How to make an offer on an online business

There is no single correct price for a small online business, and an offer that pretends otherwise is easy to argue with. Comparable sales do not produce a price; they produce a spread, and the spread is what you negotiate inside. Our own estimator never returns one number for exactly this reason — it returns the 25th, 50th and 75th percentile of the multiples that comparable sales actually cleared at.

So the first move is to replace "what is it worth?" with three narrower questions you can answer from evidence: what did similar businesses sell for, is this business similar enough for that to mean anything, and where inside the band does this specific deal belong. The rest of this page is those three questions in order.

An offer band is one multiplication, run three times. Take the business's trailing-twelve-month figure, and multiply it by the low, middle and high multiple of the comparable sales:

offer band = TTM figure × (p25, p50, p75 multiple of comparable sales)

Four details decide whether that band is worth anything, and all four are things you should be able to state out loud before you send a number.

1. Which figure. If the business has positive trailing-twelve-month profit, profit is the basis and profit comps are the comparison set. If it does not, the basis falls back to revenue — a different question with a different answer. Our calculator picks the basis from your own inputs the same way, so which multiple you are quoting is never ambiguous.
2. Twelve months of it. A monthly figure multiplied by twelve assumes a flat year. Where a listing leaves us no alternative we do that arithmetic and grade the row estimated so it stays visible; in your own offer, ask for the twelve monthly figures and see whether the multiple is monthly or annual before you multiply anything.
3. Enough comparable sales. Five is our floor: under five usable comps we publish nothing rather than a number one more deal would move. Once a sample reaches twenty we trim the extreme 5% at each end before taking percentiles, because a single 40× outlier otherwise sets the top of your band. How many comps a valuation needs goes through the floor in detail.
4. Recent sales, named venues. A band assembled from four-year-old deals on a marketplace you have never heard of is not the market you are buying in. Ask of any band — ours included — how many sales, over what dates, from which venues.

This is the check most offers skip. A band built from comparable sales only transfers to your deal if your deal is in the same weight class as those sales. Our report flags a size mismatch when the subject's own figure is four times larger or four times smaller than the median comp's — past that ratio the percentile band is describing a different market, and you should treat it as background rather than as evidence.

The table below is the input to that check: for each asset type we hold, the size of the median comparable sale and the middle half of what those comps sold for. Find your asset type, compare its median comp size against the business in front of you, and only then read the price band.

"Basis shown" is the profit comp set where we publish one, and the revenue set otherwise; your own figures decide which set your deal is actually priced against. "Comps used" is the count left after the outlier trim, drawn from the most recent 200 sales we hold for that type. The band is the 25th to 75th percentile of those comps' sold prices — the middle half, not the full range.

The band's midpoint is the default, and everything that follows is an argument for moving off it. Each of these is checkable before you send an offer, which is what makes it usable as a reason rather than as a feeling.

The long forms: why near-identical businesses sell for very different prices, which add-backs survive, and the checklist for sub-$100k deals.

1. Read your offer back as a payback period. Divide the number by the annual profit you believe. If the answer is a number of years you would not personally underwrite at this level of verification, the multiple is wrong regardless of what the comps say. The payback framing is the same arithmetic from the other end.
2. Add the costs that are not the price. Escrow and marketplace fees, migration, the first months of your own time, and any tooling the seller was getting for free. The comparable sales are prices, not total costs of ownership; your budget is the second thing.
3. Ignore the ask, and ignore the bid. An asking price is the seller's opening position, and an auction's unmet-reserve high bid is nobody's purchase. Neither belongs in your arithmetic — asking price vs sold price sets out why both mislead in opposite directions.