Ramp-up bottlenecks aren’t where you think they are.

On a site ramping up production, everyone points to the same machine, the same line, the same assumed bottleneck. Nine times out of ten, that’s not where the flow actually stalls. Here’s where I really look for the constraint — and why measuring before investing changes everything. By Mounir Telkass, founder of MT-Transition.

Callback within 2 business hours · 3 targeted profiles within 72h · 100% industry

1. The visible bottleneck is almost never the real one

Theory of constraints has said it for forty years, but on the ground it gets forgotten at every ramp-up: the station that looks the busiest, the loudest, the most closely watched isn’t necessarily the one limiting flow. I’ve seen industrial leadership teams invest in a second machine at the “obvious” station while the real constraint — an undersized inspection point upstream — kept limiting the entire line. First reflex on a site that’s plateauing: measure actual flow station by station before reaching for the chequebook.

2. The real constraint is often the people learning, not the machines running

On a recent site start-up — 650 people to recruit, 35,000 m² to bring online — the machinery was never the limiting factor: it was the team’s learning curve that set the real pace. With a structured onboarding protocol and close supervision in place from day one, the site reached its target productivity by week 6, against a sector average closer to week 16 for this type of start-up. The bottleneck was the time it took teams to build competency, not the theoretical throughput of the equipment.

3. The bottleneck hiding downstream: quality that sorts it all out

A line that speeds up upstream without quality control keeping pace gains nothing: it produces more scrap, more rework, and actual usable output stalls or drops. This bottleneck doesn’t show up on a standard production dashboard — you have to look at the real service rate at the output, not the raw line speed displayed on the floor.

4. The organisational bottleneck: who decides, and how fast

On a site ramping up, every delayed decision — a non-conforming part, a changeover, a supplier shortage — often costs more than any machine limit. The real constraint is frequently a decision chain that’s too long: three approvals for a minor adjustment, no one empowered on the floor to decide on the spot. Shortening that chain often unlocks more throughput than any capital investment.

5. Measure before you invest

Before any equipment purchase or large-scale hiring, I always have the real flow measured first: cycle time per station, work-in-progress, output service rate, decision time under disruption. This measurement takes one to two weeks. It prevents investments that, nine times out of ten, end up in the wrong place on the line.

The bottleneck that comes from your supplier, not your plant

A fifth of the ramp-ups I’ve led actually stalled outside the plant’s own walls. You speed up the line, train the teams, rebalance the stations — and flow stays capped because a critical supplier was never told about the new pace being demanded. The component that used to arrive weekly at normal cadence now needs to arrive every two days; nobody in purchasing renegotiated delivery frequency, or checked whether the supplier could actually keep up.

The symptom is misleading: it looks like a one-off stock-out, so it gets treated as an isolated incident, when in fact it will recur every cycle until supplier cadence is aligned with the target production rate. On an automotive ramp-up I supported, this diagnosis unlocked more flow than any internal adjustment: renegotiating two delivery frequencies with critical suppliers freed up more throughput than the previous three months of machine tuning.

What I look at first on a site that’s plateauing

Before making any recommendation, I systematically ask for three figures that many sites don’t have on hand in real time: actual cycle time per station — measured, not theoretical — the level of work-in-progress between each step, and the average decision time when something goes wrong. Cross-referenced, these three indicators almost always point to the real constraint within a few days, where weeks of committee discussion often only produce conflicting hypotheses. I systematically add a fourth, more qualitative angle: on the floor, who has the final say when a part is borderline compliant? If the answer is “it goes up to the quality manager, who signs off at the end of the day,” the bottleneck is neither the machine nor the operator — it’s that validation delay accumulating, station after station, across the whole line.

Why I almost always refuse the first investment requested

When I arrive on a site that’s ramping up, the initial request is almost always the same: an extra machine, a duplicated station, sometimes a whole new building. It’s usually the most visible option, the easiest to defend in a management committee — “we’re investing, so we’re acting” — and it’s also, nine times out of ten, the one that treats a symptom rather than the cause. I systematically ask for two to three weeks of measurement before any capital decision, which is unsettling at first: teams expect a quick call, not a diagnosis. But the cost of that measurement — a few weeks’ delay — is nothing compared to the cost of a misplaced investment, which ties up capital for years on a station that was never the real constraint.

On a plastics-processing site I supported, the initial request was for an additional press at €400,000. Measurement revealed that the real constraint was mould changeover time, poorly tooled and poorly trained for — a six-week SMED project unlocked more throughput than any new press could have delivered, for a fraction of the budget. This isn’t a blanket rule against investment: it’s a rule about sequencing. Measure first, invest second, and only where the measurement points.

What I see, site after site, is that the real bottleneck moves as you resolve it. Once the quality checkpoint is reinforced, the decision chain shortened and supplier frequencies aligned, a new constraint appears elsewhere — often more subtle, sometimes outside the initial scope of the ramp-up. That’s normal, and it’s actually a sign the work is progressing: a balanced line never has a fixed bottleneck, it has one that keeps moving with the pace being demanded. The right practice isn’t to “solve the bottleneck” once and for all, but to put in place a recurring flow measurement, month after month, throughout the entire ramp-up phase — not just at the start of the project.

A successful ramp-up isn’t managed by gut feel. It’s managed by measuring where flow is actually slowing down — and that’s almost never where the team assumes when they first arrive on site.

The simplest way to check this diagnosis on your own site comes down to one question: over the last four weeks, at which station does work-in-progress pile up right before it, and which station is, on the contrary, always empty first thing in the morning? The answer — almost always counter-intuitive the first time it’s genuinely measured — usually points exactly to where effort should go, and above all, where it shouldn’t.

A ramp-up that’s plateauing at your site? Let’s talk.

Callback within 2 business hours · 3 targeted profiles within 72h · 100% industry

See also