The 6 Startup Valuation Methods and the Number Behind Each

Startup valuation methods look like independent answers to one question, yet they share their inputs. The venture capital method, comparable multiples, the scorecard and Berkus methods and discounted cash flow each import one number from a market. Early investors mostly set the price from their check and the stake they want, then test it against that borrowed number.

The people who set the price say as much. Paul Gompers, Will Gornall, Steven Kaplan and Ilya Strebulaev surveyed 681 venture firms, 885 investors in all. Anticipated exits and comparable companies were the two factors most often rated important in setting a valuation, by 86% and 80% of respondents, while only 22% used net present value methods at all (Gompers et al., 2016). Methods that rest on the same few numbers make weak checks on one another, and the sections below trace each one’s borrowed number.

Exhibit 1. Each startup valuation method imports one number from a market

Method	What it computes	The borrowed number	Where that number comes from
Check and stake	Post-money = check ÷ target stake	The stake the fund needs (20% on average at early stage)	The fund’s own portfolio model
Scorecard (Payne)	Regional median pre-money × weighted adjustments	Median pre-money of recent local pre-revenue rounds	Angel-group data
Berkus	Up to a set maximum for each of five risk-reducing milestones	The maximum per milestone	The investor, guided by angel averages for the region
Venture capital method	Exit value ÷ required multiple, less later dilution	The exit multiple	Today’s public or acquired comparables
Comparable multiples	Revenue or ARR × a peer multiple	The peer multiple	Public markets, acquisitions or private rounds
Discounted cash flow	Present value of forecast cash flows, adjusted for failure	The terminal value and the survival rate	Often comparables again; sector failure data

Source: CX Cash synthesis of Gompers et al. (2016), Payne (2019), Berkus (2009, 2016) and Damodaran (2009), cited below.

Early-stage investors set a startup valuation from the check and the stake

Roughly half of the respondents set valuations from the amount they invest and the ownership they want, and the split by stage is wide: 63% of early-stage investors use that rule, against 29% of late-stage investors. Early-stage funds target an average stake of 20%, late-stage funds 27%. A fund that writes a $3M check for 20% has fixed a post-money valuation of $15M and a pre-money valuation of $12M before any method runs.

The tests applied afterwards are blunt. Cash-on-cash multiples were the most common metric, used by 63% of respondents, followed by IRR at 42%; among early-stage investors only 12% used NPV. Those who set a required multiple asked for 5x at the median, perhaps, the authors suggest, because early-stage funds compute return targets on the scenario in which the company succeeds. Only 3% named competitive pressure as the most important factor in setting valuation.

The arrangement predates the methods. In 1957 two engineers from MIT’s Lincoln Laboratory, Kenneth Olsen and Harlan Anderson, asked American Research and Development (ARD) for money to build smaller, cheaper computers. “The founders sought $100,000. ARD offered $70,000 in equity for 70 percent of the company, with a $30,000 loan to follow. With no other offers, they accepted.” (Wikipedia, ARD). The price, a post-money valuation of $100,000, followed from a check and a stake. When Digital Equipment Corporation listed in 1966, ARD’s 65% stake was valued at $38.5M, 550 times the $70,000; Wikipedia’s article on Georges Doriot, ARD’s president, puts the original stake at 77%, so the exact percentage is uncertain.

A figure attributed to this survey, and repeated in an earlier version of this article, says that 95% of VC firms call the founding team the most important factor. The paper’s figures are two. The team was named “as an important factor (by 95% of the VC firms) and as the most important factor (by 47% of the VCs)”. The team decides whether an investor proceeds; at early stage, the check and the stake decide the price.

But the scorecard and Berkus methods borrow a local median

Before a venture fund leads, angels test a startup valuation with rules of thumb that start from a market number. Bill Payne’s scorecard method, first written in May 2001 and revised in 2019, “adjusts the median valuation of recently funded companies in the region to establish a pre-money valuation of the target” (Payne, 2019).

The adjustments are weighted: the management team up to 30%, the size of the opportunity 25%, product and technology 15%, competition and sales channels 10% each, and the need for more money and other factors 5% each. In Payne’s example a strong team (125% of the norm), a large market (150%), average scores on three factors and weaker ones on competition (75%) and sales channels (80%) sum to 1.155 times a $4.5M median, or about $5.2M. The median supplies about 87% of the answer, and a team twice as strong as the norm would add 30%.

