How to Value Your Startup for an Exit: Multiples, Comparables and What You Actually Keep

Every founder carries two valuations in their head. One is the number from the last priced round, the one that appeared in TechCrunch and anchors the option grants. The other is the number a buyer will actually wire at closing. In 2025 the gap between them was wider than at any point in a decade: PitchBook data reported by Fortune in August 2025 put down rounds at 15.9% of venture deals, a ten-year high, and almost every major Q2 2025 IPO priced below its peak private valuation.

This is the second of three foundation articles in the Founder Exit Playbook. The first mapped the exit routes. This one explains how a buyer, an underwriter or a secondary investor will put a number on your company, and how your own cap table and the tax code decide how much of that number is yours. The third covers the diligence and legal process that follows.

You will read about multiples, comparables and discounted cash flow, but the single most useful section is the waterfall. Most founders only model it for the first time when a term sheet is already on the table. Do it now, with your actual preference stack, and you will negotiate differently.

Where do you stand today?

Which valuation methods do buyers actually use when a startup is sold?

Textbooks list half a dozen methods. In a live sale process, two do the work and the rest provide cover. The first is the revenue or ARR multiple: enterprise value divided by trailing-twelve-month (LTM) or next-twelve-month (NTM) revenue. It dominates whenever a company is growing fast and profit is thin or negative, which describes most venture-backed startups. Buyers quote it because it is comparable across companies at different stages of margin maturity. You should know your LTM and NTM ARR to the dollar and understand which one the buyer is using, because the gap between them at 40% growth is a 1.4× difference in headline value.

The second is the EBITDA multiple, which takes over once a company has meaningful, defensible earnings, in practice above roughly $5–10M of EBITDA. Private equity buyers in particular live here because their debt packages are sized on EBITDA, and a lender will not finance a revenue multiple. Silverpeak's Q4 2025 software M&A report puts the historical EV/EBITDA band for software at 12–17×, with 2025 averages inside it despite a 3.4-turn year-on-year decline. If you are approaching that EBITDA threshold, the decision about when to flip from a revenue story to an earnings story is itself a valuation decision.

Precedent transactions and trading comparables are the inputs that set the multiple. Precedents are what buyers paid for similar private companies; trading comps are where similar listed companies trade today. Bankers triangulate both and then argue about which peers belong in the set. Your job is to shape that set before the buyer does: a vertical SaaS company with 120% net revenue retention wants to be compared with Veeva and Procore, not with a basket of horizontal tools growing 10%. Software Equity Group data cited in the SaaSrise 2025 report shows why this matters: public SaaS companies with NRR above 120% traded at a median 11.7× revenue, those at 100–110% at 6.0×, and those below 90% at 1.2×.

Discounted cash flow appears in fairness opinions, in the buyer's internal investment committee memo and in the goodwill impairment test afterwards. It rarely sets the price for a startup because a DCF on a company burning cash is dominated by a terminal value nobody believes. The VC method, which works backward from a target exit value and required return, is how your investors priced your last round, and it is why they have a mental floor under your exit price. The Rule of 40 (growth rate plus profit margin at or above 40) is not a method at all, but buyers use it as a screen: companies clearing it sit in the upper half of every multiple table, and in 2025 that gap widened rather than closed.

What did benchmark multiples look like in 2025–26, by sector and growth?

Start with the public market, because every private valuation is a discount to it. Jamin Ball's Clouded Judgement index put the median public SaaS company at 5.0× NTM revenue in September 2025, with a brutal spread by growth: companies projected to grow above 25% traded at 23.0×, mid-growth (15–25%) at 7.3×, and low growth (under 15%) at 4.2×. SaaS Capital's own index stood at 6.7× ARR in June 2025, and Bessemer's Cloud Index median of 7.5× in early 2025 was up 25% year on year. Aventis Advisors' tracker shows the public median had slipped back to 3.4× by March 2026. The lesson: the market multiple moves by a factor of two inside eighteen months, and your exit will be timed to one point on that curve.

