The Capacity Illusion

1. Accredited capacity is extremely concentrated

Ranking every laboratory by how many accredited scope lines it holds:

One percent of the laboratories carry a sixth of the accredited capacity. Ten percent carry half of it.

That alone would be a mildly interesting statistic, and an easy one to get wrong. Different accreditation bodies publish their scopes at different levels of detail, so a concentration figure computed across all of them at once could simply be measuring which body writes longer documents. So we ran the test that mattered.

2. The pattern survives in every registry

We recomputed the concentration separately inside each registry, where publishing conventions are constant.

Eleven registries. Eleven different national systems, languages, industrial bases and regulatory traditions. In every one of them, the top tenth of laboratories holds between 37.1% and 61.8% of the accredited capacity, clustering tightly around half.

This is not an artifact of one body’s paperwork. It appears to be a structural property of how accredited testing capacity forms. The laboratories at the head of the distribution are also exactly who you would expect them to be on independent grounds: large, well-known testing organisations holding many separate certificates each.

3. The typical accredited lab is a narrow specialist

The distribution behind that concentration is severe.

• The median laboratory holds just 4 accredited scope lines.
• 766 laboratories, roughly one in fifteen, are accredited for exactly one thing.
• Only 78 laboratories out of 11,497 hold 100 or more accredited scope lines.
• The largest single laboratory in the corpus holds 346.

The average (9.1) is more than double the median (4), the signature of a long tail: a small number of very broad laboratories pulling the mean away from the typical case.

The mental model of “a testing lab” as a large facility that can handle whatever you send it describes the top few percent. It does not describe the median accredited laboratory, which is a specialist doing a small number of things, often extremely well, and often only those things.

4. What is genuinely widely available

Not everything is scarce. Some capabilities are held by large numbers of laboratories across many countries. These are lower bounds: the true figures are higher.

Drinking water is the most widely accredited sample type in the corpus. If you need routine drinking-water analysis, you are shopping in a genuinely competitive market with well over a thousand accredited options, and you should expect competitive pricing and short lead times.

Water-testing capacity by country, again as lower bounds within the registries we read: Germany 655, Italy 604, United Kingdom 408, Poland 341, France 326, United States 244. These counts reflect the eleven registries we read, not national totals. Countries are not comparable on this measure.

5. What this means for buyers, labs and regulators

If you are buying testing. The headline count of laboratories in any directory, including ours, is not the number of options you have. For a common test it may understate your options. For anything specialised, the field narrows fast, and the broad laboratories that can handle a multi-part programme are a small minority. Ask early whether a lab’s accreditation actually covers your matrix and your parameter, not just whether it is accredited.

If you run a laboratory. Being in the long tail is not a weakness. It is the normal state of the market, and half the sector’s accredited capacity sits in the top decile precisely because most labs deliberately do not chase breadth. But it does mean discoverability is the constraint. A specialist lab is invisible to a buyer searching by capability unless its scope is published, structured and findable.

If you regulate or accredit. The concentration is remarkably consistent across very different national systems. That consistency is itself worth studying: it suggests the shape of accredited capacity is driven less by national policy than by the economics of maintaining accredited scope.

What to take away

• Ten percent of accredited laboratories hold half of all accredited testing capacity, in every national registry we examined.
• The typical accredited laboratory is a specialist with four accredited tests, not a general-purpose facility.
• A directory’s laboratory count is not the number of options you have for your test.

Methodology

What the corpus is. 104,612 accredited scope lines from 16,751 certificates held by 11,970 organisations across 24 countries, as published by eleven national accreditation bodies: DAkkS (Germany), A2LA (United States), ACCREDIA (Italy), COFRAC (France), PCA (Poland), UKAS (United Kingdom), SWEDAC (Sweden), RvA (Netherlands), INAB (Ireland), DANAK (Denmark) and FINAS (Finland).

Data vintage. The scopes analysed here were collected in August 2026. Accreditation scopes change continuously, so these figures describe the registries as published at that time.

This is not a global study, and we do not claim it is. Eleven registries is not the world. There is no data here from China, Spain, India, Switzerland or Belgium, among many others. Every finding is scoped to the registries named above.

Every number is a lower bound, by design. We count a laboratory toward a capability only where its published scope states that capability explicitly. Ambiguous entries are excluded rather than assumed. So every figure here is a floor: we report “at least N laboratories”, never “only N”.

Why we report abundance rather than scarcity. Abundance is the direction this evidence supports. A capability held by a thousand laboratories is robust to any single scope being read imperfectly; a claim that only a handful of labs can do something is not. We publish the claims the method can carry.

Two things this report deliberately leaves out. We do not report trends, because our collected data does not yet include reliable dates. We also do not report how many laboratories serve external customers versus in-house needs: we audited that classification while preparing this report and set it aside for rework rather than publish a number we might later have to retract. Both will return when the data supports them.

Definitions