Digital Transformation in Packaging Dies in the Pilot

"Digital transformation" is not a thing you can buy, and in packaging it is not a thing most companies have done. It is a phrase that covers everything from a new label printer to an ERP replacement, which is why it survives on conference agendas and rarely survives contact with a converter's Tuesday.

What is measurable is narrower and more useful: of the companies that start, how many get past the pilot.

Three Quarters Never Leave the Pilot

McKinsey has a name for this. Its digital-manufacturing research describes companies stuck in "pilot purgatory": technology proven on one line, in one plant, by one enthusiastic team, and then never scaled. As of late 2020, about 74% of surveyed companies reported being in it.

That figure is five years old, which raises the fair objection: the technology has changed beyond recognition since, so surely the number has too. It has not. McKinsey's 2025 survey, published in November, found adoption close to universal and scaling still rare, with most organisations reporting they had not begun scaling AI across the enterprise at all. Different technology, same wall.

The printing and packaging numbers underneath it look the same. In the Printing Outlook 2026–27 survey, only 10% of printers said they actively use AI or machine learning in their operations, with a further 22% piloting. Another 22% described themselves as "never-AI-ers", which is not undecided but opposed.

A small group has shipped something. A similar-sized group is trying. The largest group has not started, and a fifth has ruled it out. Five years of better technology moved almost none of that.

The Word Is Doing Too Much Work

Part of the reason projects stall is that nobody agrees what was bought.

Japan's Ministry of Economy, Trade and Industry publishes one of the few government attempts to quantify the cost of not moving, and the printing industry there has taken it seriously enough to write about it in its own trade journal. The ministry's DX Report warns of what it calls the 2025 cliff: legacy systems that are complex, aged, and black-boxed, roughly 90% of IT budgets consumed maintaining them rather than building anything, and an estimated economic loss of up to ¥12 trillion a year from 2025 onward if that continues.

Whatever you make of the arithmetic, the description of a legacy system is worth borrowing, because it is a checklist rather than a mood:

• Nobody has documented it
• Data cannot easily be got out of it
• Any change has wide blast radius and needs long testing
• The people who understood it have left
• Maintenance costs more each year
• The vendor's support has ended

A converter can answer those six honestly in an afternoon. That is a more useful starting point than a transformation strategy, because it tells you which system is actually the constraint, and it is frequently not the one anyone was planning to replace.

The Clock Nobody Puts in the Plan

There is a deadline in this that has nothing to do with software, and most plans ignore it.

The pricing knowledge in most converters is not written down. It is held by two or three people who know which jobs to be careful with, which customers argue, and what the last three quotes for this box actually cost. When they retire, that leaves the building, and no amount of digital ambition replaces it afterwards.

This is not an American peculiarity. Japan has been measuring the same thing for two decades, and its manufacturing sector shows the shape of it plainly: workers aged 34 and under fell from 3.84 million in 2002 to 2.59 million in 2023, while workers aged 65 and over rose from 0.58 million to 0.88 million. The over-65 share of the manufacturing workforce nearly doubled, from 4.6% to 8.3%.

Europe faces the arithmetic one level up: by 2050, 22 of the 27 EU member states are expected to see their working-age population shrink. A converter in Germany or Italy is not competing for estimators against other converters. It is competing against every other employer in a smaller pool.

This changes the ordering of a transformation programme. Capturing how your estimators price, while they are still there to correct it, is time-sensitive in a way that a plant-floor sensor project is not. One has a deadline set by demographics. The other can wait a year without penalty.

The Buyers Already Moved

The demand side did not wait for the industry to be ready.

NAPCO Research's survey work puts print providers' mean turnaround for label and packaging work at 11 days, against a mean brand-owner requirement of 7, with 38% of brand owners wanting 2 to 5 days.

That is not a preference gap, it is a four-day structural deficit, and it is the reason "we are looking at digitising" keeps turning into "we lost that account."

Adjacent research on industrial part buyers is blunter still: in a survey of 417 buyers, 67% expected a quote in under 24 hours and 75% made the decision within a week of receiving quotes. Those buyers are not waiting out your programme plan. They are choosing whoever answered.

And the Money Got Tighter

The other thing that changed is what a board will fund.

Packaging M&A multiples have compressed sharply: Capstone Partners puts median EV/EBITDA at 6.5× year-to-date 2026, against 10.3× for full-year 2025 and a 2018–2026 average of 9.3×. Whatever that says about valuations, it says something simple about capital discipline: this is not a market in which a two-year internal programme with benefits at the end gets waved through.

Which is fine, because the two-year programme was the thing that was failing anyway.

What Actually Ships

The pattern that works is unglamorous and mostly organisational.

Pick the constraint, not the roadmap. One system, chosen because it fails the legacy checklist above and because its failure is visible to customers. For most converters that is the quote desk, not the plant.

Put one team on it live. Not a sandbox: real quotes, real customers, while everything else runs unchanged. A pilot that cannot lose you money also cannot teach you anything, and it is why so many of them prove nothing and end in purgatory.

Make the second team a decision, not a date. They join when the first team prefers the new way. If that never happens, you have learned something cheap and early, which is the entire point.

Keep the change in your hands. If adjusting a surcharge or a spec field requires a vendor ticket, the system will drift out of date faster than your prices do, and the transformation quietly reverts to a spreadsheet.

We are not neutral about this ordering. We run a packaging business on this software, and we came out of enterprise ERP, content, and CRM implementations before that, where the long version failed often enough to be predictable. It rarely failed on the software. It failed in the twelve months between deciding and using, which is time this industry no longer has to spend.

If you want the detail of how that sequencing runs in practice, we wrote it down.

The Honest Summary

Digital transformation in packaging is not a technology problem, a vendor problem, or a generational one. It is a scoping problem. Programmes fail; sequences ship. The companies pulling ahead are not the ones with the most ambitious plan. They are the ones with something small already live, doing real work, while the plan is still being written somewhere else.