PE diligence on a container glass acquisition: the 14-day pass

Every CIM I've read in the last ten years carries a 'normalised run-rate' line that sits 8-12 points above what the plant's own SCADA historian actually logged over the trailing 12 months. Sellers call that normalisation. I call it the gap a private equity due diligence team has to close before the deal closes, not after. Fourteen days. Not fourteen months.

The run-rate number in the CIM is never the number on the floor

Sellers normalise run-rate by stripping out anything they can label a one-off: the furnace campaign-end rebuild, the flood in section three, the six-week job change backlog left behind by a SKU rationalisation. Some of that stripping is fair. Most of it isn't. On a five-line container glass plant I did buy-side technical diligence on in 2019, in the US Southeast, the CIM showed a normalised OEE of 91.5%. The trailing actual, pulled straight off the historian rather than the seller's monthly rollup, was 79.2%.

11 of those points came from job change variance alone, the same 30-60% cross-shift variance you see on almost any plant that hasn't systemised its changeover with something like a Job Change Tool. On a well-run line, the hot-end superintendent owns recipe lock and the operator doesn't touch a set point without sign-off. Ask to see that sign-off log during the plant walk. If it doesn't exist, the 'normalised' number in the CIM is a guess dressed up as a KPI.

Look, the data says one thing and the floor says another, and a 14-day pass exists to reconcile the two before the wire goes out. That gap matters more right now in the US, where capacity rationalisation among incumbents like O-I, Ardagh, and Anchor Hocking is pushing more bolt-on deal flow toward independent and regional container glass plants, exactly the assets that haven't systemised a changeover or catalogued a mould shop in a decade.

A CIM tells you what the plant did on its best week. Diligence has to find out what it does on a Tuesday.

Campaign life is a mould-shop problem before it's a melting problem

PE analysts default to furnace campaign life because it's the number the seller volunteers up front: a furnace campaign of 8-12 years, rebuild scheduled whenever it's scheduled. That's the wrong campaign to interrogate first. The one that bites six months post-close is mould campaign life, and it lives in a spreadsheet the mould shop foreman keeps, not in the data room.

Not a furnace problem. A mould-shop problem. Push a mould set two years past its design cycle count and you get choked necks, settle waves, and a checkline reject rate that creeps rather than spikes, so nobody flags it until the quarterly numbers come in soft. The mould shop owns campaign tracking on a well-run line. If nobody on the floor can tell you the cycle count on the plant's highest-volume mould set inside five minutes, that's your first red flag, and it usually means spare inventory is thinner than the asset register suggests too.

That same 2019 plant ran two generations of forming equipment on one floor: a newer line on a Heye Smart H1 holding gob weight CV under 0.4% with closed-loop control, and an older line still running late-1990s Emhart IS machines on open-loop timing. The CIM blended both into one plant average. The older line's actual run-rate was 20 points below the newer one's, and its mould sets hadn't been re-catalogued since the plant's last ownership change.

Coating drift is a capex bill nobody wrote down

Coating drift doesn't show up as a line item anywhere in a CIM. Hot-end coating protects the mould cavity, and when a plant lets the spray interval slip to save on tin tetrachloride, cavity wear accelerates and you start seeing stones and cord on ware that should run clean. Cold-end coating drift is worse for a buyer because it stays invisible until a customer's incoming-line audit catches it, months after close.

And nobody puts coating drift on a maintenance ticket until the rejects show up on the checkline, by which point it's the new owner's problem, not the seller's.

Target coating thickness on most cold-end lines runs close to 0.5 microns, applied inline right before the ware hits the lehr exit (and yes, the seller's quality manager will tell you the coating booth is on schedule, check the spray log dates yourself rather than take the word for it).

What a 14-day pass has to pull off the floor

A technical pass this short only works if it's structured. On every deal we've supported, the list looks roughly like this:

• Pull 12 months of SCADA historian data instead of the seller's monthly rollup
• Walk the mould shop and check cycle counts against design life, not against the rebuild calendar
• Check hot-end and cold-end coating logs against the actual spray schedule, not the maintenance plan
• Sit in on one 0600 shift handover and see what the outgoing crew actually records
• Compare gob weight CV (target ≤0.4%) and forehearth zone profile (target ±2°C across the zones) against the recipe card, not the spec sheet

That 0600 handover is worth the trip on its own. On most lines I've audited, the swabbing data and mould temperature notes from the night shift don't survive the handover about 70% of the time, which means the day shift is running blind on the exact variables that drive campaign life and coating wear.

What operator-credentialed diligence buys that banker calls don't

Financial due diligence will find the working capital problems and the customer concentration risk. It won't tell you whether the plant can hold its number the Tuesday after close. That's a forming-floor question, and answering it needs someone who has run a hot end, not modelled one from a spreadsheet.

Lean Glass was built by an operator, not a consultant who never ran a line. Our founder started on the floor at O-I Brisbane in 2005, made plant manager by 2019, and then led a $220M USD greenfield build in the US through 2022. We run this as a standalone 14-day pass or fold it into a broader strategic advisory engagement, and every finding is operator-credentialed rather than vendor-sourced, because the OEM quoting your rebuild has no reason to tell you the mould shop is two years behind on cycle tracking.

That's the core of container glass PE due diligence: vendor-neutral, operator-led, and done in 14 days, not 14 months.