We don't run QoE engagements. We sit one layer upstream, building the data pipelines that score targets before a firm ever signs an LOI, which means a QoE shows up downstream of where our work lives. From that vantage one thing stands out: most of the operator metrics that surface inside a QoE — customer concentration, accruals shape, working-capital drift — have public proxies in regulatory filings, state registrations, and licensing databases the screen could have read months earlier. The interesting question isn't whether to run a QoE. Every buyer eventually does. The interesting question is what a deal team decides in the two weeks before they commission one, and what they do with the artifact after it lands.
That decision set is mostly invisible in the public literature, because the public literature is written by the providers. Read the top results for quality of earnings M&A: every page is written by the firm that hopes to bill for the work. Useful, but one-sided. This is the buyer's side of the same conversation, written from the data layer that sits next to it.
What actually triggers the commission
A QoE is the diligence step where someone finally has to pay. The trigger is usually one of three things: an LOI gets signed and the exclusivity clock starts, the lender requires a third-party report before issuing a term sheet, or the deal team noticed something in the seller's CIM that nobody can reconcile. The third reason is the most useful, because it tells you what the QoE is for in your specific deal. A scoped engagement that answers a specific question costs less and runs faster than the boilerplate one.
Morgan & Westfield notes that sell-side QoEs are becoming standard for all but the smallest transactions, which is true and also a tactical opening for the buyer. If the seller already commissioned one, the buy-side QoE shifts from "tell me what the numbers are" to "tell me where the seller's report soft-pedals." That's a different and cheaper engagement.
Scope decisions to make before signing the engagement letter
The provider will quote a default scope. You want a deliberate one. A mid-market QoE typically runs three to six weeks and costs between twenty and seventy-five thousand dollars, and the range exists because scope varies wildly. Four calls matter:
The period. Trailing twelve months is the default, but in a business with seasonality or a recent contract win, the more honest cut is sometimes a typical-quarter run rate. The provider will do whichever you specify. They will not pick for you.
The basis. Cash versus accrual is worth deciding before the engagement letter goes out. The CFA Institute's framing, that QoE goes beyond GAAP by adjusting for non-recurring items and normalizing revenue streams, is the polite version. The blunter version: a cash-basis target with no proof-of-cash workpaper can hide a quarter of aging receivables, and that's hard to catch without one.
The add-back posture. Aggressive, moderate, or conservative. The seller's broker has already calculated EBITDA on the aggressive end. Telling the provider where you want to land changes both the number and the conversation in IC.
Carve-outs. Customer concentration above some threshold, related-party revenue, and any business line you wouldn't roll forward post-close. Windes is direct on the first one: if forty percent of revenue comes from one client, the QoE will flag it as a valuation discount rather than wave it through. You want that flag early.
What the deliverable actually contains
The QoE is rarely a single number, even though it gets treated that way. The report bridges reported EBITDA to an adjusted figure and sets the working capital peg for the purchase agreement, and the peg is where post-close disputes start. A well-scoped engagement also gives you a proof-of-cash, a customer-concentration cut, and an aging schedule granular enough to argue with.
The non-recurring items that come out are usually familiar: gains and losses on asset sales, PPP forgiveness, employee retention credits, family members on payroll who don't operate the business, owner personal expenses run through the company, one-time litigation costs. Morgan & Westfield lists most of these explicitly. What the report doesn't do, and this is the part deal teams sometimes miss, is tell you whether the seller will accept those adjustments in the purchase agreement. That negotiation is a separate problem.
What this looks like from the data layer
The most underused output of a QoE is the pattern across deals, not the number for the current one. The operator-side issues a QoE surfaces — accruals problems, customer concentration, working-capital drift, related-party revenue — have public proxies. The CFA Institute's red flag list is a reasonable starting point: unusually high accounts receivable, excessive or obsolete inventory, customer concentration, hidden liabilities. Each of those shows up, in some form, in regulatory filings, state registrations, court records, and licensing databases. None of it requires an accountant to spot at the screen stage; it requires someone who's willing to wire the data sources into the scoring layer so the same problems get caught before an analyst spends a week on the model.
This is the loop the providers can't close: they produce one report at a time while the buyer is running screens by the hundred. The operator signals that come out of the QoE belong inside the sourcing layer, scored on every target before the LOI, not after it.
The $40K tutor
Here's the part that bothers us, watching from outside the engagement. A firm commissions a QoE. The report comes back. If it's clean, the deal closes and the lessons get filed. If it's ugly, the deal dies and the lessons get filed harder. Either way the $40K bought a single decision on a single target, and the screen that surfaced that target keeps surfacing the same shape of target next quarter, because nothing about the screen changed.
The QoE is a $40K tutor, and the tuition usually gets paid without anyone capturing the lesson. The asymmetry is brutal: a clean report confirms only that target; an ugly one, properly captured as features feeding the next screen, can quietly disqualify the next twenty targets that share the same shape before anyone wastes an analyst-week on them. Nobody bills you for that compounding, so almost nobody builds the layer that does it.
