Thesis. The recurring revenue a founder presents and the recurring revenue a buyer will pay for are two different numbers. The difference is found during exclusivity, by the buyer’s accountants, at the moment the founder has nowhere else to go. Every restated dollar leaves the price several times over. The founders who keep their price are the ones who found the difference first, while it was still theirs to explain.

01. Week Seven

Helen knew her numbers. She had built them, invoice by invoice, for eleven years. Her software managed rostering and compliance for aged-care providers across Australia and New Zealand. Revenue was $16.1 million, of which $14.2 million was recurring. Growth was 18%. Churn was low, renewals were steady, and the offer, when it came from a North American acquirer that had bought more than a dozen companies like hers, valued the business at five and a half times ARR, its annual recurring revenue. It was one of the numbers software buyers use to begin the argument about value. The multiple was the part of the letter Helen remembered between readings. It was the part written to be remembered. She considered the price fair, said so, and signed the letter of intent.

Exclusivity was granted the same week. Sixty days, extended once. The diligence lists arrived within days, hundreds of items deep, and Helen answered them at night after running the company all day. She was tired, but the process felt like progress. The buyer was responsive and warm. Nothing suggested a problem.

The call came in week seven, late on a Tuesday. On the screen was a partner from the accounting firm the buyer had retained, and behind him a spreadsheet Helen had never seen but recognised instantly. It was her revenue. Every customer, every invoice, every renewal, rebuilt from the raw records her own team had uploaded to the data room. The partner was courteous, even apologetic. We get to a different number, he said.

Their number was $12.6 million.

Nothing in the gap was dishonest. Onboarding and configuration fees, invoiced once in a customer’s first year and never again, had remained inside the ARR schedule since 2019: six hundred thousand dollars. First-year discounts, granted in side letters while the contracts showed list price, had been counted at the full rate: four hundred thousand. And two enterprise customers who had threatened to leave the previous winter had been kept with price reductions, while the ARR schedule still carried them at the old rate: six hundred thousand more.

The $1.6 million was real money. It simply was not recurring. At the multiple in the letter of intent, the difference was worth almost $9 million. And the buyer had found it first.

02. Two Versions of One Number

ARR is often the most important number in a software sale, and one of the least standardised. Reported revenue answers to accounting standards and, where the accounts are audited, to an auditor. ARR answers to the company’s own definition, and the definition drifts. It drifts in small, innocent decisions made years apart: a services invoice routed through the subscription line because the billing system made it easy, a discount recorded gross because the contract showed list price, a rescued renewal counted as retention because the

customer, after all, had stayed.

No single decision is a lie. Each is defensible on the day it is made. But the decisions are rarely returned to and reconciled, and the drift rarely runs in the direction that makes a company look less recurring than it is. After a decade, the reported number and the rebuilt number are different numbers. The founder is usually the last person able to see the gap, for the same reason nobody can proofread their own writing. She wrote it.

Buyers know this, and so they do not read the number. They rebuild it. The exercise may sit inside a quality of earnings review, alongside the buyer’s work on normalised earnings and working capital, or it may be commissioned separately as a revenue-quality analysis. Either way, it works from the bottom up: customer by customer, contract by contract, separating recurring revenue from implementation work, netting concessions, and testing whether a retained logo is also retained revenue. The output is the buyer’s version of ARR, and it becomes the version against which the final price is tested.

The exercise is also getting heavier. Software Equity Group’s 2026 research among more than 200 acquirers and investors describes a market in which a small tier of premium assets still attracts competition, while the broader field faces longer processes, deeper diligence and wider valuation outcomes. The rebuild is no longer a formality performed near the end. It is increasingly where the transaction is decided, and it runs on the seller’s own records.

The point of the rebuild is not usually to catch fraud. It is to find drift, and drift begins in classification decisions so small that nobody thought them worth recording. That is what makes this failure so democratic: it does not select for careless founders. It selects for founders who defined their own numbers, which is all of them.

ARR is not an accounting standard. It is a claim.

03. How $1.6 Million Became $9 Million

Here is the arithmetic that founders discover late. When a buyer restates a dollar of recurring revenue, the seller does not lose a dollar. She loses the dollar multiplied by whatever the deal is priced at, and nobody in the room needs to negotiate for that to happen. Helen’s $1.6 million restatement, at five and a half times, took almost $9 million off the value of the business on its own. The money did not vanish from the company. It vanished from the claim.

Then comes the second cost, which is larger and harder to see. The first restated number changes how every other number is read. A buyer who has caught the revenue line being generous does not conclude that the revenue line alone was generous. He re-tests the churn calculation, the pipeline, the margin, the forecast. Diligence extends. The question lists lengthen. The tone cools. What was being verified is now being audited, and everything the founder says for the rest of the process is priced with a discount for the possibility that it, too, is a little generous.

