Seller's discretionary earnings, and which add-backs survive

SDE = net profit + owner's compensation + non-transferring costs

Seller's discretionary earnings is what one full-time owner-operator gets out of the business in a year, before that owner decides how to pay themselves. You start at net profit, add back what the owner took as salary or distributions, and add back costs that exist because of this owner rather than because of the business — a personal phone billed to the company, a one-off legal fee for setting the entity up, the accountant's fee for a return the buyer will file differently.

Worked through, on a business whose books show $30,000 of net profit after paying its owner $40,000 and after a $5,000 one-time rebuild of the site: SDE is $75,000. Nothing about the business changed between those two numbers — only the question being asked. Net profit answers "what did this return on top of a paid operator?"; SDE answers "what would this return to me if I were the operator?"

That is a legitimate question, and at this deal size it is usually the right one: the buyer of a $60,000 content site is buying a job as much as an asset. The problem is not the concept. It is that the add-back list is written by the party being paid, from books nobody has audited, and every dollar added to the denominator is multiplied on its way to the price.

One test settles most of them: will the buyer have to spend this money to keep the business running as it is? If yes, it is a cost, whatever it is called. If no, it is an add-back. Apply that honestly and the usual list sorts itself.

Usually survives

• The owner's own pay, when the buyer intends to do the work themselves. Add back a salary, not a salary plus the contractor who actually did the job.
• Genuinely one-off spend — a redesign, a migration, an entity setup, a single legal matter now closed — where there is an invoice and a date, and it does not reappear in the prior year's books.
• Personal costs run through the business: a phone, a car, a family member on payroll for no role, a conference that was a holiday. Each needs a receipt trail, not a claim.
• Interest on debt that does not transfer, and any owner loan being repaid out of the business.
• Depreciation and amortization on assets the buyer is not going to have to replace during the payback period. If they will, it is a real future cost wearing an accounting label.

Usually does not survive

• Work the buyer must replace. A virtual assistant, an editor, a support contractor, an agency retainer — if the business needs the hours, adding them back prices a business that does not exist.
• Spend cut just before the sale. Ads paused, content commissioning stopped, tooling downgraded. Trailing profit rises, the business quietly starts shrinking, and the buyer inherits both.
• "Non-recurring" costs in their third year. Something that has appeared in each of the last three annual accounts is recurring, and the word for it is a cost.
• A second owner's labour added back at zero. SDE is one operator. Two people running it means the buyer hires, and the hire is a cost.
• Founder equity, unpaid time, or "what I would have earned elsewhere." Opportunity cost is not an expense the business incurred.
• Anything with no invoice. An add-back you cannot document is a request to take the seller's word for part of the price.

Buyers argue about multiples and accept the earnings figure. The arithmetic says that is backwards. Price is multiple × earnings, so a percentage error in either moves the price by the same percentage — but the multiple is negotiated in the open and the earnings figure arrives pre-computed in a spreadsheet.

Take a business with $50,000 of defensible SDE, priced at 2.5× — $125,000. Now suppose $10,000 of the add-backs do not survive: a contractor the buyer will have to rehire. Real SDE is $40,000, the honest price at the same multiple is $100,000, and the buyer who negotiated the multiple from 2.7× to 2.5× congratulated themselves on saving $10,000 while overpaying by $25,000. Contesting one add-back was worth more than the entire multiple negotiation.

The same arithmetic runs in reverse for sellers. An add-back you can evidence is worth its face value times the multiple; an add-back you cannot evidence, discovered mid-diligence, costs you the buyer's trust in every other number you gave them.

We record the profit figure the source page states, and we do not normalize it. A Flippa listing that reports "profit" almost never says whether it is net of the seller's own time, and no small-marketplace listing we have seen publishes an add-back schedule. So our multiples are computed against stated profit, which at this deal size is usually something SDE-shaped and occasionally something else entirely.

That is a limit we publish rather than paper over. The table below is the honest version of our coverage: how many sold deals we hold per category, and how many of them state any profit figure at all — the sample every median is computed from.

Where the count of deals stating a profit figure is under five, we publish no median at all. Categories like domain names mostly change hands with no operating business attached, so there is no earnings figure to add anything back to — see what is a domain name worth.