Most of it never becomes a deal
The companies I sat on the board of are the ones the fund bought. The number I walked through, took apart and wrote up is several times larger. Most operational diligence ends in a recommendation not to proceed — and that is the half of the job that saves money rather than spends it.
So the pattern library is much bigger than the deal list. When a partner asks whether I have seen a failure mode before, the answer often comes from a company nobody bought.
When a partner calls me
The management presentation was excellent and nobody in the room has walked the plant. The operating thesis rests on a capacity number the company has never actually hit. You are ninety days from close and the hundred-day plan is a document, not a set of names. The pack has been green for three quarters and the cash says otherwise.
None of those is a data problem. They are all the same problem: nobody independent has been inside the operation.
Four moments, four different asks
01 — Before the deal
Is the operational thesis real, and what does the gap cost?
Operational diligence, technology and manufacturing readiness, IP and switching-cost validation.
02 — First hundred days
Get control without breaking what works.
Hundred-day plan, CEO onboarding, delegation of authority, governance stood up.
03 — The hold period
Make the plan happen, and know early when it is not.
Value creation plan and KPI tree, quarterly tracking, portfolio-wide compliance, embedded operating cadence.
04 — Exit
Know what the next buyer’s diligence will find, eighteen months before they run it.
Before the deal
A week to ten days inside the business: the plant walked, the product definition, the supply base and the order book taken to pieces. The report is a variance against the investment thesis, not a description of the company: every finding lands as confirmed, better than expected, or not there, and each gap is costed in money and months.
Two patterns recur. The value is often not where the memorandum puts it: in an equipment business it frequently sits in the controller and the software rather than the iron, because that is where the switching cost lives. A controller company that happens to make machines and a machine company that happens to make a good controller are different theses, and the memorandum rarely says which you are buying. And on management: a director title at a large parent often means process, not ownership. What did they decide, and what did they coordinate?
Sometimes the thesis is right and the business is the wrong shape for it — a profitable core carrying segments that lost money under the last owner. That is a different plan, worth knowing before the price is set: the business shrank; the overhead did not.
The first hundred days
A new owner, often a new CEO, and a plan written before anybody had seen the inside of the business. The failure mode is not moving too slowly. It can be moving too fast before the business has stabilized and everyone knows the new rules. Get control without breaking what works and embed the new governance structure.
Spend real time inside the business. Visit the key customers. Visit the key suppliers. Talk to the warehouse supervisor at 6 am. Then revisit the plan. The post-close report is the diligence report re-run: what the thesis said, what was found, what has moved. Then every line of the plan carries two named leads — one from the sponsor and one from the company. Not an owner and a helper; two names, both accountable. Embedded into personal incentives, even if the incentive is tiny. That way these plans get executed rather than admired.
Value creation planning
Most value creation plans are a list of good intentions with a number at the bottom. The useful version is a tree: enterprise value at the top, the three or four drivers that move it, and beneath those the operational measures a plant manager can affect on a Tuesday. If a line cannot be traced down to something somebody does differently, it will not happen. Objectives are defined jointly with the sponsor rather than submitted to them. The method and the thinking behind it are written up.
The hold period
A board that has not decided how it will work will decide it by accident, in the first bad quarter. A pack format that stays the same every month, so the board reads changes rather than re-reading the business. Committee charters written before the committee is needed. Governance is cheap in month one and expensive in month fourteen.
Two things happen during a hold that the plan did not assume. The market turns — in equipment it does not taper, it falls off a cliff, and the number you need that Monday is the committed-versus-cancellable register kept from month one. Or the business turns out to have changed shape under the previous owner. The first is exposure; the second is structure. They look alike from the board pack and need different work.
Exit
The next buyer’s operational diligence will find what mine finds. Months out, run it against yourself: is the product written down so it survives the founders leaving; is the fixed cost sized to the core rather than to the company it used to be; can the register answer the Monday question. A company that has done that work can be priced by the seller with confidence. One that has not is bought on the buyer’s walk-through at the buyer’s price.
Portfolio-wide compliance
The work nobody wants until the moment they want it very much. Environmental and safety, CSR, import and export — designed once and rolled out, with reporting a fund can aggregate. The same program installs at eight companies for barely more than the cost of one.
Where this has been done
Board seats and operating roles across venture-backed, private-equity-owned, family-owned and listed businesses, and through an IPO — capital equipment, energy storage, plastics machinery, materials testing, the industrial aftermarket. The plan formats are the ones institutional sponsors already use; the operations half is what I add. The methods are the work; the structure they build is what the fund is left holding. Where most of them came from is written up: one machine a day to a hundred a month, and then the 2008 cliff.
