How VCMC values private companies

One number per private company: what its whole equity is worth today, the price it would raise or sell at now, lowered by how far it has slid toward winding down. On every page the word for it is valuation. Five more figures come with it.

A company page also shows the last round's price, the company's phase (after a round, before its first round, no priced round yet, or valued on company size), its headcount and its headcount trend, and a chart of the valuation over time: each point on it is what the model would have shown on that date, using only what was known then.

A valuation is not a price anyone paid, and it is not an appraisal or investment advice. A company acquired after its last funding round keeps the value it had on the day of the deal, worked out from what was known then, and is left out of every total and ranking. Public companies are not on VCMC.

A valuation starts from the company's funding rounds, taken in date order. The model prices seed and pre-seed rounds, Series A to H and later, and growth rounds; bridge rounds and venture rounds of unknown stage stay on the company page but set no price. Rounds on the same day are taken in stage order. Three checks catch errors in the data before anything is priced.

The typical round: seed $3M, Series A $12M, Series B $30M, Series C and later $60M, growth $70M.

A round with a disclosed post-money valuation is priced at it. A round with only an amount is priced at the amount divided by the share of the company a round of that stage usually sells.

price of a round = amount raised / share of the company that stage usually sells

The share depends on the stage, on when the round was raised (seed and pre-seed rounds before 2018 sold 25%; seed rounds of 2021 and 2022 sold 18%) and on the region: outside North America the same round is taken to sell a slightly larger share, so the same amount gives a lower price. A European Series A of $20M is priced at $20M / (18% x 1.09) = $102M.

Later rounds are also priced from the round before. A second round of the same stage that is no larger than the first (a top-up) is priced at no more than the first round's price, or 70% of it when its amount is not known. Any other later round is priced a second way too: the previous round's price times the usual step-up between the two stages, moved by the company's headcount growth in between. The two prices are blended, the step-up counting for up to 50% when the rounds are close together and for nothing once they are 3 years apart, and not at all when the round is under 0.8x the previous one. A round with neither an amount nor a valuation takes the step-up from the round before, or, as a company's first round, the typical post-money valuation of its stage and region.

A round priced from its amount that comes to less than $50K per employee does not fit the company's size: the company is then valued on its size instead, at no more than 10x what the round implies, and its page reads "valued on company size".

While the company is alive, its valuation rises from the round's price toward the price of its next round, reaching it one year after the round. The next round's price is the round's price times the usual step-up to the next stage, moved by the company's headcount growth since the round.

price of the next round = price of the round x step-up to the next stage x (headcount now / headcount at the round) ^ 0.22

Doubling the team since the round raises the next round's price by 16%. The headcount at the round is the reading nearest the round's date within 180 days; without one, headcount growth is left out.

If 3 years pass with no new round, the rise fades back toward the round's price by about 10% a year: fully when headcount is flat or falling, less as it grows, and not at all once it has grown about 35% or more over the last 18 months, nor while its trend is unknown. A company that has gone quiet is not assumed to keep climbing.

On each date the model reads the company's latest headcount on or before that date. Today it also has the company's current headcount: when that and the latest reading in its history differ by more than 3x, VCMC cannot tell which one is right and treats the headcount as unknown. A company with no readings uses its current headcount.

The trend compares the latest reading with the first over the last 18 months. It sets the company's odds after a round, below, and it is shown on the company page.

For the winding-down probability the model also measures the trend as a number: the rate at which headcount changed over the same window, fitted through every reading in it.

When headcount falls, the valuation slides from the rising path toward a winding-down path, gradually and never as a switch. How far it has slid is the winding-down probability: any fall in headcount over the last 18 months starts the slide, and a fall of about 18% completes it, up to 97%.

The probability remembers. For up to 24 months after the last round, a past fall still counts, weighing 37% as much after a year, so a company that stops shrinking recovers over about a year instead of jumping back. A new round starts the count again.

The winding-down path itself starts from the round's price. After a year in which nothing new is learned, it splits in two: an up path that reaches the next round's step-up 2 years later, and a down path that loses about 45% a year, to no less than 10% of the price. Where the company sits between the two depends on the odds that a company of its stage, this long after its round, raises again, is sold, closes or stays quiet within three years, moved by its headcount trend (both tables are in the figures below).

valuation = rising path, slid toward the winding-down path by the winding-down probability

The range slides with it: its high end moves from the next round's price toward the up path, its low end toward the down path.

A company that has raised venture money but has no priced round yet, and whose founding year is known, is valued at its industry level: what a company of its industry, region and era is priced at in its first seed round, measured on 2,310 first seed and pre-seed rounds. It starts from $1.81M and is multiplied by factors for the era, the region and the sector, and by 1.118x for each doubling of headcount above or below 16 people.

industry level = base x era x region x sector x headcount factor

Up to age 6 the industry level sets the company's value. After that, the industry level counts for less each year, in proportion to how many companies still raise their first seed at that age, and the company's size and revenue count for the rest; with neither on record, the industry level alone sets the value at any age until the first round. On the chart, the dates before a company's first round are valued this way too: before its first round, a company is valued as a startup because it later raised. A company with only bridge rounds or rounds of unknown stage is valued the same way, and its page reads "no priced round yet".

