Startup valuation multiples: what revenue multiple is fair?

A fair revenue multiple for a small software startup is usually a few times annual revenue, and often less than two times for tiny, owner-run products. Profit-based pricing is more common under about $1M in revenue, and larger companies move to revenue multiples that depend heavily on growth and retention. The headline number matters less than what sits behind it.

This is general education, not financial advice. Real valuations depend on a business's details and should be reviewed by a qualified professional.

Why multiples differ so much

Public SaaS stocks and tiny bootstrapped tools are different animals. A public company with thousands of customers, a management team and predictable growth gets a very different price from a one-person tool with 300 subscribers. Mixing the two leads buyers to either overpay or insult sellers.

Three questions drive almost every multiple:

1. How stable is the revenue? Recurring, low-churn revenue is worth more than one-off sales.
2. How fast is it growing? Growth is rewarded, shrinking revenue is punished.
3. How much does it depend on the owner? A business that runs without the founder is worth more.

Two ways to price a small software business

Seller's discretionary earnings (SDE)

SDE is the business's profit plus the owner's salary and personal perks. It answers "how much does one working owner take home?" Most deals below roughly $1M in annual recurring revenue are priced this way. FE International's 2026 guide on SaaS valuation multiples puts private SaaS under $1M ARR at about 2.5x to 4x SDE, which it also describes as roughly 2x to 3x ARR.

Annual recurring revenue (ARR)

ARR is the yearly value of your recurring subscriptions. Once a company has a team and the owner's pay is no longer the main profit driver, buyers price on ARR. The same guide gives about 3.5x to 5.5x ARR for companies between $5M and $10M ARR, and higher ranges for larger ones. Companies above $10M often get priced on both ARR and EBITDA, which is profit before interest, taxes, depreciation and amortization.

A quick reference table

Business size	Typical basis	Range reported by FE International
Under $1M ARR	SDE	2.5x to 4x SDE (about 2x to 3x ARR)
$5M to $10M ARR	ARR	3.5x to 5.5x ARR
Above $20M ARR	ARR	5x to 7x ARR, higher with strong retention and growth

Treat the table as a map of the territory, not a quote. Marketplace data adds more color. Different platforms report different averages, and sample sizes vary, so one source's median may not apply to your listing.

What pushes a multiple up

FE International lists several factors that raise what buyers will pay. The ones that matter most for small deals are:

• Strong net revenue retention. Customers who stay and spend more over time are the best sign of a healthy product. Figures above 100 percent are a good sign, and well above that is better still.
• Healthy gross margin. Software margins above about 75 percent are typical for good products, with services revenue separated from software revenue.
• Low customer concentration. No single customer should be a large share of revenue.
• Owner independence. Documented processes and a small support load make the business easier to run.
• Financials that tie to billing. If the books reconcile with payment data, buyers trust the numbers.
• Several interested buyers. A competitive process tends to lift price.

What pulls a multiple down

• Churn hidden by new sales. If growth only comes from replacing lost customers, the true value is lower.
• Heavy dependence on the founder. Relationships, code knowledge or support held by one person is a risk.
• Customer concentration. One large customer leaving can wipe out profit.
• Rising costs. For AI products, inference and API costs that grow faster than revenue are a warning sign.
• Thin differentiation. AI wrappers with shallow workflows and little proprietary data are easier to copy and often priced lower.
• Unverifiable books. Cash-basis records that cannot be checked quickly make buyers cautious.

How to test whether an asking price is fair

Step 1. Rebuild the profit yourself

Do not accept the listing's figure. Take trailing twelve months of revenue from the payment provider, subtract refunds and fees, then subtract real costs: hosting, AI and API usage, tools, contractors and support. Add back only the owner's pay and clearly personal expenses, and ask for proof of each add-back.

Step 2. Apply a range, not a single number

Take your rebuilt SDE or ARR and multiply by the low and high ends for the right size band. If the asking price sits above the top of the range, the seller needs a clear reason such as unusual growth or very low churn.

Step 3. Adjust for risk

Make a short list of risks and shade the price accordingly. Ask yourself:

• What would happen if the top customer left?
• What if the underlying AI model provider doubled its prices?
• What if the founder disappeared after closing?

Step 4. Check the payback period

Divide the price by annual profit. If the answer is four years, you are betting the business holds steady for four years just to get your money back. For fast-moving AI categories, many buyers want a shorter payback.

Step 5. Compare with real peers

Look at what similar products charge and how much traffic they get. The traffic checker models monthly visits for any domain, and the revenue rankings show how modelled revenue compares across AI startups. These are estimates built from public popularity lists, so use them for context rather than proof. Our methodology page explains exactly how they are built.

Why revenue multiples alone can mislead

A revenue multiple ignores cost. Two startups with the same $200,000 in annual revenue can have very different profit. One might keep $150,000 and the other $20,000 after paying for compute and staff. Paying the same multiple of revenue for both makes no sense.

Revenue multiples also hide quality. A dollar of annual contract revenue from a customer who has renewed for three years is worth more than a dollar from a customer in a one-month trial. Always ask what sits inside the revenue figure.

Special notes for AI startups

AI businesses add three extra variables.

• Model dependence. If the product relies on a single provider's API, price changes or policy changes can hit margins without warning.
• Cost per unit. Check how much each customer's usage costs. A flat monthly plan with heavy usage can run at a loss.
• Defensibility. Ask what stops a competitor from copying the product with the same model. Proprietary data, integrations and distribution are stronger moats than a clever prompt.

Many AI tools are very young, which means a short revenue history. Price that uncertainty in rather than assuming last quarter's growth will continue.

Practical negotiation tips

• Start from your own rebuilt numbers and show your working. Sellers respond better to a logical offer than to a lowball.
• Trade price for structure. A lower upfront price with an earn-out tied to retention can bridge a gap. See financing a startup acquisition.
• Ask for a short transition period to be included in the price.
• Walk away when the numbers do not tie out. A cheap price is not a bargain if the revenue is not real. See red flags in startup listings.

If you want to browse deals where revenue is verified from the payment provider, start at the Buy a startup marketplace.

FAQ

What is a good revenue multiple for a small SaaS?

For very small, owner-run products, revenue multiples are often in the low single digits and many are priced on profit instead. As revenue, growth and retention improve, buyers pay more. Use a published range for your size band and adjust for risk.

Should I value a startup on revenue or profit?

For small businesses, profit (SDE) is usually the better guide because it shows what an owner actually earns. For larger or fast-growing companies, buyers lean on ARR and growth.

Do AI startups get higher multiples than other software?

Not automatically. Buyers pay more for AI products with proprietary data, deep workflow integration and healthy margins. They pay less for thin wrappers around a third-party model that can be copied quickly.

Are marketplace averages reliable?

They are useful as a rough guide, but samples vary and listing prices are not always final sale prices. Compare several sources and treat averages as a range.