidelity Private Shares explains, the filing discloses the total offering amount, the amount actually sold, and the amount still remaining to be sold. A company that announced a "$40 million Series B" and filed a Form D showing $26 million sold against a $40 million offering did not raise what the press release said.

Then calibrate against stage benchmarks. Carta's graduation-rate data shows only about 15% of the 2022 seed cohort raised a Series A within two years, well below the 25–30% that was normal pre-2022. Carta's State of Private Markets tracks down-round and bridge activity by stage. Knowing the base rate for a company's cohort tells you how much of the founder's confidence is signal.

Finally, check exits and headcount. Public engineering departures over the last six months, the recruiter's own tenure, and whether the company appears in Layoffs.fyi or the Computerworld layoff timeline are all free signals. A team that lost two staff engineers in a quarter has a reason, and you are allowed to ask what it was.

How to Ask Without Torpedoing the Offer

The fear that stops most engineers is that asking looks mercenary. The opposite is true: founders who have run a real process expect these questions from senior hires, and the ones who bristle are telling you something you needed to know. Three delivery rules make the difference:

• Frame it as commitment, not suspicion. "I'm planning to be here for years, so I want to understand the equity well enough to treat it as real compensation" opens every door that "what's your runway" slams shut.
• Batch the questions into one message. Send them in a single email before the offer call, so the call is spent on answers rather than on someone hunting for a spreadsheet.
• Ask the founder, not the recruiter. Recruiters rarely have preference-stack data and will guess. At senior level you have earned fifteen minutes with someone who has seen the cap table.

What you do with a refusal matters more than the refusal. "We don't share that" on fully diluted share count is a red flag at any stage; "we don't share aggregate preference" is common at seed and unusual at Series C. Put the ask in writing so the answer is documented alongside the offer.

Scoring the Offer: A Decision Framework

Resolve the offer against a cash-equivalent baseline, not the recruiter's headline. Price the equity three ways: at zero, at the current 409A less exercise cost, and at a realistic exit that clears the full preference stack. If it is not competitive in the first scenario, you are being paid in optionality alone -- a legitimate choice only if you made it deliberately.

None of this argues against joining startups. Concentration risk is how outsized outcomes get created, and the engineers who capture them took that risk knowingly, on terms they understood. The failure mode is not risk -- it is unpriced risk: a $60,000 salary discount for an instrument whose value nobody would state, whose exercise cost you never calculated, and whose preference stack you never asked about.

Spend the week. Ask the questions. The company that answers them well is a better company to join, and the company that will not is the one the diligence was for.