What nobody has measured: the case for an African maintenance benchmark
This report draws together the publicly available data on maintenance in African industry. There is less of it than there should be — and the shape of the gap tells you what a benchmark would need to contain.
- No published benchmark measures maintenance performance in African industry from operating data — what exists are proxies drawn from firm surveys and infrastructure statistics.
- The closest proxy is outage data: across 137 countries, sales lost to electricity outages are estimated at $82 billion a year, with self-generation adding $65 billion. [1]
- The World Bank records that data on the African engineering and technician skill mix is seriously lacking — the workforce is uncounted, and so is its work. [2]
- Industrial capacity is expanding through special economic zones and cross-border corridors, so the population a benchmark would measure is growing before the measure exists. [3]
A benchmark is a simple thing to define and a demanding thing to build: a defined population of operations, a consistent set of measures, and a method that can be repeated so the numbers can be compared over time. Manufacturing productivity has such series in most regions. Maintenance in African industry has none, and this report assembles what public data does exist so the size and shape of the gap is visible.
Everything below is drawn from published institutional sources, each cited. Where a figure could not be traced to the body that produced it, it does not appear.
What can be measured today: the proxies
The best-evidenced proxy for maintenance conditions is electricity reliability, because firms report its consequences in surveys and regulators publish its indices. World Bank researchers, analysing survey responses from more than 143,000 firms across 137 countries, estimated annual sales losses from power outages at $82 billion, the cost of running self-generation at $65 billion, and total losses from underutilised capacity at $151 billion — presenting all three as lower-bound estimates. [1]
The companion indicators track, by country and year, the share of firms experiencing outages and the value lost as a share of sales. [4] [5] These are production measures, not maintenance measures — but every interruption they count is also a restart transient, a thermal cycle and a duty hour on standby plant. The maintenance share of those losses is not separately published anywhere, which is precisely the kind of number a benchmark would exist to produce.
The workforce is also uncounted
Maintenance performance is inseparable from the people who deliver it, and here the data gap is documented rather than inferred. Reviewing engineering capacity in sub-Saharan Africa, the World Bank cited the ILO-recommended ratio of one engineer to five technicians to twenty-five craftspeople — and noted that data on the actual skill mix in African countries is seriously lacking. [2]
What is measured is the training pipeline, and it is thin: vocational programmes account for 1.8% of lower-secondary enrolment in sub-Saharan Africa and under 15% at upper secondary, with enrolment declining at both levels between 2000 and 2022. [6] A workforce that cannot be counted, produced by a pipeline that is measurably narrowing, is maintaining an asset base that is measurably growing.
Why the gap matters now
Industrial capacity is arriving at pace. Special economic zones and industrial parks are being established across the continent to attract manufacturing investment, alongside ports, rail and power projects intended to bring down the transport and energy costs that have held competitiveness back; cross-border projects such as the DRC–Zambia battery corridor are moving from announcement toward execution. [3]
Every one of those facilities will be maintained under a regime nobody is measuring. The practices an operation establishes in the first years of an asset’s life tend to persist for its whole life — which means the cost of the measurement gap compounds: each year without a benchmark is a cohort of new plant whose baseline is never recorded.
What the absence costs
- Budgets are set by convention — last year plus inflation — because there is no external reference for what maintenance should cost per asset class in these operating conditions.
- Repair-versus-replace decisions are argued from opinion, since no regional cost-per-asset distribution exists to locate an asset against.
- Insurers, lenders and investors price maintenance risk blind, which in practice means pricing it conservatively — a cost carried by every operator, including the good ones.
- Policy debates about industrialisation proceed without the one number that links capital investment to whether the capital keeps working.
What a benchmark would need
The building blocks already exist in public form. The measurement definitions are published: an independent professional society maintains standardised maintenance and reliability metrics expressly so organisations can measure consistently and compare validly, and the international asset-management standard frames what a documented maintenance programme must contain. [7] [8] The survey method exists too — the firm-level enterprise surveys behind the outage estimates show that operational data can be collected at scale in these markets. [1]
What is missing is the intersection: nobody has applied standardised maintenance metrics to a defined African industrial population and published the result. The first credible version would not need to be large. A few hundred operations, three or four metrics with published definitions, one region, repeated annually — that would be more than exists today.
What an operator can do meanwhile
The absence of an external benchmark does not prevent an internal one. An operation that records cost per asset, downtime by cause and completion against schedule — using published metric definitions rather than house conventions [7] — builds a private baseline that any future benchmark can be compared against. It also builds the discipline that makes such data trustworthy, which the survey literature shows is the harder half of the problem.
The first version of any benchmark is a baseline. The only way to have one in five years is to start recording now.
References
- World Bank, Underutilized Potential: The Business Costs of Unreliable Infrastructure in Developing Countries, Policy Research Working Paper 8899. documents1.worldbank.org
- World Bank, Africa's Need for Engineering: Sub-Saharan Africa seriously lacks engineers, technicians and craftspeople. documents1.worldbank.org
- Institute for Security Studies, African Futures: Africa Manufacturing Forecast. futures.issafrica.org
- World Bank, Firms experiencing electrical outages (% of firms), indicator IC.ELC.OUTG.ZS. data.worldbank.org
- World Bank, Value lost due to electrical outages (% of sales for affected firms), indicator IC.FRM.OUTG.ZS. data.worldbank.org
- ILOSTAT, Empowering Africa's youth: the need for more vocational training and on-the-job learning. ilostat.ilo.org
- Society for Maintenance and Reliability Professionals, Best Practices, Metrics and Guidelines. smrp.org
- ISO 55000:2024, Asset management — vocabulary, overview and principles. iso.org
Related reading
The most useful numbers are the ones you record yourself
These reports draw on the best public data available. A live demo shows how SaharaDesk builds the same evidence — cost per asset, downtime by cause, compliance on record — from your own operation.