Comparable Analysis

Pricing a company requires something to calculate from. An old manufacturer has ten years of profit and a set of factories and equipment, and a number can be built out of those. A software company three years into its life has neither. Profit hasn't arrived yet, and what the company owns amounts to code and people.

That leaves projecting future earnings and calculating from those, which stalls almost immediately. Whether you assume revenue doubles or triples by year five changes the answer entirely, and nothing justifies one assumption over the other. So the space between "our company is worth $23 million" and "based on what?" stayed empty in negotiation after negotiation. Whoever spoke with more conviction, or had the least need for this particular deal, tended to win.

Comparable analysis fills that gap with facts collected from outside. Rather than arguing about this company's future, it takes what companies resembling this one actually sold for and uses those prices as the reference. It is the same move as checking what a comparable unit in the same building sold for last month before setting a price. Two parties can forecast differently, but a deal that already closed looks the same to both of them.

The work runs in five steps.

1. Assemble the comparison set. Select companies that resemble yours on business model, whether the customer is an individual or a company, growth stage, and market geography. Five to ten is a normal size.
2. Collect each company's deal price and the revenue figure that price was set against. Public companies have filings; private ones turn up in announcements attached to funding and acquisition news.
3. Compute a multiple for each one: deal price divided by annual recurring revenue (ARR). Annual recurring revenue is subscription income stated on a yearly basis — the portion expected to arrive again next year, as opposed to one-time sales. Pricing a company by applying such a ratio is called working from a multiple.
4. Drop the outliers and take the median. If six companies produce multiples of 6.2, 7.0, 7.4, 8.1, 9.3 and 18.0, treat the 18.0 as a deal with special circumstances, remove it, and take 7.4 as the reference. The median beats the average here because one unusual deal drags an average upward on its own.
5. Apply it to your own figure, then adjust for the differences. Annual recurring revenue of $2.3 million multiplied by 7.4 gives $17 million. Adjust upward if you are growing faster than the comparison set and downward if more of your customers are leaving.

The method fits best exactly where a company has few usable numbers of its own, which is why early-stage startups lean on it. It also behaves well in a negotiation. To push back, the other side has to show that the comparison set is wrong, and doing that means bringing deals of their own. The conversation moves from conviction against conviction to evidence against evidence.

There are three limits.

1. Genuinely similar companies are hard to find. Two collaboration tools can look alike and price completely differently depending on whether they sell to individuals or to companies. What one customer pays, how long contracts run, and what selling costs all differ, so two firms at the same revenue can carry multiples twice as far apart.
2. The sample is small. Private deals often close without disclosed terms, so filtering for a real match can leave five or six data points. Adding or removing one of them moves the median visibly.
3. It mirrors the market. The method transfers whatever price the market is currently assigning, so an overheated market inflates your number and a frozen one deflates it. Citing a deal from 18 months ago means importing a price from a market that no longer exists.

For those reasons nobody walks into a negotiation holding this method alone. It gets run alongside discounted cash flow (DCF), which projects future cash and converts it back to present value, and the resulting figures define a range to argue within. The agreed number goes into the term sheet as a single line, and due diligence then checks whether the revenue behind it is real.

"We set this round's valuation at the median of six comparable deals."

The number rests on recent transactions rather than on the founders' confidence.

"That one sells mostly to enterprise customers, so it's hard to keep in the comparison set."

Different business model, different multiple, so drop it from the sample.

"Last year's deals came out of a different market, so we can't use them as they stand."

Timing changes the multiple even for an otherwise identical company.

As Pagedeck prepared its next round, Grace and the finance lead built a comparison set. Searching for collaboration document tools produced twelve companies; removing the ones selling to individual users and the ones living on large enterprise contracts left five. The median multiple came to 7.4, which against annual recurring revenue put the company at $17 million.

The investors ran the same exercise on their own terms. Adding one company and dropping another pulled the median to 6.6, and that gap became the negotiating distance. Because the difference sat in the sample rather than in the arithmetic, the discussion that followed was about why a particular company belonged in the set.

The most common assumption is that you can lift a similar company's multiple and use it as is. The last step of the procedure is a single multiplication, so it looks that way. But a multiple is not a fixed property of a company; it moves with growth rate and with how many customers leave. A company tripling each year and one sitting still do not both earn a 7.4. Multiply without adjusting and you get a figure that looks well-sourced, right up until diligence surfaces the growth and churn numbers and the whole valuation comes down.

Using public company multiples unmodified is the other frequent error. Public filings are easy to obtain, which makes those companies tempting to include. But public shares can be sold whenever the holder wants, while private shares require finding a buyer. That difference takes a cut out of the price, and the cut grows as the company gets smaller. Skip the discount and your number lands far above where deals actually close, and negotiations built on it usually stop at the first meeting.