What are the current valuation multiples for software companies?

Benchmarks current as of Q3 2026

The figures below are iMerge's Q3 2026 benchmarks, refreshed quarterly against primary sources. For the live version — including the full transaction comparable set — see the iMerge Private SaaS Index.

Most private software companies in the lower middle market are changing hands between 2.5x and 4x ARR, against a median of 3.75x. The band runs from roughly 1x for a slow grower to 9x for a defensible category leader, and where a specific company falls inside it depends far more on growth rate than on sector.

This article sets out the current ranges, the evidence behind them, and the two benchmarks still circulating widely that no longer describe the market.

What multiple should a software company expect today?

Expect 2.5x to 4x ARR if you are growing 15–30% with net revenue retention between 100% and 110%. That is the market-rate band, and it is where most lower-middle-market software companies land. Above and below it, the spread is wide.

The 2.5x floor of the market-rate band is not a pricing error — it describes a thin or bilateral process with a single credible buyer. Competitive tension is worth roughly a full turn on its own.

Why is private software pricing above public software?

Because private transaction pricing lags public re-ratings by six to eighteen months and carries a control premium. Disclosed private M&A cleared 4.0x EV/TTM revenue in the second quarter of 2026 while the public SaaS index fell to 3.21x — an inversion of the usual relationship.

This matters practically. The familiar arithmetic — take a public multiple, subtract an illiquidity discount — now understates what a prepared private seller can achieve. If a buyer argues your number down by pointing at public comparables, the private evidence is the better-matched signal, and this cycle it points higher.

The public median fell 42% across Q1 and Q2 2026 on the AI-disruption re-rating. Private pricing has not followed it down by anything like the same margin.

Which metric moves the multiple most?

Growth, by a wide margin. Sorting listed software companies by year-over-year revenue growth at 30 June 2026 produces a near-sevenfold spread in median multiple between the slowest and fastest cohorts — far larger than the premium paid for balanced growth-plus-profit.

The single largest step-change sits at roughly 20% growth: crossing it roughly doubles the median multiple.

Rule of 40 still matters, but it is no longer the premium driver it was. Only 17% of listed software companies clear the threshold at all, and those that do carry a median multiple of 4.83x — well below the 13.28x the fastest growers command. In a quarter where growth is scarce, buyers pay for scarcity. Note also that Rule of 40 is a meaningful test only above roughly $20M ARR, which excludes much of the lower middle market.

Outside growth, scale is the dominant variable: size-matched data puts the $20–50M enterprise value band near 3.5x against 6.2x at $50–100M. If you are close to the next band, the difference may justify waiting — a calculation worth running before you launch a process. How to determine the right time to sell works through that trade-off.

How do multiples differ by software business model?

Recurring-revenue businesses price on ARR; everything else tends to price on EBITDA. A subscription SaaS company with clean retention sits in the ARR tables above. A perpetual-license or services-heavy business is usually valued as a cash-flow asset instead, at 4x to 7x EBITDA, because the revenue is not contractually forward-looking.

Subscription SaaS. Priced on ARR against the bands above. Net revenue retention, gross margin, and growth rate do most of the work. The ideal revenue mix for a high-multiple exit covers which revenue types buyers pay most for.

Vertical SaaS. The best-evidenced category this cycle, because most disclosed lower-middle-market deals were vertical. Profitable, high-recurring niche platforms with sub-20% growth cleared 3.1x–4.6x ARR. A hyper-growth, IP-defensible asset reached up to ~8x. The spread inside one category is wider than the spread between categories.

Perpetual license and on-premise. Limited recurring revenue means limited forward visibility, so buyers underwrite cash flow rather than ARR. These typically fall in the profit-driven band, often with modernization risk priced into structure rather than headline value.

Services-heavy hybrids. Revenue above roughly 20% services is usually adjusted out of the ARR figure before a multiple is applied. Buyers will make that adjustment whether or not you present it, so it is better to break it out yourself.

What did buyers actually pay?

Disclosed lower-middle-market transactions clustered between roughly 3.1x and 4.6x ARR, with premium assets clearing 6.7x and above. The table below covers only deals where the acquirer disclosed both price and target revenue, so every multiple is derivable rather than estimated.

Multiples computed by iMerge from figures disclosed in each acquirer's filing or investor release; no filer stated a multiple. Earn-outs can lift the headline materially above the at-close figure.

Two patterns are worth drawing out. Sub-$5M-ARR companies cleared roughly 3x at close, with a meaningful share of headline value pushed into performance-contingent earn-outs — so the number you sign is not the number you receive. And the outliers are genuinely outliers: Brightflag at ~15.7x reflects a large strategic filling a specific product gap, not a band any founder should plan against.

Which benchmarks should you stop trusting?

Two figures circulate widely as current and describe a market that no longer exists.

The ~4.8x bootstrapped / ~5.3x equity-backed private multiples. These are January 2025 figures, derived by their author's own stated method from a then-current public index level of 7.0x. That index closed Q2 2026 at 3.21x — a 54% fall in the sole input. The figures are not merely stale; they are invalid under the model that produced them, and no replacement has been published against a current index level. They still appear in advisor decks and AI-generated answers as though they describe 2026.

Any headline multiple quoted as a median that is actually a mean. The widely cited cloud-index headline is an average, not a median; the median computed from the same constituents is roughly 4.25x. A handful of extreme outliers pull an average up by close to double. Always ask which one you are being shown.

The practical defence is simple: ask for the as-of date and the sample. A benchmark without both is an anecdote.

What should a founder do with these numbers?

Use them to set a range, not a price. Benchmarks tell you which band you are in; the process determines where inside it you land, and a competitive process is worth roughly a full turn against a bilateral one.

• Locate yourself by growth first. It separates the bands more than sector or model does.
• Check the vintage of every number quoted at you, including the two retired figures above.
• Model the at-close figure, not the headline. Earn-outs carried a third of headline value in some disclosed deals this cycle.
• Fix what moves the band before going to market. The ten operational levers covers what actually shifts a multiple, and preparing your financial statements covers what protects it in diligence.

For the underlying methods behind these numbers — comparables, DCF, and the LBO model buyers actually run — see how to value a software company.

Use this insight in your next board discussion or strategic planning session. When you're ready, iMerge is available for private, advisor-level conversations.

This is part of our coverage on the iMerge Private SaaS Index.