A company that has never raised venture money is valued from its size and revenue alone, since nothing in the data tells a young startup from a small business that will never raise: its headcount times the revenue per employee typical of its sector, times a revenue multiple for the sector and a level for its region, or, where its revenue fits its size, its revenue times the sector's multiple. Its range runs from 1.5x lower to 1.5x higher, widened further because size says little about value. With neither a headcount nor a usable revenue it has no valuation. Its page reads "before first round".

The same winding-down probability applies, from the company's headcount trend; the odds that a company with no round closes are in the figures below.

Revenue, from Veridion's company data, is an input to the model only. VCMC never shows a company's revenue; to keep it that way, the revenue multiples on this page are shown as bands and the revenue per employee the model uses is withheld.

Revenue counts only when revenue per employee lies between $20K and $1.5M. When revenue times its sector's multiple is above the valuation, the valuation moves halfway toward it, on a log scale. When it is below and the last round is 3 years old or more, the valuation moves 10% of the way down. After a round priced from its amount, a revenue value more than 2x the valuation also lifts the high end of the range toward it, by at most 2x.

VCMC holds one revenue figure per company, dated when it was collected, so the chart's earlier dates use no revenue.

A company known to be closed is valued at the floor: 1% of its last round's price, or of its value without a round. Its winding-down probability is 100%, its grade is dead, and it has no next round and no predicted next raise. A company with nothing to value it from has no valuation, closed or not.

VCMC does not read website status yet, and it holds no list of companies verified as closed, so today no company is marked down this way: a falling headcount is the only sign of winding down the model reads. When website status is read, it works as follows.

Every figure the model uses, with where it comes from: measured on data, or chosen. A figure marked as tested was picked among several by testing each against past rounds and sale prices. Revenue multiples are shown as bands and the revenue per employee is withheld, so that no company's revenue can be worked out from its valuation.

Each test hides a price that is known, values the company from what was known before it, and compares the valuation with that price. The tables read the same way.

How to read the tables

"Half within" is the factor that half of the valuations came within of the price, either way. "Within 2x" is the share of prices within a factor of two of the valuation. "Price against the valuation (median)" is how far the typical price sat above or below the valuation.

Test 1: the next priced round

We test the model the way it is used. For 998 companies that raised a priced round again, we hid that round and asked the model what the company was worth the day before, from what was known then: only earlier rounds, the headcount on or before that day, and no revenue. The answer is compared with the round's pre-money: its price less the money it raised. Half of VCMC's valuations were within 1.58x of the round's price before the new money, and 68% within a factor of two.

The rounds before 2024 and the rounds from 2024 are shown apart. Companies that raise again tend to price above the valuation: 1.46x above at the median. That is expected, because only companies doing well raise again: the companies in this test are survivors, and a company that never raises again leaves no price to compare with.

Test 2: sale prices

A second test looks at sales: 651 companies acquired after their last round at a disclosed price, each valued the day before the sale. Half of the valuations were within 2.09x of the price and 48% within a factor of two; at the median the price was the same as the valuation. 586 more were acquired at a price that was not disclosed.

Sale prices are a weaker test than new rounds. Recorded sale prices are often flawed: a rumoured price, one investor's proceeds stored as the price, a stake recorded as the whole company. The flaws gather in the cases that decide the result, companies years past their last round. A sale price is also not the value the day before: a buyer pays for control, and the price may be for the business rather than its equity. And a company that closes is never sold, so this test cannot see the companies a valuation should mark down.

Test 3: the price at the round

A third test hides the price of each of 5,266 disclosed rounds and values the company on the round's own date from its amount and stage alone, the way a new round without a disclosed valuation is priced: half were within 1.50x and 73% within a factor of two.

Test 4: companies verified as closed

A fourth test checks 17 companies verified by hand as closed. With no website status, as VCMC runs today, the model holds them at a median 37% of their last round's price, $6.3B together. With their website status, 16 of the 17 go to the floor and together they hold $284M.

The range and the grade

How often the price fell inside the range: 36% of next-round prices, 53% of sale prices and 63% of prices at the round. The range is not built to hold a share of outcomes: its low end is the company winding down and its high end the price of its next round, each widened only by how precisely the starting price is known. So it is narrow, and many real prices fall outside it.

The grade follows the range. On 15 Oct 2026, of the venture-backed companies VCMC values, 1% were graded A, 8% B, 29% C and 61% D. The range widens with the time since the last round:

The homepage shows the newest priced rounds against the valuation the day before.

Companies without a round are valued with no sign of whether they will raise. The industry level applies only to companies that have raised venture money, since nothing in the data tells a young startup from a small business that never will; a startup before its first raise is valued on its size and may be valued too low.

Some companies have no valuation because their headcount readings disagree: when today's headcount and the latest reading in its history differ by more than 3x, the headcount is unknown, and a company with no round and no usable revenue is then left without a valuation.

Companies that raise again price above the valuation, 1.46x above at the median on 998 rounds. The test can only see companies that went on to raise; read it together with the closed companies, where the error runs the other way.

The odds behind the winding-down path come from older data: US companies of 2010 to 2013. They shape only the winding-down path.

A company acquired with no record of the deal in VCMC's funding data is still valued as a private company. A headcount that does not match the company, such as a wrong profile, moves its valuation until it is corrected.

A round with no disclosed valuation is priced at the share its stage usually sells; a round that sold more or less than usual is priced too high or too low. At the round itself this pricing is within a factor of two 73% of the time.