Recurring-Revenue Loans Explained: How Software Buyouts Get Levered on ARR When There Is No EBITDA to Lend Against — the Net Revenue Retention Test, and the Covenant That Converts to Leverage the Day Growth Stops

Software Buyouts Break the Rule That Leverage Is a Multiple of Earnings

Ask how much debt a company can carry and the reflex answer is a multiple of EBITDA — five, six, sometimes seven turns of earnings, the arithmetic that sits under every LBO model. Point that arithmetic at a high-growth software company and it returns almost nothing. The company might have £40M of annual recurring revenue growing 30% a year at an 80% gross margin, and still report EBITDA close to zero — not because it cannot make money, but because it chooses to spend every pound of gross profit on sales and marketing to capture a market before a competitor does. Five times an EBITDA of nothing is a loan of nothing. On the earnings test, the business is unleverageable.

Yet software has been close to a third of global buyout value in recent years (approximate, and it swings with the cycle), and sponsors like Thoma Bravo, Vista and Hg have built firms out of buying exactly these companies. They finance them with a different instrument. Rather than lend against profit the target does not report, the lender lends against the revenue it does — the recurring-revenue loan, sized as a multiple of ARR, underwritten on the durability of that revenue rather than on earnings that have not arrived yet. The whole product turns on one question: when is revenue safe enough to lend against as if it were profit?

The Loan Is Sized Against ARR, and the Multiple Looks Small Until You Convert It

A recurring-revenue facility is quoted against ARR, not EBITDA. Leverage is expressed as debt divided by annual recurring revenue — commonly somewhere around 0.5x to 1.5x ARR on the senior piece, wider on a stretched unitranche (approximate; it moves with retention, growth and gross margin). Written that way, the number looks tame. A loan of 1.0x revenue against a company growing 30% a year sounds like almost no leverage at all, and next to a business trading at 12x ARR in the equity market it barely registers.

That framing is the first thing to distrust. A revenue multiple and an earnings multiple are not comparable, and the gap between them is the entire risk of the instrument. Take the same £40M-ARR company at 1.0x ARR — a £40M loan. If the business one day runs at a 20% EBITDA margin, that is £8M of earnings and the loan is 5x EBITDA: ordinary. If it only ever reaches a 10% margin — £4M of earnings — the same £40M loan is 10x EBITDA: aggressive. And if the margin never comes at all, the EBITDA multiple is infinite. The revenue multiple did not change. What changed is the profit the revenue was supposed to become. The lender is not really lending 1x revenue; it is lending an EBITDA multiple it cannot yet see, and betting on the margin that resolves it.

Net Revenue Retention Is the Number the Whole Loan Rests On

If the collateral is revenue, the underwriting question is how much of that revenue survives. The metric that answers it is net revenue retention — the revenue this year from the customers you already had last year, after churn, downgrades and expansion, expressed against what those same customers paid a year ago. Above 100%, the installed base grows without a single new customer: existing accounts expand — more seats, more usage, more modules — faster than others leave. Below 100%, the base leaks, and every new sale is running to stand still.

NRR is what lets a lender treat software revenue as bond-like rather than forecast-like. Contracted, subscription-based, high-gross-margin revenue that renews at 110%+ is not a projection; it is closer to an annuity that grows. A lender underwriting a base with 120% NRR knows that even if the company signs no one new and simply defends what it has, the revenue backing the loan rises. That is a fundamentally different risk from lending against next year's bookings. It is why the recurring-revenue loan exists at all — and why NRR, not growth, is the first covenant in the package. Best-in-class SaaS ran net retention comfortably above 120% in 2021; the median public software company sat above 110%, and both compressed toward 105–110% through the 2023–24 slowdown as expansion stalled and churn crept up (approximate, and the direction matters more than the decimal).

A Worked Facility: £40M of ARR Debt on a Company With No EBITDA

Put the pieces together on one deal. A sponsor buys a vertical-software company: £40M ARR, growing 30% a year, 80% gross margin, 118% net revenue retention, and EBITDA of roughly zero because every pound of gross profit funds new-customer acquisition. No EBITDA lender will touch it. A recurring-revenue lender sizes a facility instead.

The elegance is that the loan tracks the company's own maturation: it is underwritten on revenue durability while the business is still spending to grow, and on earnings once the business is meant to have stopped. The danger is hidden in the same mechanism, and it is worth its own section.