Payne notes that others call it the Benchmark Method, and the surveyor’s sense of the word fits. A benchmark was a horizontal mark chiselled into stone, into which an angle iron could be set to hold a levelling rod. Each benchmark’s height is carried through a network from a fundamental benchmark tied to a vertical datum, typically mean sea level (Wikipedia, Benchmark (surveying)). A scorecard measures a company’s height above the local median. It cannot say where the median sits.

Dave Berkus built his method in the mid-1990s to value start-ups without their revenue projections. His premise, offered without a source, is that “fewer than one in a thousand start-ups meet or exceed their projected revenues in the periods planned” (Berkus, 2009). It credits up to $500,000 each for a sound idea, a prototype, a quality management team and strategic relationships. A product roll-out or sales earns the same again, “allowing for a pre-revenue valuation of up to $2 million (or a post rollout value of up to $2.5 million)”.

The milestones look independent of the market; the maxima are not. Berkus cut them after the bubble, and in 2016 told users to assign maximum values “appropriate to the industry and to the times”, noting that the Angel Capital Association’s HALO Report put the US average pre-money above his ceiling. A big-data start-up in Silicon Valley might call for $1.5M per element, the same company in Nebraska $500,000 (Berkus, 2016).

Exhibit 2. The same pre-revenue company, measured from five datums

Datum	Source and date	Pre-money implied
Payne’s starting average for pre-revenue deals	Payne, 2011 revision	$1.5M before adjustments
Berkus maxima at $500,000 per element	Berkus, 2009	Up to $2.0M ($2.5M after roll-out)
Payne’s starting median: midpoint of average pre-seed ($4M) and seed ($5M) deals	Payne, 2019 revision	$4.5M before adjustments
Berkus maxima at $1.5M per element (his Silicon Valley example)	Berkus, 2016	Up to $6.0M ($7.5M after roll-out)
Median pre-money of new primary seed rounds on Carta	Carta, Q3 2025	$16M

Sources: Payne (2011); Payne (2019); Berkus (2009, 2016); Carta (2025). The $6.0M and $7.5M maxima are our arithmetic. Angel rounds and venture-led seed rounds are different markets.

Payne’s own starting point tripled between his two revisions, and Carta’s seed median is 3.6 times the later one. A founder who brings an angel benchmark to a venture-led seed round, or the reverse, misprices the company by more than any scorecard factor can move it.

Therefore the venture capital method moves the question to the exit multiple

Once a venture fund leads, exit considerations rank first, and the venture capital method turns them into a startup valuation. As Aswath Damodaran of New York University sets it out, the method forecasts revenue or earnings to an exit two to five years ahead and values the company then at the multiple “that publicly traded firms in the sector trade at”. It discounts that value at a target rate “generally set high enough to capture both the perceived risk in the business and the likelihood that the firm will not survive”, then adds the new money (Damodaran, 2009).

The borrowed number is the multiple, and it moves. Jamin Ball’s Clouded Judgement newsletter divides enterprise value by next-twelve-months revenue for public cloud software companies. Its median was 20.8x in February 2021 and 4.2x on 25 September 2026 (Clouded Judgement, 2021; 2026). Growth sorts the companies: 18.0x for those expected to grow faster than 22% over the next twelve months, 6.7x between 15% and 22%, and 3.3x below 15%, cut-offs Ball himself calls arbitrary.

Exhibit 3. One hypothetical plan, priced at five exit multiples

Suppose a seed-stage SaaS company, hypothetical, plans $20M of ARR in five years and is raising $3M. The investor requires 5x, the survey’s median, and expects two later rounds of 20% each to leave it with 64% of its stake.

Exit multiple (EV ÷ next-12-months revenue)	Datum	Exit value	Post-money today	Stake bought by $3M
3.3x	Growth under 15%, Sep 2026	$66M	$8.4M	35.5%
4.2x	Median, Sep 2026	$84M	$10.8M	27.9%
6.7x	Growth of 15% to 22%, Sep 2026	$134M	$17.2M	17.5%
18.0x	Growth over 22%, Sep 2026	$360M	$46.1M	6.5%
20.8x	Median, Feb 2021	$416M	$53.2M	5.6%

Source: multiples from Clouded Judgement (12 February 2021; 25 September 2026); the company is hypothetical and the rest is CX Cash’s calculation. Post-money = exit value × 64% ÷ 5.