Private M&A multiples sit below public ones and are far more dispersed. Software Equity Group recorded a 4.1× median and 6.0× mean revenue multiple for private SaaS deals in 2024. Aventis, using Mergermarket data on 543 SaaS transactions since 2015, shows a long-run median of 4.7× and a 2025 median of 3.8× on a small sample, with a clear size effect: deals under $50M cleared at roughly 3.0–3.3× revenue while deals above $50M cleared at 5.1–6.2×. SaaS Capital's 2025 private valuation estimates were 5.3× ARR for equity-backed companies and 4.8× for bootstrapped ones. On EBITDA, Aventis shows SaaS M&A medians of 19.3× in 2024 and 28.3× in 2025's small sample; Silverpeak's broader software set averaged 14.9×.

Outside SaaS, multiples compress. Consumer subscription and e-commerce businesses typically change hands at 1–3× revenue or 6–12× EBITDA unless they have a brand a strategic badly wants. Marketplaces are valued on net revenue (take rate), not gross merchandise value, which is where many founders' expectations break; a marketplace with $200M GMV and a 12% take rate is a $24M-revenue company in the buyer's model. Fintech lenders are valued on book or on earnings, not on origination volume. Services-heavy businesses, including many AI-enabled services companies, are capped by their gross margin: below 60% gross margin, software multiples simply do not apply, regardless of what your pitch deck calls the company.

The AI premium is real but narrower than it looks. Bain's 2026 M&A report notes that tech M&A rose 77% in 2025 and that almost half of strategic technology deal value above $500M came from AI-native companies. Kroll's Winter 2026 software update recorded a record 2,897 announced software deals and $291B of value in 2025. But the premium accrues to companies with proprietary models, data or distribution and genuine usage growth; a wrapper on a third-party model with 30% gross margin is a services business in disguise and will be priced as one after the quality-of-earnings review. Treat the AI label as something the buyer will test, not something you can assert.

What real exits teach: three cases

Why do strategic buyers pay more than financial buyers, and what is control worth?

A financial buyer, usually a private equity fund, prices your company on its standalone cash flows plus whatever leverage a lender will provide. Their model has a required return, typically a 20–25% gross IRR over four to six years, and the price is whatever makes that return work. A strategic acquirer prices your company on what it is worth inside their business: your revenue through their sales force, your product filling a gap in their suite, your engineers, or the simple removal of a competitor. That synergy value is real to them and unavailable to a sponsor, which is why strategics can rationally outbid. Kroll's Winter 2026 data quantifies it for software: strategic buyers paid 5.6× LTM revenue in 2025 versus 4.4× for financial sponsors, a 27% premium, and the strategic multiple was itself up 14% from 4.9× in 2024.

The premium is not automatic. A strategic pays it only when you are on their roadmap and the process makes them believe a competitor might get you first. If you walk into one strategic alone with no alternative, they will price you like a sponsor and take longer. Our dedicated articles on selling to a strategic and selling to private equity go into process design. For valuation purposes, remember that the strategic premium is paid for scarcity and fit, and that a sponsor's lower headline often comes with cleaner cash terms and less earnout.

Control premium is the related concept for listed companies and for majority stakes. It is the amount above the trading price (or above a minority valuation) that a buyer pays for the right to run the business. BVR/FactSet's control premium study has historically shown rolling medians in the low-to-mid 20% range across all transactions, with wide dispersion by sector and by how contested the deal was. Slack's shareholders received roughly $45.86 per share from Salesforce against a pre-rumour price under $30, a premium above 50%. For a private founder the point is practical: a 100% sale commands a control premium; a secondary sale of a minority stake usually carries a discount of 20–40% to the last round price instead. Same company, two very different per-share numbers.

How will the buyer adjust your number before closing?

The multiple gets the headlines; the adjustments get the money. The first is the quality of earnings (QoE) review, in which the buyer's accountants, typically a Big Four or specialist firm, rebuild your revenue and EBITDA from the ledger up. They will test revenue recognition, strip one-off and non-recurring items, reclassify customer-success salaries from sales into cost of revenue if that is where they belong, and recompute your ARR on their definition rather than yours. Expect the QoE to find something. A 5–15% haircut to reported EBITDA is routine; a change in how you count ARR can be worse, because it hits the multiple as well as the base.

