Home / Insights / The real cost of unplanned downtime
Manufacturing & OEE · Original research 9 min read Published 19 July 2026

The real cost of unplanned downtime: what 190 field complaints reveal

By Frank Guo · Technology & Product Leadership, addanode

TL;DR — We analysed 190 verbatim complaints from plant engineers, maintenance teams and operators across public communities, review platforms and industry reporting. The picture on unplanned downtime is consistent: incidents are becoming less frequent but more expensive per event, with recovery times extending; the cost per hour runs to thousands of rand on ordinary lines and far more on cold chains; small faults become long stoppages through spares and skills gaps; and the biggest losses — changeovers, unrecorded minor stops, start-up rejects — never appear in manually-kept records at all. The common thread: what you can't see, you pay for.

About this research

This article aggregates a research base of 190 verbatim complaints and discussions collected in 2026 from public sources — engineering communities, product-review platforms (G2, HelloPeter, Trustpilot, Google reviews) and industry articles — clustered into 189 recurring pain points and scored for severity and frequency. Of those clusters, 32 concern maintenance and downtime directly, rising to 39 when processing, bottling and packaging line failures are included. Quotes below are reproduced verbatim and anonymised; sources are on file. Where a quote itself cites a statistic, we attribute it as reported rather than as our own measurement.

Finding 1: downtime is consolidating — fewer events, each one worse

The most strategically important pattern in the base contradicts the comfortable story. Plants are not necessarily having more breakdowns — but each one hurts more. As one unattributed industry analysis in our research corpus reports — directional, not audited: "monthly incident frequency has fallen 40% since 2019, yet annual downtime costs per FMCG plant have more than doubled over the same period. Plants running at higher utilization have less capacity to absorb lost time, and the higher cost of energy, labor, and materials wasted during each stoppage compounds the financial impact of every event."

A second source in the base puts the same trend in one line: "incidents are becoming less frequent but more expensive per event, recovery times are extending."

The mechanism matters for anyone budgeting maintenance: higher utilisation means no slack to catch up in; leaner teams mean slower diagnosis; costlier energy and materials mean every idle hour burns more. Reliability programmes that measure success by incident count are looking at the number that's improving while the number that matters deteriorates.

Finding 2: the per-hour and per-event arithmetic

Across the base, the cost language is consistent: for ordinary production lines, unplanned downtime is discussed in thousands per hour; on constrained or high-value lines, far more. The extreme case is temperature-controlled product, where one industry source in our corpus (unattributed) reports: "Product loss begins immediately when temperature control fails, with refrigeration failures running from $50,000 to over $500,000 per incident."

One concrete South African-relevant example from the base shows how mundane the trigger can be: "a poorly configured VFD drive caused a 6 hour outage. This not only burned up precious hours, but also put a strain on the budget." Six hours, one drive parameter. Your own number — hourly volume × contribution margin on the constraint — is the first figure any downtime business case needs, and our guide to what an OEE system costs shows how quickly monitoring pays against it.

Finding 3: small faults become long downtime through spares and skills

A cluster of complaints describes the same escalation: the fault was minor; the stoppage wasn't. In one engineer's words: "No spares or training turns a small fault into long downtime." Related complaints describe spare-parts systems that can only be searched by internal part numbers (so the right part can't be found under pressure), and low-cost generic parts whose hidden reliability risks surface later as repeat failures.

The implication is uncomfortable but useful: a meaningful share of downtime cost is not a maintenance-engineering problem at all — it's an information problem (which part, where, what's degrading) meeting a logistics problem (is it on the shelf). Condition data that converts "it failed" into "it's failing, order the bearing now" attacks the escalation directly — which is the case our condition monitoring work is built on.