The same plan is worth about 80% less at the 2026 median than at the 2021 median, and the gap between the slow- and fast-growth medians is 5.5 times. Damodaran’s objection is that the method stops at the exit and borrows today’s multiple for a sale years away; leaving the later cash flows unestimated does not remove their uncertainty. Because the founder pushes the forecast and the multiple up while the investor pushes them down, “the projected value becomes a bargaining point between the two sides rather than the subject of serious estimation.”

But a headline private round can overstate its multiple twice

Private comparables, the other common input to a startup valuation, come from three markets, and none matches the public median. SaaS Capital, a lender to software companies, finds private multiples lower, all else equal: its January 2025 model put private companies at 4.8x ARR if bootstrapped and 5.3x if equity-backed, when its public index stood at 7.0x, or 24% to 31% lower (SaaS Capital, January 2025).

Private SaaS acquisitions priced at a median 4.0x trailing revenue in the second quarter of 2026, by Software Equity Group’s count of deals with disclosed multiples (SEG, 2026). The rounds that make headlines sit far above both: Bessemer’s 2025 Cloud 100 of top private cloud companies averaged 20x, and its AI companies 24x (Bessemer, 2025).

A headline round can overstate its multiple twice. Its post-money valuation includes the cash just raised, and it prices every share at the price of the newest preferred share. William Gornall and Ilya Strebulaev show the second effect with Square. Its October 2014 Series E sold preferred shares at $15.46, and 388 million fully diluted shares at that price gave the reported $6B. The Series E holders were promised at least $18.56 a share in an IPO, ahead of every other shareholder, and when Square listed in November 2015 at $9 they received extra shares. The authors’ model valued Square at $2.2B after the round. “Square’s reported post-money valuation overvalued the company by 171%.” Across their sample of 135 unicorns the overstatement ranged from 5% to 250%, and 65 of them lost unicorn status once such terms were priced (Gornall & Strebulaev, 2017).

Surveyors learned to keep one datum. Great Britain’s Ordnance Survey measured heights from mean sea level at Liverpool, recorded over a short period in 1844. In 1921 it adopted mean sea level at Newlyn, recorded between 1915 and 1921, and by 1950 some 40% of the lower-order levellings still used the old one (Wikipedia, Ordnance datum). A comparables table that mixes 2021 rounds with 2026 public multiples, or a private post-money with a public enterprise value, makes the same error. The scale differs: Liverpool and Newlyn differed by 0.13 feet, while the public software median fell by about 80% between February 2021 and September 2026.

DCF values the cash flows, but the terminal value and the failure rate decide the answer

Discounted cash flow is the one method that does not start from a price, and few venture investors use it: 22% overall and 12% at early stage, on the survey’s figures. For an early startup valuation the reason is structural. Damodaran writes that “it is not unusual for the terminal value to account for 90%, 100% or even more than 100% of the current value of a young company”. The common shortcut for that terminal value is a multiple, which he calls “inconsistent with the notion of intrinsic value.”

The venture capital method’s target rate hides a second borrowed number, the failure rate. Damodaran’s table puts typical targets for a start-up at 50% to 70% a year, while early-stage venture investors earned 21.4% a year over the 20 years to 2007. Survival data close most of that gap.

In Bureau of Labor Statistics data analysed by Amy Knaup and M.C. Piazza and reproduced by Damodaran, 31.24% of new information-sector establishments founded in 1998 survived five years. A 50% target over five years asks for 7.6 times the money; weighted by that survival rate, it asks for 2.4 times, or about 19% a year, close to what early-stage funds earned. The survival data cover all new businesses, not venture-backed ones, so the match is indicative. A single rate also assumes the risk of failure is the same every year. Across all firms in the same data, the annual failure rate held between 17% and 19% for four years and fell to 9.5% in year seven (our arithmetic).

Damodaran’s worked example shows the divergence on one company. His hypothetical anti-virus start-up, Secure Mail, is worth $202.34M under the venture capital method, using a 50% target rate and a revenue multiple taken from two public competitors. Valued by discounted cash flow as a going concern at $177.56M and weighted by a 60% chance of survival, it is worth $106.54M. The choice of method moves the value of one company by $95.8M, or 47%.

Damodaran would replace the venture capital method with a DCF

Damodaran rejects the premise that a startup valuation can only be a price. His paper argues that the venture capital approach “is flawed and should be replaced”, and that “there is no reason why young companies cannot be valued systematically”. He values the company as a going concern and then adjusts for the probability of failure, “rather than try to adjust the discount rate for this likelihood”.