04. A Room With One Door

What the buyer’s accountants found mattered less than where Helen was standing when they found it. Had the same three findings surfaced two years earlier, inside her own review, they would have been housekeeping: reclassify the services revenue, restate the schedule, note the concessions, carry on. Surfaced in week seven of exclusivity, they were something else entirely. They were the price, moving.

Readers of this series met the shut-down move in Issue 16, the request for exclusivity that closes the field before a market can form. Daniel granted it to his first and only buyer and never learned what a second one would have paid. Helen’s version was quieter. She had other interest, early and unformed, and she stood it down herself when she signed, because exclusivity was asked for and refusing felt like bad faith.

A restatement does not just reprice the revenue. It reprices the founder’s word.

Notice the asymmetry she accepted, because almost every founder accepts it without noticing. In Helen’s letter, the commercial terms bound neither party. The price was subject to diligence. The structure was subject to diligence. Everything the buyer had offered was provisional. The one clause with real teeth was the one that bound her: exclusivity. The buyer’s promises could move for sixty days while she could not, and by the time the promises moved, the alternatives were no longer alternatives.

Now the mechanics of the room took over. Some buyers who find a gap in diligence are surprised by it, still want the business, and adjust in good faith. Plenty of deals close that way. Others arrive intending to find one, because for them the letter of intent was never a price. It was a starting position, offered warmly, that diligence exists to erode. Nobody publishes a reliable count of how often deals are repriced between the letter of intent and the close, because the deals that adjust quietly are the ones nobody announces. What is not in doubt is the direction. The adjustments run one way.

Helen had no way to know which kind of buyer she had, and standing in the room, it made no difference. Her choices were identical either way. Accept the new terms. Argue back part of the way, from a position of documented weakness. Or walk, back to a market that had been told she was under offer, to explain to the next buyer why the last one left. I have sat with founders at exactly this point in a sale. The silence after the new number sounds the same in every deal.

The adjustment, when it came, did not arrive as a price cut at all. The rebuilt number justified taking almost $9 million off the headline. The buyer did not take it. It held the headline in place and moved $4 million of it into an earnout contingent on two years of performance, inside the buyer’s business, under the buyer’s brand, on the buyer’s sales model, with Helen expected to stay and deliver it. The escrow grew. The working capital target tightened. On the announcement, the deal would look unchanged. In the bank, it would not be. The price had not disappeared. Its certainty had.

A flaw disclosed early is context. A flaw discovered late is a discount.

05. Making the Number True

The version of this story that ends well is duller, and it begins twelve to twenty-four months earlier. The seller commissions the rebuild herself, before any buyer exists, and asks for it to be done the way the other side will do it: from the invoices up, unsentimentally, with no credit for good intentions. The definitions get written down. The one-time revenue is separated from the recurring. The concessions are surfaced, attached to their stories, and priced into the number the seller takes to market, instead of being extracted from it in week seven.

Founders resist this, and the resistance tends to take the same form: my numbers are clean. Perhaps they are. But the claim cannot be tested from the inside, by the person who defined the terms being tested. The founder who is right pays a defined cost and gains a number that has already survived the exercise a buyer will run.

The founder who is wrong can preserve millions for the cost of an accounting review. Few things in a sale are priced so far below what they return.

The process defences from Issue 16 still apply, and they apply with more force here. Exclusivity kept short. Alternatives kept alive until the price is truly found. A buyer who knows the seller can leave writes up his findings differently from a buyer who knows she cannot. But process alone cannot rescue a number that does not hold. Competitive tension wins a full price for what is real. Nothing wins a full price for what is not.

06. Confidence, Not Revenue

Because a buyer does not buy revenue. A buyer buys confidence that the revenue is what the seller claims, and that confidence is either built by the seller in the years before the process, or extracted from her, dollar by restated dollar, in the weeks after the letter of intent. One of those is a choice. The other is a room with one door.

Helen’s company was worth what she believed it was worth. Her number was not. The buyer did not find a weaker business. It found a weaker number, and it priced the business off the number. The company took her eleven years to build. The number could have been made true before the process began. The difference was paid at the closing table, by the only person in the room who had not seen it coming.

THE PRE-SALE REVENUE TEST

Four questions predict most of what a buyer’s rebuild will find. Can every dollar of ARR be tied to an active contract, net of concessions, with a reason it will recur? Is one-time work, implementation, onboarding, training, separated from subscription revenue? Do the schedules reflect what customers actually pay, side letters and saved renewals included? Could an outside accountant rebuild the same number without your explanation?

A buyer’s accountants will rebuild your numbers, and the rebuilt version is the one that gets paid for. Nexxit helps founders find what that rebuild will find, first: what recurs and what does not, where concessions hide, whether the records would hold. The Assessment is the entry point to Nexxit, Cube Capital’s acquirability programme.

Founders and transactions described in Founder Thesis are composites drawn from recurring patterns in software M&A.

Source: Software Equity Group, 2026 Buyers’ Perspectives Report, a survey of more than 200 private equity investors, strategic acquirers and software chief executives.