Why near-identical businesses sell for very different prices

A comp-based valuation is a claim about a distribution: businesses like yours recently changed hands somewhere in this band. Report only the middle of that band and you have thrown away the part that tells a buyer or seller what they are actually facing. So our estimates publish the interquartile range — the middle half of comparable sales, from the 25th to the 75th percentile — with the median inside it, rather than a single figure with a confidence note nobody reads.

The correct reading of a wide band is literal. It means comparable businesses genuinely sold at prices that far apart, and that the deal-specific facts decided more of the outcome than the category did. It does not mean the midpoint is your number and the band is an error bar around it.

These are sold prices, not multiples: the 25th percentile, median and 75th percentile of what the comps in each category actually changed hands for, on whichever basis that category can currently support. The last column is simply the top of the middle half divided by the bottom of it — one number for how much room there is between two comparable sales.

Loading live quartiles from /api/stats…

A category appears on the profit basis where it clears the five-comp floor there, and on the revenue basis otherwise; categories clearing neither are listed with no quartiles rather than with filled-in ones. The prices are those of the comps themselves, so this is the spread inside a sample — not a range for any particular business, and not a prediction for yours.

Three things narrow the band before you ever see it, and all three are deliberate:

1. The tails are trimmed. Once a sample reaches twenty comps we drop the top and bottom 5% of multiples before computing anything, because at this end of the market a stray figure is more often a parsing error or a mislabelled headline than a real outlier. Real extreme sales get trimmed alongside the errors.
2. Only comps with a usable multiple count. A sale that states a price but no earnings carries no multiple, so it never reaches these quartiles — and sales with unusual prices are disproportionately the ones that get reported without their numbers.
3. Only published sales exist at all. Auction venues publish closes, brokers mostly do not, private deals never do. Every figure here is a sample of what gets published, not of what gets sold. See where sold-price data comes from.

So treat the quartile gap as a floor on the dispersion rather than a measurement of it. If the middle half of comparable sales already spans a wide band, the full range of outcomes for a business like yours is wider still.

A comp set knows a business's category, size and sale price. Almost everything that decides where inside the band a particular deal lands is invisible to it. We do not hold these as fields and we do not claim to have measured their effect — but they are what the residual spread is made of, and they are the right things to argue about once the band is on the table:

• Transferability. Whether the asset survives leaving its owner — accounts, contracts, supplier relationships and admin access that move cleanly, or don't.
• Concentration. One client, one product, one keyword or one traffic source carrying most of the revenue. Same earnings, different business.
• Platform dependence. Earnings that exist at the discretion of a marketplace, an ad network or a search algorithm are worth less than earnings that do not, and the discount is not uniform across categories.
• Owner hours. A business needing forty hours a week is being priced partly as a job. The stated profit rarely nets that out — see which add-backs survive.
• Trend. The same trailing-twelve-month figure can be a recovery or a decline. Buyers pay for the direction, not the average.
• Terms and competition. All-cash on a short close is not the same price as the same headline with an earnout attached, and a contested auction is not the same process as a private approach. What a headline sale price includes takes this apart.