Normalised EBITDA is the output: your earnings adjusted for owner compensation at market rate, for pro forma cost savings the buyer is willing to credit (rarely all of them), and for run-rate effects of recent price changes or hires. Then comes the move from enterprise value to equity value. Net debt is deducted: venture debt, deferred revenue treated as debt-like by some buyers, accrued but unpaid bonuses, deferred taxes, and unpaid transaction bonuses. Cash is added back, but only cash above the working capital peg, the normal level of net working capital the buyer has agreed the business needs. If you close below the peg, the shortfall comes out of the price pound for pound; if above, you are paid for the surplus. Negotiating the peg is worth as much as a quarter-turn of multiple.

These mechanics appear in the share purchase agreement as a closing-accounts adjustment or, increasingly in sponsor deals, as a locked-box price fixed at a historical balance sheet date with leakage protections. Locked box gives you certainty earlier and shifts cash-flow risk to the buyer; closing accounts gives the buyer a second bite. Which you get depends on jurisdiction and leverage. UK and continental European deals lean locked box; US deals lean closing accounts. Whatever the mechanism, model it on your own numbers before the LOI, because a buyer who has agreed a headline and then discovers a $2M debt-like item will not raise the price to compensate.

When should you sell: on the growth curve or at the top of the market?

There are two timing questions and founders usually conflate them. The first is where your company sits on its own growth curve. Buyers pay for the growth in front of them, not behind them, which is why a company growing 60% with $8M ARR can command a higher multiple than one growing 25% at $20M ARR. The time to run a process is when you can show twelve months of accelerating or at least stable growth, strong retention and a pipeline that supports the NTM number. Selling after growth has visibly decelerated, which is when most founders feel the urge, is selling at a lower multiple on a number the buyer no longer trusts.

The second question is where the market sits. The public SaaS median went from 5.0× to 6.7× to 7.5× and back to 3.4× between 2025 and March 2026 depending on which index and month you pick; the 2021 peak was above 15×. You cannot time this precisely, but you can refuse to ignore it. A founder who spent 2022 waiting for 2021 multiples to return sold in 2024 at 4×. Carta's data shows that 1.2% of startups on its platform were acquired in the first half of 2025 alone, 502 companies, which it described as high relative to prior years: when the window opens, volume follows, and so does competition for buyer attention.

The right answer is usually to prepare continuously and to launch when both curves are favourable rather than waiting for either to peak. Have the data room ready, the QoE pre-done by your own advisers, and the cap table waterfall modelled. Then a strong inbound or a hot market becomes an option you can exercise in weeks rather than a scramble that takes two quarters, during which the market you were trying to catch may have moved. The acquisition preparation guide covers the operational side of this readiness.

What is the waterfall from a $100M headline to the founder's bank account?

Here is the section to read twice. Assume a buyer agrees a $100M enterprise value for your company. You are a founder holding 25% on a fully diluted basis. You raised $30M across seed, Series A and Series B at 1× non-participating liquidation preference; preferred holders own 45% fully diluted. There is a 12% option pool, a $3M convertible note that converts at closing, $5M of venture debt, and $3M of cash in the bank. The headline is $100M. Let us walk down.

First, enterprise value becomes equity value. The $5M of venture debt is repaid, the $3M cash is added back, and the closing balance sheet lands $1M below the working capital peg. Equity value is $97M. Transaction fees come next: a banker at 1.5% of value ($1.5M) and legal, accounting and QoE support of roughly $1.2M. Net distributable proceeds are $94.3M. The buyer then holds 10% of that, $9.4M, in escrow for 15 months as security for warranty claims, which SRS Acquiom's 2025 study shows is the median across 2,200 private-target deals. If there were an earnout, part of the price would also be contingent on future performance; in SRS Acquiom's non-life-science sample, 22% of 2024 deals included one.

Now the preference stack. Preferred investors can take their $30M back or convert to common and take their 45% share. At $94.3M distributable, 45% is $42.4M, so they convert, and everyone shares pro rata. Your 25% is worth $23.6M gross, of which $21.2M arrives at closing and $2.4M sits in escrow. Had the Series B instead negotiated participating preferred, investors would take their $30M first and then 45% of the remaining $64.3M, a total of $58.9M, leaving $35.4M for common and options. Your share of the common pool would fall to about $16.1M. Same company, same headline, $7.5M less for you because of one word in a term sheet three years earlier.

