Container Glass Confidentiality: Why We Sign NDAs First

"What exactly are you going to do with our redraw numbers?" A plant GM asked me that on my second visit, before he'd let me near the batch house. Fair question. Redraw rate and percent-defective-to-warehouse are the two figures most container glass plants will not hand over without a signed mutual NDA, because both numbers tell a competitor exactly what your yield economics look like.

NDA-by-default is not a formality

Most consultancies treat the NDA as paperwork the lawyers push through before the real work starts. We treat it as the real work. Every engagement at Lean Glass starts with a mutual NDA signed before anyone walks the hot end. Not a formality. A precondition.

That matters more in container glass than people outside the industry realise, because a proper audit needs sign-off access across six roles on a shift: batch house operator, furnace and glass technologist, IS machine minder, forming supervisor, cold-end QC inspector and lehr operator. The hot-end superintendent owns recipe lock on that roster, and nobody touches a set point without their sign-off. Giving an outside auditor visibility into recipe changes is a different negotiation to handing over a furnace pull-rate chart, and that's exactly where scope talks get contentious.

Our founder, Zaid Hassoneh, ran that exact roster of six roles himself long before he audited anyone else's plant, starting on the floor at O-I Brisbane in 2005 and making plant manager by 2019. He knows on sight what's genuinely proprietary and what's just habit dressed up as a secret.

In 2021 I audited a mixed-OEM plant outside Dammam running three lines off two furnaces, part of the wider push under Saudi Vision 2030 to build domestic container glass capacity instead of importing packaging, the kind of investment that's put producers like Obeikan Glass on the map. The GM's condition for access was blunt: no photography within ten metres of the batch house, no video near the colour dosing station, every KPI published indexed rather than absolute (and yes, he watched us delete the photos off the camera roll in front of him before we left). We agreed before the scope was even final. That's how this work has to run if you want the plant to let you back in.

Why the case studies never name the plant

Every case study we publish is anonymised properly, not the kind where you can still guess the site from the pull rate and the region mentioned in the same paragraph. Furnace geometry, campaign age and the specific pull-rate-to-design-capacity ratio get stripped before a word of the report leaves site.

Those two figures, campaign life and pull-rate-to-design-capacity, are the ones plant GMs guard hardest in any benchmarking conversation. A well-run regenerative furnace runs ten to fifteen years between cold repairs. Publish that number against a named plant and you've told every competitor in the region how close that furnace is to a forced rebuild, and how much capital runway the plant has left.

A case study that still lets a competitor guess the plant isn't anonymised. It's just badly disguised marketing.

On that Dammam line, tightening batch house weighing tolerance back to target closed a redox drift that had pushed seed counts from under two per 100 containers to over five, a swing that showed up a full day before the cold-end reject rate moved. Redraw rate on that line dropped -18% inside six weeks. That number is real. It never appeared with the plant's name attached, and it never will.

The gap OEM-affiliated advisors won't negotiate

An OEM-affiliated consultancy walks in already conflicted. If the recommendation at the end of the audit is a furnace rebuild or a refractory brand, and that recommendation happens to favour the parent company's own equipment line, the client has no way to separate good advice from a sales funnel. Whether the plant runs an old Emhart 8-section on 1980s cam-shaft controls or a recent Heye Smart H1 makes no difference to that conflict. The confidentiality standard should be identical regardless of what badge is on the machine base, and with an OEM-tied advisor it usually isn't.

A genuinely vendor-neutral shop has no equipment to sell, so the entire negotiation over data access is about the plant's interests, not ours. That's the basis of what an independent, vendor-neutral container glass consultant actually offers, and it's worth saying plainly rather than burying it in the small print of a services PDF.

And this is where the data-handling discipline actually gets tested, not in the contract, but on the floor at 2am when a section is choking and someone hands you a laptop with the live gob weight trend on it. Gob weight is normally held to ±0.5g of target on the IS machine. Drift past that band and you get checker defects downstream in the lehr, thermal-shock cracks that surface as customer complaints two months later. We see that number, we report the trend and the corrective action, and we do not screenshot it into a slide deck with the plant's name on it or reference it in a proposal to a different client down the road.

What actually stays in the room

The clean-room principle is simple to state and harder to run well. Furnace geometry, campaign age, redraw and defective-to-warehouse figures, colour formulations, anything from a batch house weighing tolerance report (target ±0.5% on most well-run lines) stays with the plant. What travels into a case study or a benchmarking deck is the shape of the finding: a defect mode, a root cause, a KPI delta, stripped of anything that identifies the site.

A shift handover is where this gets tested in practice too. On most lines I've worked, the 0600 handover misses the night-shift swabbing data seventy percent of the time, which means the incoming crew is often chasing a defect with half the picture. When we flag that gap in a report, we flag the pattern, not the names of the two operators on shift that week. That distinction is the entire difference between an audit that fixes a process and one that gets someone written up.

The trust problem is the actual product

None of this data-handling discipline shows up on an invoice. It's not a line item. But it's the reason a plant lets you back in for a second audit, and a third, and eventually lets you run strategic advisory work across the whole site instead of one furnace campaign. Trust compounds slower than a KPI chart and breaks faster than one too.

Look, the paperwork isn't the interesting part of this job. The interesting part is that a plant running on gas priced well under two dollars a million BTU, against the three-to-four-dollar international benchmark European and US plants pay, is sitting on a structural cost advantage most outside auditors never ask about, let alone protect once they've seen it. That's the kind of number that belongs in a confidential report and nowhere else.

So next time someone asks what we're going to do with their redraw numbers, the answer is the same as it was outside that batch house near Dammam: nothing you haven't approved, in a form nobody can trace back to you, under a contract we signed before we ever walked the floor.