The Conversion Covenant Is the Cliff the Whole Deal Is Underwritten Toward

The flip from an ARR test to an EBITDA test is not a formality — it is the moment the deal is judged. On the day it converts, the £40M loan stops being measured against revenue and starts being measured against profit, and the company must clear the 6.5x leverage test to avoid a default. Whether it clears depends entirely on a margin that did not exist when the loan was written.

Run the two outcomes. Suppose ARR has grown to £70M by year three. If the company now runs at a 20% EBITDA margin, that is £14M of earnings, and £40M of debt is 2.9x — it clears comfortably, and the sponsor refinances into ordinary leverage. If instead growth pulled margins down and it reaches only a 10% margin — £7M of EBITDA — the same £40M loan is 5.7x, right up against the covenant, with no room for a bad quarter. And if growth stalled and the promised operating leverage never materialised, the company hits the conversion date levered at a multiple of EBITDA that trips the test on the day it applies. The company did not fail; it simply arrived at the covenant that assumed it would have matured, and had not. The recurring-revenue loan does not remove the leverage risk of a software buyout. It defers it to a fixed date and dresses it, until then, as a revenue multiple.

Who Lends It — and Why SVB's Collapse Handed the Market to Private Credit

Recurring-revenue and venture lending was, for two decades, the near-private preserve of Silicon Valley Bank, which built the playbook for lending against ARR and cash runway to companies no cash-flow lender would consider. When SVB collapsed in March 2023 — the second-largest bank failure in US history at the time — the single largest provider of this credit vanished in a weekend. The demand did not. It moved, almost wholesale, into private credit: direct-lending funds and specialist software-credit arms — Blue Owl, Golub, Ares, Vista's own credit business, Hercules and others — stepped into the gap, often delivering the facility as a unitranche they hold to maturity rather than syndicate.

That shift is not incidental to the product; it fits it. A recurring-revenue loan is bespoke, covenant-heavy, and underwritten on metrics — NRR, gross churn, magic number, the Rule of 40 — that a syndicated market prices badly and a single relationship lender prices well. The lender that will hold the paper is the lender that will do the diligence on retention cohort by cohort. So the instrument that finances the most modern corner of the buyout market is delivered, increasingly, by the least liquid corner of the debt market. The collateral is a software subscription base; the counterparty is a fund that intends to be there when the covenant converts.

The Verdict: A Loan Against Revenue Is a Bet on the Margin That Revenue Becomes

The recurring-revenue loan solves a real problem cleanly. A company can be genuinely valuable, genuinely fundable, and report no EBITDA because it is rationally spending its gross profit to grow — and lending against durable, contracted, high-retention revenue is a defensible way to finance it. Where the metric is honest — NRR comfortably above 100%, gross margins in the 80s, churn low and cohorts stable — the ARR the loan sits on is closer to an annuity than a forecast, and the modest revenue multiple is what it appears to be.

Where it is not honest, the same structure hides the leverage rather than removing it. A 1x-ARR loan on a company that never reaches a real margin is not conservative debt; it is a high EBITDA multiple wearing a low revenue multiple, with the reckoning fixed for the conversion date. For a student, the discipline is the one this site keeps returning to: refuse the flattering number and name the structure underneath it. The question a lender asks is not "how fast is it growing" but "how little of this revenue leaves, and what margin does it become when the growth is supposed to stop." Answer that and you have understood the instrument. Miss it and you have mistaken a deferred leverage bet for a revenue-backed one.

Take Your Preparation Further

A recurring-revenue loan only makes sense against the leverage it replaces. See how debt capacity is normally sized off earnings in the beginner's LBO guide; where the facility sits relative to senior and junior debt in the LBO debt stack; why the covenant package matters and how a maintenance test bites in covenant-lite explained; who holds this paper and why in private credit and direct lending explained; and how the same lenders deliver it as one instrument in unitranche financing explained.

To build the debt schedule, the covenant tests and the conversion mechanics inside a live deal model, download the LBO Model Template — PE Ready, and for the financing and valuation mechanics on one page, see the free Valuation Methods Cheat Sheet.

Ready for personalised feedback? Book a 1-on-1 mentoring session with an experienced IB/PE professional.