Damodaran has the target rate right, and an intrinsic value can be estimated: a survival-adjusted DCF with a growth-based terminal value is the only method in Exhibit 1 that can disagree with the market. As an account of how a venture price forms, his case is thinner. His own discount rate starts from the betas of public companies in the same business, and his survival rates are sector averages. The fund across the table has a limited life, “typically ten years with three years of automatic extensions” in the survey’s words, so the exit price is what pays it. The DCF tests the datum; it rarely sets the price.

Three questions to put to any valuation before the first meeting

Every startup valuation can be checked with three questions: which market number the method imports, the date of that number, and the market it came from.

Which number does the method import? In most early rounds it is the investor’s target stake, since most early-stage investors work from the check and the stake they want. A founder can therefore negotiate the stake as well as the price. A board can ask for the borrowed number to be reported beside every valuation, with the comparables and the revenue base behind it.

When was that number set? An exit multiple carries a date. Priced at the September 2026 median, a plan is worth about a fifth of what the February 2021 median implied, so the founder’s question to an investor is which month the model’s multiple comes from.

Which market did it come from, and on what terms? A private round’s price comes bundled with its terms. A post-money valuation with an IPO guarantee is a higher price than the same number without one, and the dilution calculator below tracks the founders’ stake round by round. Investors and boards can value holdings by share class rather than by post-money: in Gornall and Strebulaev’s Square case, the gap between the two was larger than the company’s fair value. They can also separate the probability of failure from the required return, as Damodaran proposes.

Startup valuation methods do not average into a truth, because they share their inputs. The check and the stake set most early prices; the scorecard and Berkus methods adjust a local median; the venture capital method and comparables import a multiple; and a DCF often borrows its terminal value from the same comparables. Most of that picture rests on what 885 investors told Gompers and his co-authors in a survey first published in 2016, and rounds priced since then may weigh the inputs differently.

Sources

• Berkus, D. (2009). The Berkus Method: valuing the early stage investment. Berkonomics (blog), 4 November 2009.
• Berkus, D. (2016). After 20 years: updating the Berkus Method of valuation. Berkonomics (blog), 4 November 2016.
• Bessemer Venture Partners (2025). The Cloud 100 Benchmarks Report 2025.
• Carta (2025). State of Private Markets: Q3 2025.
• Clouded Judgement (Jamin Ball, Altimeter Capital). Newsletter issues of 12 February 2021 and 25 September 2026.
• Damodaran, A. (2009). Valuing Young, Start-up and Growth Companies: Estimation Issues and Valuation Challenges. Stern School of Business, New York University.
• Gompers, P., Gornall, W., Kaplan, S. N. and Strebulaev, I. A. (2016). How do venture capitalists make decisions? NBER Working Paper 22587; published in the Journal of Financial Economics 135(1), 169–190, 2020, and summarised in Harvard Business Review, March–April 2021.
• Gornall, W. and Strebulaev, I. A. (2017). Squaring venture capital valuations with reality. NBER Working Paper 23895; published in the Journal of Financial Economics 135(1), 120–143, 2020.
• Payne, B. (2011). Scorecard Valuation Methodology: Establishing the Valuation of Pre-revenue, Start-up Companies, January 2011 revision.
• Payne, B. (2019). Scorecard Valuation Methodology (Rev 2019). Angel Capital Association blog, 21 October 2019.
• SaaS Capital (2025). 2025 Private SaaS Company Valuations, 24 January 2025; Lucas, R., SaaS Valuation Multiples: Understanding the New Normal, 6 August 2025.
• Software Equity Group (2026). SaaS M&A and Public Market Report, 2Q26.
• Wikipedia. American Research and Development Corporation; Benchmark (surveying); Georges Doriot; Ordnance datum.

Figures without a cited source, and those marked “our arithmetic”, were derived by CX Cash from the cited numbers; Exhibit 3 is hypothetical, and Secure Mail is Damodaran’s hypothetical company. The survival rates come from Damodaran’s reproduction of Bureau of Labor Statistics data for businesses founded in 1998, and the DEC terms from Wikipedia, whose articles disagree on the stake. The public and private multiples match those on CX Cash’s SaaS valuation multiples page. This piece is education, not investment advice.