The three SaaS quality-of-earnings adjustments standard QoE misses.

We have written about how we underwrite a software acquisition for a 25-year hold and about the diligence checklist we run end-to-end. The quality-of-earnings workstream sits inside the financial diligence stream of that checklist, but its scope is much narrower and its consequences much larger than any other single workstream. The earnings number that survives QoE becomes the denominator on the multiple, the basis for the bank financing, and the steady state our model rolls forward for the next two decades.

Why the standard QoE misses SaaS

Standard QoE methodology was built for businesses with relatively short revenue cycles, working capital that turns in weeks, and capital expenditure that lands on the balance sheet. B2B SaaS inverts all three. Revenue is multi-year and recognized over time. Working capital is dominated by deferred revenue and contract-related accruals. The largest capital expenditure — engineering investment in the product — is expensed in the period it occurs and lives nowhere on the balance sheet. A generic QoE applied to a SaaS target will produce a clean report and a wrong earnings number.

The point of a SaaS QoE is to reconcile the cash the business generates to the steady-state earnings a long-term owner will harvest. The standard methodology reconciles the trailing twelve months. They are not the same number.

The three adjustments that matter most

1. Capitalized versus expensed engineering investment

Most lower-middle-market SaaS targets expense engineering in full. A few capitalize the portion that qualifies under accounting standards. Neither presentation gives a long-hold buyer the information they actually need, which is the portion of the current engineering spend that is investment in the next decade of the product versus the portion that is maintenance of the current surface. We pull the engineering roster, the project list, and the time-tracking data, and we rebuild the engineering spend into investment, maintenance, and coercion buckets, using the same framework we run inside every Cobalt Glacier brand. The portion that is maintenance is a real cost of running the steady-state business. The portion that is investment is option value we want to underwrite separately.

2. Deferred revenue and the working-capital trap

A growing SaaS business prints deferred revenue every quarter as multi-year contracts are billed annually and recognized over time. The standard QoE often treats the growth in deferred revenue as a working-capital benefit, which makes cash generation look better than the underlying business warrants. We unwind deferred revenue back to a recognition-basis cash conversion and separately track the contribution of upfront billings as a financing item, not an operating one. On a fast-growing target, the swing is often two-x the headline EBITDA number. On a target that has slowed, the unwind tells you the business is generating less cash than the income statement suggests, and the target's owners frequently do not know it.

3. Customer-acquisition cost economics

Customer-acquisition cost is fully expensed in the period incurred, but on a permanent-capital clock, the right way to think about it is as an investment with a multi-year payback. We rebuild CAC by cohort, compute payback on a contribution-margin basis (not a revenue basis), and isolate the steady-state CAC the business would run if it stopped underwriting new logo growth tomorrow. That steady-state CAC is the number that belongs in the long-hold earnings model. The difference between current CAC and steady-state CAC is, again, investment we underwrite separately. This is the same logic we apply to NRR as the only B2B SaaS metric that compounds — once you isolate the recurring engine, the growth spending becomes optional rather than mandatory.

The adjustments we make less often

Owner compensation normalization, related-party expenses, non-recurring legal and professional fees, and one-time revenue items are all standard QoE territory and are usually well-handled by the credentialed firm. We accept their numbers on these items roughly nine times out of ten. Where we diverge is on the SaaS-native items above, and the divergence is rarely about disagreement on methodology — it is about a question that was never asked.

The common mistakes

• Accepting a sell-side QoE at face value. The sell-side report is a professional product. It is also a marketing document, and the adjustments inside it are chosen to support a process. We rebuild the model from source data even when the sell-side report is from a firm we respect.
• Anchoring on trailing twelve months. TTM is a snapshot of a business mid-investment cycle. The earnings number that matters to a twenty-five-year owner is the steady-state run-rate after the investment cycle stabilizes. We model both and underwrite to the steady state.
• Treating the deferred revenue benefit as permanent. It is not. When growth slows, the benefit reverses, often abruptly. Any model that does not run the reversal scenario is missing the most important downside case.
• Confusing CAC payback with profitability. A short CAC payback is a great metric. It is not the same as the business being profitable at steady state. We separate the two and look at them in parallel.

How we use the answer

The SaaS QoE produces three numbers we carry forward into every other workstream: a steady-state earnings number, a steady-state cash conversion ratio, and an investment spend the business is currently absorbing inside the income statement. The first number drives the multiple. The second drives the bank financing. The third drives the conversation with the founder and the operating team about what changes after close and what deliberately does not. The three numbers, together, are the underwriting answer. Without them, the deal is being priced on a number the long-hold owner will not actually own.

How the rebuild shows up in the model

The output of the SaaS QoE flows into the financial model in three specific places. The steady-state earnings number replaces TTM EBITDA as the denominator on the entry multiple. The steady-state cash conversion ratio replaces a generic seventy-five percent assumption with a target-specific number that the bank financing is sized to. The isolated investment spend becomes a separate line item in the first three years of the post-close operating plan, with a decision point at the end of each year about whether to continue funding it at the current level or to harvest some portion back to the FCF line. The model is no more complicated than it would be without the rebuild. It is meaningfully more honest about what a long-hold owner is actually buying, which is the only thing the financial model is supposed to do. The rebuild also produces a small number of follow-up questions for the operating team that frequently become the most useful conversations in the first ninety days after close.

The bottom line

Quality of earnings in B2B SaaS is a different exercise than quality of earnings in a general industrials target. The three adjustments that matter most — engineering investment, deferred revenue mechanics, and CAC economics — are systematically missed by standard methodology, and the cumulative impact on the steady-state earnings number is typically meaningful. We rebuild the QoE from source data on every target, even when the sell-side report is well-prepared, because the number that comes out of the rebuild is the one we will own for the next two decades.

If you are an LP or co-investor evaluating Cobalt Glacier's underwriting process, start a conversation with our investor team. The quality-of-earnings rebuild is one of the workstreams we walk through in detail because it is the part of underwriting where the largest amount of value is created or destroyed before a deal is signed.