What is a business with no sold comps worth?

Loading live coverage from /api/stats…

The two comp columns count only sales inside the window a report is built from — the most recent block of deals for that type — that published enough for a multiple to exist on that basis. The gap between those columns and the rows column is the whole subject of this page. Where a basis holds fewer comps than our floor we publish nothing on it, for the reasons how many sold comps a valuation needs sets out.

“There are no comps for my business” describes three unrelated situations. Treating them the same is how people end up borrowing a number from a market that has nothing to do with theirs.

a. The category is too thin to say anything

A handful of sales exist, but not enough for a median to mean anything: one unusual deal would move it, and you would have no way to tell that it had. This is the situation our floor is built for, and it is why some rows in the table above report a count and still support no figure. The remedy is to widen deliberately and say you did — a broader asset class with real comps, read as a rough band rather than a valuation, beats a narrow one with two sales in it.

b. There are plenty of sales, but no earnings to divide by

This is the largest single case in our record rather than an edge case, and it is not thinness at all: sold prices exist in quantity, and almost none of them arrive with a profit or revenue figure attached. A multiple is simply not computable over them. The remedy here is not to find better comps — the comps are fine — but to stop asking for a multiple and read the price distribution directly, which is how the domain market has always been priced. What is a domain name worth is that case in full, and what a profit multiple actually means covers when the tool applies at all.

c. The comps exist but none of them are your deal

The category clears the floor, the multiples are published, and every sale behind them is four times your size, or five years old, or monetised a different way. A comp set you do not fit is worse than an empty one, because it looks like evidence. What makes a sold comp comparable is the four-way test, how recent a comp needs to be is the date half, and why near-identical businesses sell for different prices is what remains unexplained even after a good match.

Nothing on this list is as good as a matched set of recent sales. They are ordered by how much weight each can carry, and each entry says how it fails, because a substitute you cannot criticise is one you have stopped examining.

1. Sold prices without multiples. If sales of the asset class are published but earnings are not, the distribution of prices is still evidence — the middle half of it tells you what the market pays for that kind of thing, independent of any earnings claim. How it fails: only sales worth writing up get written up, so a published price distribution skews high and says nothing about the ordinary end of the market.
2. An adjacent category, quoted as an adjustment. A comp set one step out — same monetisation, different niche; same size band, different asset type — can anchor a band if you state the step and the direction you are adjusting for. How it fails: the adjustment is a judgement with no evidence behind it, so it can absorb any conclusion you want. Write the size of it down before you compute anything, not after you see the answer.
3. The cost to rebuild it. What it would take in money and months to get to the same traffic, code, inventory or list is a genuine floor-ish reference for a buyer who has the option of building instead. How it fails: it is a cost, not a value. Plenty of assets cost more to build than anyone will ever pay for them, which is exactly the situation an unprofitable business is usually in.
4. The payback period you would personally underwrite. Price divided by the monthly earnings you actually believe is a number you can feel, and at this size it is a better instrument than a multiple because it forces you to name the risk horizon. How it fails: it prices your tolerance, not the market's — a seller is under no obligation to accept your discount rate. Payback period is the arithmetic.
5. Asking prices — last, and only as a ceiling. If asks are all that exists, they still bound the problem from above: an asset rarely sells for more than similar assets are openly listed at. How it fails: an ask is a number one party chose, including on the listings that will never sell. Asking price vs sold price is why they keep arriving dressed as evidence, and averaging them produces the most confident wrong number available.

With no comp set the goal changes. You are not producing an estimate that can be defended as accurate; you are producing a range whose reasoning someone else can inspect and argue with. That is a lower bar and a far more useful artefact. Four lines, in writing, before you say a number out loud:

• What the number is built from. Name the substitute — “six auction sales of similar names”, “the adjacent category, adjusted down for size” — and the count behind it.
• The band, not the point. A low and a high you would each defend, with the reason they are that far apart. A single figure claims a precision no substitute can support.
• What would move it. The one or two facts that, if verified, would narrow the band — usually an earnings claim you can check rather than take.
• What you could not check. Written down as a list, because it is the part that decides the structure in the next section.

A band presented this way survives a negotiation that a bare figure does not. “This is priced against six sales, here they are, and here is why I adjusted down” is a position the other side has to engage with on the evidence. How to make an offer is the mechanics of doing that without anchoring yourself.

When the evidence is thin, the instinct is to handle it by cutting the price. That is the crude tool, and it fails in both directions: too big a cut kills a fair deal, too small a cut leaves you carrying a risk you never agreed to. The better instrument is the shape of the transaction, because it can be made to pay out on the facts you could not verify.

• A holdback against the claims you could not check. Part of the price released once the thing that was asserted has been observed for a period.
• An earnout tied to the specific unverifiable figure, not to a vague notion of performance — and written so that both sides read the same number off the same dashboard.
• A staged handover, with assets in the buyer's control before funds release, and an explicit account-by-account list of what “assets” means.
• The claims you are paying for, written as warranties. If an earnings figure is the reason for the price, it belongs in the agreement rather than only in the listing.