The Case That It’s a Volume and Complexity Problem
The intuitive explanation is also a defensible one: more output, more variation in how that output is produced, and more interdependent steps in producing it, simply create more surface area for defects to occur and more difficulty catching them before they matter. Jay Galbraith’s information-processing view of organisation design supports this directly — as task interdependence and volume rise, the information-processing burden on any quality system rises correspondingly, and a quality system designed for a smaller, simpler operation will structurally struggle to keep pace unless it evolves alongside the operation it is meant to monitor. On this account, quality control becomes harder at scale for essentially the same reason coordination becomes harder at scale: there is more to track, more interdependency between the parts, and less of it is directly observable to any one person.
The Case That It’s a Measurement and Feedback Problem
A different, equally well-evidenced account locates the difficulty elsewhere: not in the volume of what is being produced, but in what happens to the reliability of the measures used to monitor it. V. F. Ridgway’s research at Cornell’s Johnson Graduate School of Management found that performance and quality measures reliably produce behavioural distortion — people and teams adjust their behaviour toward what is measured, not necessarily toward the underlying outcome the measure was meant to represent. Ridgway’s own examples are instructive here: interviewers measured on volume interviewed faster and placed fewer people; investigators given a fixed monthly quota of cases selected easier cases as the deadline approached. Quality measures are exactly this kind of measure, and this account holds that quality gets harder to see, not necessarily harder to achieve, as an organisation scales — because scale typically means quality measurement becomes more indirect, more removed from direct observation, and therefore more exposed to exactly the kind of distortion Ridgway documented.
On this account, the core difficulty is not that there is intrinsically more to go wrong. It is that the people closest to the work are increasingly measured and evaluated by proxies for quality rather than by quality itself, and proxies are what get optimised under pressure — sometimes at the direct expense of the thing they were meant to represent.
Why the Disagreement Persists
Neither explanation fully displaces the other, which is why the disagreement persists rather than resolving. The volume-and-complexity account is clearly true as far as it goes — a larger, more interdependent operation genuinely does have more places for defects to originate. But it does not, on its own, explain why some large organisations sustain excellent quality while others do not, at comparable scale and complexity — a gap the measurement-and-feedback account addresses directly, by pointing to differences in how reliably quality is actually being measured and fed back to the people who can act on it.
Equally, the measurement-and-feedback account does not fully explain why quality problems specifically correlate with scale, rather than appearing at random regardless of size — a correlation the volume-and-complexity account explains more directly, since genuine complexity does rise with scale in a way that measurement distortion, in principle, does not have to.
What It Means for a Leader
The two accounts point toward different diagnostic questions, and distinguishing between them in a specific organisation is more useful than assuming one applies universally. If quality problems track closely with genuine increases in volume and interdependency — more products, more variants, more interdependent steps — the volume-and-complexity account is likely doing most of the explanatory work. If quality problems persist or worsen even where volume and complexity have been relatively stable, but the distance between where work happens and where it is measured has grown, the measurement-and-feedback account deserves closer attention. In most real organisations, both forces are present simultaneously and reinforcing each other — rising complexity increases the information-processing burden at the same time as it increases the distance between direct observation and formal measurement — which is precisely why quality control becoming harder at scale is so consistently reported and so rarely resolved by addressing only one side of it.
There is a further complication worth naming: the two forces do not just coexist, they actively reinforce each other in a way that makes the underlying problem harder to diagnose than either account alone suggests. Rising complexity creates more genuine opportunities for defects (Position A), while simultaneously stretching the distance between the work and its measurement, making those same defects harder to detect reliably (Position B). An organisation experiencing both simultaneously may see a quality metric that looks stable — because the measurement itself is degrading at roughly the rate that underlying defects are rising — right up until the gap between the two becomes large enough to surface as a visible failure, at which point it often looks sudden, even though both underlying trends had been building for some time.
A Third Consideration: Who Is Closest to the Defect
A further dimension, distinct from either the volume-and-complexity or the measurement-and-feedback account, is worth introducing directly: the physical and organisational distance between where a defect originates and where it is finally caught. In a small operation, the person best placed to notice a quality problem is often also the person doing the work — defect and detection sit close together, in time and in person. As an operation scales, these two roles frequently separate: the person producing the work and the person responsible for verifying its quality become different people, often in different parts of the organisation, sometimes separated by several intermediate steps. This separation is not itself a failure of either account examined above; it is a structural consequence of specialisation, the same dynamic Galbraith’s framework describes more generally. But it adds a genuinely distinct source of difficulty: even a perfectly designed quality measure, applied by someone with no direct exposure to the conditions under which the defect actually occurred, will tend to catch a narrower slice of what is actually going wrong than the same measure applied by someone closer to the work.
For how this pattern presents in other kinds of organisation, see other sectors.
Questions worth considering
- Do quality problems in this organisation track most closely with genuine increases in volume and complexity, or with growing distance between the work and how it’s measured?
- How direct is the current path between someone doing quality-relevant work and someone able to see the result of a quality measure based on it?
- Are quality metrics currently measuring the outcome that matters, or a proxy that has become easier to measure than the outcome itself?
More on this topic: Operations.
Your team is facing similar challenges?
Subscribe to The Riddle Ledger
Perspectives are sent as they are published, with no promotional email, and you can unsubscribe at any time.