A small fault, and a long stoppage The complaints describe the same escalation every time: the fault was minor and the stoppage was not. A meaningful share of downtime cost is not a maintenance-engineering problem at all — it is an information problem about which part is degrading meeting a logistics problem about whether that part is on the shelf. Condition data attacks step two, which is what stops the escalation 1 The fault itself Minor: a drive parameter, a bearing, a failed sensor. → Minutes of work 2 Which part is it? An information problem — which part, where, and what is actually degrading. → Not in the manual 3 Is it on the shelf? A logistics problem, in a store searchable only by internal part number. → Often not 4 Who can fit it? A skills problem, on this shift, at this hour of the night. → Not always on site 5 The stoppage Hours rather than minutes. One documented case ran six hours from a single drive parameter. → The real number
A small fault, and a long stoppage An information problem about which part, meeting a logistics problem about the shelf. Condition data attacks step two 1 The fault itself Minor: a drive parameter, a bearing, a failed sensor. → Minutes of work 2 Which part is it? An information problem — which part, where, and what is actually degrading. → Not in the manual 3 Is it on the shelf? A logistics problem, in a store searchable only by internal part number. → Often not 4 Who can fit it? A skills problem, on this shift, at this hour of the night. → Not always on site 5 The stoppage Hours rather than minutes. One documented case ran six hours from a single drive parameter. → The real number
The escalation described in the finding above, step by step. Most of the cost is added after the fault, not by it — which is why condition data that names the degrading part attacks the second step rather than the first.

Finding 4: the biggest losses never make the log

When plant teams in the base describe where availability actually leaks, the named culprits are rarely dramatic breakdowns: "The most common causes are extended changeover duration, unrecorded minor stops on packaging lines, and startup rejects after each product transition. Food-specific requirements — allergen sanitation, HACCP monitoring, and mandatory cleaning — add OEE loss."

Note the word unrecorded. These losses share one property: they are invisible to manually-kept records. Nobody logs a 40-second jam; changeovers get written down as one vague block; start-up rejects vanish into scrap counts. That invisibility is why they persist — and it's the argument we develop fully in why nobody trusts your production dashboard: the losses you can't see are the ones you never work on.

What the plants themselves measure

Across the base, the metrics named by plant teams form a consistent stack: OEE, throughput, unplanned downtime, changeover duration, First Pass Yield and batch rejection rates. The gap isn't knowing what to measure — it's that the measurements are assembled by hand, after the fact, from memory. Machine-sourced capture closes that gap on any line, including old lines with no PLC, and in South Africa it has to keep recording through load shedding — outages are precisely when restart losses cluster.

What this means for a South African plant

  • Re-baseline your downtime cost annually. If the consolidation trend holds, the per-event cost you calculated two years ago is stale — and understated.
  • Judge reliability by cost and recovery time, not incident count. Fewer, longer, costlier events can hide inside an improving incident metric.
  • Attack the escalation, not just the fault. Spares findability and early-warning condition data shorten the path from fault to fix.
  • Instrument the invisible buckets. Changeover duration and minor stops only shrink once they're machine-timestamped facts.

Frequently asked questions

What does unplanned downtime cost per hour in manufacturing?

Field discussion consistently puts it at thousands of rand per hour on ordinary lines and tens of thousands on bottleneck lines, with temperature-controlled product the extreme case — unattributed industry reporting in our base cites $50,000 to over $500,000 per refrigeration failure, directional only. The only number that matters for your business case is your own: hourly volume × contribution margin on the constraint.

Why are downtime costs rising if breakdowns are getting rarer?

Because the cost per event is climbing faster than frequency is falling: higher utilisation leaves no slack to absorb lost time, leaner teams stretch recovery, and costlier energy and materials make every idle hour burn more. Unattributed reports in our base describe incident frequency falling while annual downtime cost per plant rose — directional, not audited.

What are the most overlooked causes of OEE loss?

The ones no one logs: extended changeovers, unrecorded minor stops on packaging lines, and start-up rejects after product transitions — plus, in food plants, the OEE cost of allergen sanitation and mandatory cleaning. They persist precisely because manual records can't see them; machine-timestamped capture is what makes them workable.

How was this research compiled?

From 190 verbatim public complaints and discussions collected in 2026 across engineering communities, review platforms and industry articles, clustered into 189 recurring pain points and scored for severity and frequency. Quotes are reproduced verbatim and anonymised; statistics inside quotes are attributed as reported by their sources, not as our own measurements.

Know your downtime number — before it grows again.

Tell us about the line that hurts most. We'll scope machine-sourced downtime capture and condition monitoring that turns these findings into your own numbers — through load shedding, on the machines you already own.

Book a Discovery Call See OEE management →

Related reading

8 min read

What does an OEE system cost?

Indicative ZAR bands per line, and payback.
8 min read

Why nobody trusts your dashboard

Manual entry kills boards; machine data revives them.
Solution

Asset & condition monitoring

Catch the failing bearing before the stoppage.

Primary sources