Operational due diligence expanded into a formal discipline for a defensive reason. After the 2008 crisis and the Madoff fraud exposed how little anyone had checked the plumbing behind the returns, limited partners stopped taking a manager’s back office on trust and started auditing it. That origin still shapes what the phrase means, and it explains a gap that costs buyers real money on the deal side: the M&A version of operational due diligence inherited the checklist without inheriting the one question that now decides mid-market returns.
The word carries two jobs, and the deal side gets the thinner one
There are two things called operational due diligence, and they are not the same work. The first is fund-level: an LP assessing whether a manager’s operations, controls and service providers can be trusted with capital. The second is deal-level: a buyer assessing whether a target’s operations can carry the growth the investment thesis is priced on. The first is about counterparty risk. The second is about whether the business actually runs the way the model assumes it does.
Deal-side operational due diligence has a solid, well-worn scope. It reads the organisation chart, the key-person dependencies, the core processes, the supplier concentration, the facilities, and the management team’s ability to execute the plan. It is genuinely useful, and a good operational read has saved plenty of buyers from a target that looked scalable on paper and turned out to be held together by three people and a spreadsheet. The weakness is not in the discipline but in where it stops, because it treats the technology estate as a risk column, something to be flagged for security exposure or licensing liability, rather than as the place where most of the suppressed EBITDA in a mid-market service business is actually sitting. At Digital Forms we call the pattern it misses the Manual Wall: the point where a company keeps buying capacity in people because the systems underneath were never built to carry the volume.
Why this gap is expensive now, when it wasn’t a decade ago
For most of the last cycle, a buyer could underwrite a return without looking too hard at how the work got done. Cheap debt and rising multiples did a lot of the lifting, so an operational blind spot was survivable. Bain’s recent Global Private Equity Report gave the change a memorable shorthand, that 12 is the new 5, meaning a buyout that used to clear its target on around 5% annual EBITDA growth now needs closer to 10% to 12% to reach the same multiple over the hold. When the return has to be built inside the business rather than supplied by the market, anything the diligence failed to see becomes a direct hit to IRR.
Two structural shifts sharpen the point. Holds have lengthened, with Bain putting the median buyout hold at roughly seven years, so a manual process left unexamined at entry is not a fixed cost, it compounds against the fund every quarter the asset is owned. And the accountability for fixing it has moved: operating partners now sit close to the centre of value creation rather than the periphery, which means the person who owns the number is also the person who inherits whatever the operational read never opened. A blind spot that was academic in 2015 is a scored line on the value creation plan today.
What does operational due diligence actually cover in an M&A deal?
A conventional deal-side operational read works through a recognisable checklist. It covers the management and organisation structure, the maturity of core processes and standard operating procedures, customer and supplier concentration, capacity and scalability constraints, regulatory and compliance exposure, and the integration or carve-out complexity if the deal needs one. Systems usually appear on that list, but they appear as an entry, not as a thesis. The question asked is whether the systems are a liability, whether they are secure, supported, licensed, and unlikely to fall over.
That is a fair question, and it is not the one that moves the number. The question that moves the number is how much of the day-to-day work inside those systems is being done by people because the software was never wired to do it. A target can pass every line of a standard operational checklist, with clean SOPs, a capable team and no alarming supplier concentration, and still be carrying a payroll line that exists only because two platforms don’t talk to each other. Nothing in the traditional scope is built to surface that, because it reads process quality and headcount as facts to be recorded rather than as a cost to be recovered.
Why operational and technology due diligence keep missing each other
The gap is structural, not a failure of any single advisor. On most deals the two workstreams run in parallel and report separately. Technology due diligence, when it happens at all, comes back as a risk report: architecture, security posture, technical debt, scalability of the stack. Operational due diligence comes back as a people-and-process assessment. Each is competent inside its own boundary, and neither owns the seam between them, which is exactly where the money is. The seam is the set of workflows where a manual operational process persists because a technical capability is missing, and it belongs to both reports and neither.