Finally, tax. Treatment varies by jurisdiction and by how your shares were acquired, so this is mechanics, not advice. A US founder paying federal long-term capital gains at 20% plus 3.8% net investment income tax, before state tax, would keep roughly $18M of the $23.6M, and far more if the shares qualify as Section 1202 QSBS, which can exclude up to $10M of gain per issuer (raised for post-July 2025 issuances under the 2025 US tax legislation). A UK founder may access Business Asset Disposal Relief on a limited lifetime amount; elsewhere, rates range from zero to over 30%. The honest summary: a 25% founder in this structure takes home roughly 18% of the headline, and a founder in a participating structure or a higher-tax jurisdiction takes home less.

What else can shrink the number?

Several items are easy to forget. Stacked preferences in a company that raised at falling valuations can mean later investors sit senior to earlier ones, so the waterfall must be run series by series, not as one preferred class. SAFEs and notes with valuation caps convert into more shares than their face value suggests when the exit is below the cap, diluting you at the moment of liquidity. Unvested options with acceleration consume part of the pool; the buyer may also require a retention pool carved from the price. Management carve-out plans, transaction bonuses and change-of-control payments all reduce distributable proceeds, and in the US a Section 280G excise tax can apply to executives whose payments exceed a safe-harbour multiple of their historical compensation.

Why is your paper valuation not your exit valuation?

The last round's post-money is not a market price for the whole company. It is the price one investor paid for one slice of preferred stock with a package of rights attached: a liquidation preference, anti-dilution protection, information rights, sometimes a board seat and a veto. Those rights are valuable, which means the common stock sitting beneath them is worth less per share than the preferred, and the company's 409A valuation will say so. When a buyer offers one price for every share, the preference structure that justified the round price falls away, and the whole-company price almost always sits below the round's headline, even for a healthy business.

Then there is the market. The 2021 cohort raised at multiples the public market no longer supports. Fortune's August 2025 report on PitchBook data listed Q2 2025 IPOs almost all pricing below their peak private valuations: Chime at $9.1B versus a $25B peak, Hinge Health at $6.2B versus $23B, Circle at $5.8B versus $7.7B. Lincoln International's Q2 2025 venture-growth valuation primer tracked the same pattern across private marks. Down-round exits happen for a simple reason: a company that raised at 30× ARR in 2021 and tripled revenue since is still worth less than its last round at today's 5–7× public multiple. The business did everything right and the valuation still fell.

This is where anti-dilution and preference terms bite. Full-ratchet or weighted-average anti-dilution provisions issue investors extra shares if the exit (or a prior down round) prices below their entry, pushing common further down the waterfall. Senior preferences mean the newest money is paid first. In a sale at or near the total preference stack, common stock, which is what you and your employees hold, can be worth close to nothing while the headline still reads in the hundreds of millions. The IPO route has its own version of this problem: ratchets and IPO-price protections in late-stage rounds that issue investors more shares if the listing prices below their entry.

Figma is the live illustration of how far the swing runs in both directions. Adobe agreed to buy it for $20B in September 2022, roughly 50× its revenue at the time, then terminated the deal in December 2023 after UK and EU regulators objected, paying a $1B break fee. Twenty months later, Figma listed on the NYSE at $33 a share, opened at $85 and closed its first day at $115.50, a market value of almost $68B according to CNBC's reporting, on quarterly revenue of about $250M. The company's intrinsic progress between those dates was real but gradual; the valuation moved by $48B because the buyer, the market and the multiple all changed. Treat every valuation as a snapshot of a transaction, not a property of the company.

The process, step by step

Red flags and founder risks

Frequently asked questions

Related reads from Kurums

Winding Down a Startup: Asset Sales, Fire Sales and the Orderly Shutdown

Oct 14, 2026

The exit nobody pitches at demo day but most founders eventually run. How to decide, which wind-down route preserves value, who gets paid in what order, and how to shut down without personal liability.

read more

Management Buyouts and Employee Ownership: Exiting Without Selling to an Outsider

Oct 14, 2026

The insider exit family explained for founders: who buys you out, how the debt gets paid, what trustees and lenders will accept as a price, and what reaches your account when the buyer is your own team.

read more

Taking Your Startup Public: IPO, Direct Listing and SPAC as a Founder Exit

Oct 14, 2026

An IPO does not cash you out; it changes the currency you hold. This guide covers readiness, process, cost, the three listing formats and how founders actually sell after the bell rings.

read more