Zimbabwean industry runs at 56% of capacity — and a third of it is buying more.
Both of those come from the same survey, and put side by side they are hard to reconcile. If a plant is using a bit over half of what it already owns, the first question isn't how to add capacity. It's where the other 44% goes — and on most Zimbabwean sites nobody can answer that with a number, because nothing measures it.
CZI, 388 manufacturers, all ten provinces.
The Confederation of Zimbabwe Industries published its sixteenth annual manufacturing survey in June 2026, covering 388 formal manufacturers across all ten provinces, of which 72% were not CZI members — so it reads as the sector describing itself rather than an association describing its own membership.
55.9%
Capacity utilisation in 2025, up from 52.3% in 2024 — the fourth-highest reading since CZI began tracking it. Government figures put the first quarter of 2026 near 57%.
~35%
Share of firms that invested in expanding capacity during 2025. Output grew 13%, turnover 12% and employment 6% across the surveyed firms.
18.4% vs 8.4%
Turnover growth in 2025 among the roughly one third of firms that upgraded technology, against those that did not. More than double, in the same economy and the same year.
That last figure is the one worth sitting with. It is not a vendor's claim and it is not our number — it is the national manufacturers' association reporting what happened to its own respondents. The firms that spent on technology grew at more than twice the rate of the firms that didn't.
Buying capacity you already own.
Capacity utilisation is a plant-level self-assessment: roughly, what you produced against what you believe you could produce. At 56% it says the average Zimbabwean manufacturer thinks it is using a little over half its plant. In the same year, about a third of manufacturers spent money adding more plant.
Sometimes that is exactly right — a genuine bottleneck, a new product, a market that has moved. Often it is not, and the tell is simple: if you cannot say where the missing 44% went, in hours, you are buying capacity to solve a problem you haven't diagnosed. New plant added to an undiagnosed loss usually reproduces the loss at larger scale, and now with more depreciation attached.
There is also a national debate running on this exact point — Zimbabwe's financial press has been arguing through 2026 about whether capacity utilisation is even the right measure of industrial health. The criticism has force. It is self-reported, the denominator is elastic, and it flattens very different situations into one percentage. But the answer to a soft number is not a different soft number. It is a measured one.
Where the missing time usually is.
When a line is measured properly for the first time, the losses almost always sort into the same six buckets — and the ranking is almost never what the shift supervisor predicted:
- Short stops nobody logs. Two minutes, forty times a shift. Individually beneath anyone's notice, collectively the single largest loss on most packaging and processing lines. They are invisible precisely because each one is too small to write down.
- Changeovers that take longer than anyone thinks. The scheduled figure and the measured figure typically differ by a wide margin, and the measured figure varies enormously between crews — which is useful, because it means the good result is already being achieved by somebody in your own plant.
- Reduced-rate running. The line is running, so it counts as uptime, but at eighty per cent of nameplate. No alarm fires for this and no log records it. It is often the largest single loss on continuous plant.
- Startup and shutdown ramps. Every stop costs more than its duration. A plant that stops six times a day pays six ramp-ups, and only the stops appear anywhere.
- Quality losses and rework. Production that consumed capacity and produced nothing saleable. Counted as output on the board and as loss in the accounts.
- Waiting — for material, for the forklift, for a decision, for the electrician. Usually organisational rather than technical, and usually the cheapest to fix once it has a number attached.
Combined into a single figure, this is what the manufacturing world calls OEE — availability multiplied by performance multiplied by quality. The term matters less than the discipline behind it: every hour the line did not produce is attributed to a named cause, automatically, without anyone deciding what to write down. A line at 90% availability, 90% performance and 98% quality is not running at 90-odd per cent — it is running at 79%, and the three multiply that way on every plant.
What the survey says, and what a line measures
| CZI 16th survey (2025) | Figure | What a measured line adds |
|---|---|---|
| Capacity utilisation | 55.9% (52.3% in 2024) | Where the other 44% went, shift by shift |
| Output / turnover / employment | +13% / +12% / +6% | Growth from capacity you already own |
| Firms investing in expansion | ~35% | Buying capacity before measuring the existing |
| Technology upgraders vs the rest | turnover +18.4% vs +8.4% | Measurement is where the upgrade starts |
| Sample | 388 formal manufacturers, ten provinces, 72% non-members | A national picture, not a members' club |
One line, four signals, a few weeks.
Putting a number on lost output does not require instrumenting the plant. It requires instrumenting one line properly:
- Count — good units and total units, taken from the machine where possible rather than from a tally sheet.
- Run state — running, stopped, or running slow, sampled often enough that a two-minute stop is a data point rather than a rounding error.
- Stop reasons — captured automatically where the machine reports a fault, and entered by the operator on a simple screen where it doesn't. The second kind is where the organisational losses reveal themselves.
- Energy on the same line — which converts everything above into cost per tonne, the number a board reads without translation.
Four to six weeks of that data is usually enough to change the conversation, because the ranking of losses is nearly always a surprise. The common outcome is that the expansion case gets smaller and better targeted — you find you need one machine rather than a line, or a changeover method rather than a machine.
The instrumentation itself is covered under instrumentation and control in Zimbabwe — much of it can usually be read from equipment you already own.
Capacity and lost output — common questions
What is Zimbabwe's manufacturing capacity utilisation?
CZI's sixteenth annual manufacturing survey, published in June 2026 and covering 388 formal manufacturers across all ten provinces, puts capacity utilisation at 55.9% for 2025, up from 52.3% in 2024 — the fourth-highest level since CZI began tracking the indicator. Government figures put the first quarter of 2026 at approximately 57%, against 47.7% a year earlier.
Is capacity utilisation a reliable measure?
Only roughly. It is self-reported, the definition of full capacity varies between firms, and a single percentage hides very different situations — a plant limited by demand and a plant limited by breakdowns can report the same number. That is why it works as a sector indicator and fails as a management tool. Measured production and downtime data does the job it cannot: it says which hours were lost and to what.
We know our problem is power, not the line. Does this still apply?
Then measure it and use it. Zimbabwe has run around 188 consecutive days without national load shedding, so the picture has changed from the years when a shedding schedule dominated planning — but an ageing distribution network still produces local faults. Instrumented lines log those interruptions as timestamped events with real production consequences attached, which is a considerably stronger basis for a discussion with your supplier, or for sizing standby plant, than a recollection that it happens often.
How long before the data is worth anything?
Four to six weeks on one line is normally enough to rank the losses reliably, because you need to see a representative spread of products, crews and shifts. The first week is often the most striking, since it usually reveals a loss category nobody had counted at all.
Will operators accept being monitored?
It depends entirely on what the data is used for, and that is decided in the first fortnight rather than by the technology. Where downtime data is used to argue for maintenance, spares and better changeover methods, operators tend to become its strongest advocates — it finally evidences what they have been reporting verbally for years. Where it is used to rank individuals, it stops being accurate almost immediately, because the reasons entered become the safe ones.
What does it cost to measure one line?
Driven by how many signals can be read from the machine you already have, whether power and network reach the line, and how many stop-reason inputs you want. We quote a written cost band after a remote assessment — a few photographs of the machine's control panel and its nameplate is usually enough to size it. Starting with one line is also the cheapest way to find out whether the rest of the plant is worth instrumenting.
Before you buy capacity, find out what you already have.
Tell an engineer which line you'd expand first. We'll tell you what it would take to measure it, and what four weeks of data would likely show.