This is the same unpriced variable that our private equity engagements are built around. Commercial and financial diligence get covered thoroughly, then the technology read comes back as a document that gets filed, and the value creation plan gets written by people who won’t be in the room to deliver it. The operational upside hiding in how the business runs, the automation gaps, the manual reconciliation, the rekeying between systems, gets left on the table because no one whose remit is the deal thesis ever quantified it. When the technology read and the operational read are done by the same eye, the seam stops being invisible, and the manual work starts showing up as a modelled number instead of a footnote.
Where the blind spot shows up in the numbers
In a mid-market service business the tell is usually growth that looks healthy from the outside. Revenue is climbing and headcount is climbing with it, and two lines rising together read as a company scaling well. Underneath, payroll is often growing faster than throughput, because each new hire is quietly brought on to do work that software should be handling. A team of forty processing cases through a workflow that touches a dozen disconnected screens will ask for more people long before it asks for a better system, and a standard operational read will record the hiring plan as evidence of momentum rather than as evidence of a structural constraint.
Consider the shape without a logo attached. A specialist claims or servicing operation grows its volume by a third, hires proportionally to keep up, and reports stable margins, which the diligence dutifully notes. What the diligence does not note is that a large share of that headcount exists to move data between a customer platform and a billing platform by hand, to generate output documents one at a time, and to chase the exceptions that a connected system would never have created. That work is invisible to a checklist because it has no line item. It is simply how the business has always run. The upside only becomes legible when someone counts the hours going into it and prices what recovering them is worth, which is the difference between an operational read that describes a business and one that can be underwritten.
What a technology-aware operational read looks like
The fix is not a heavier checklist, it is a different output. A read worth underwriting produces three things an investment committee can actually price. The first is an operational risk register that tags each finding by how it should affect the deal, whether it is a price chip, a representation-and-warranty issue, or a genuine red flag. The second is a conservative, confidence-rated model of the EBITDA upside that recovering the manual work would release, expressed as a number the buyer can defend rather than a promise. The third is a sequenced first-quarter plan that says which moves come first and why, so the value creation work has a starting line the day the deal closes rather than a discovery phase.
The discipline that produces this is the same whether the asset is a target or one already in the portfolio. When we run it on an owned asset the entry point is a Profit Leak Diagnostic, a short read that finds where the manual work is concentrated and what it costs, and the reason we built it that way is that operators consistently underestimate how much margin is trapped in processes they have stopped noticing. On a live deal the same read feeds pricing and structure directly, because a quantified manual-work number is a legitimate input to what the asset is worth, not a post-close surprise.
How operating partners can close the gap before close
The practical move is narrow and it changes little about the deal process. When commissioning the operational read, add one explicit requirement: the technology estate is to be assessed as an operating-upside question, not only as a risk column, and the manual work inside the core workflows is to be counted and priced. That single instruction pulls the seam between the two workstreams into someone’s remit, and it converts a vague sense that the business is people-heavy into a figure that belongs in the model.
The second move is about who executes the finding. A read that quantifies the upside is worth far more when the people who scoped it are the people who deliver it, because the context does not have to be rebuilt from a filed report. That is why our engagements are designed to land in delivery rather than stop at recommendation, whether that is a fast first build through an Operations Sprint that puts a working change live in weeks, or an embedded team that carries the value creation plan through the hold. The operating partner who owns the EBITDA number is better served by a partner who stays to move it than by an advisor who has already moved on.
The question the model should have to answer
Most diligence can tell a buyer whether a target’s operations are competent. Fewer reads can tell the buyer how much of that competence is people compensating for software that was never built, and that figure is now large enough, and the return math tight enough, that leaving it uncounted is a pricing decision whether the deal team means it to be or not. The next time an operational read comes back clean, the useful thing to ask is not whether the business runs well. It is how much it is paying, in payroll, for the parts that a system should be running instead.
