The Problem Is Assembly, Not Capability
Most organisations do not lack the parts of a good capability system. They lack the assembly that makes the parts work as one. And the reason the parts do not work as one is specific: none of them was ever derived from the strategy. Each was built from the work as it already stood, and the strategy was laid over the top afterwards.
That is what assembly means here. Not tidying up the parts, but re-deriving them so the work is architected to the strategy. The full argument is at peoplecapabilitysystem.com.
What does the problem is assembly, not capability mean?
It means the components of a working capability system already exist in most mid to large organisations, disconnected from each other and from the strategy. The gap is not a missing part. It is that nothing was intended to the strategy, so there is no line running through any of it.
Capability frameworks exist. Competency frameworks exist. Skills taxonomies exist. The 70-20-10 model exists. Decades of engagement research exists. The psychosocial hazard duty exists. The competence requirements inside ISO 9001 exist. Each is well evidenced on its own.
What is missing is an architecture that derives them from the bets and connects them to each other, so that one design produces both the performance lift and the evidence that regulators separately ask for.
The inventory is already in the building
Take stock of a mid to large organisation.
There is a capability framework somewhere, from a project two years ago, in a slide deck. There are competencies inside old role profiles that have drifted from the work. There is a learning catalogue that maps to nothing. There is an engagement survey. There are position descriptions written by hand. There is a skills field in the HR system that nobody trusts.
Nothing on that list is missing. Everything on that list was built from the work rather than from the strategy.
Talent acquisition commissioned one thing. Learning commissioned another. Workforce planning built its own view because neither of the first two gave it what it needed. Each function got a structure that described the work it could see, in its own language, and none of them started from a bet.
Assembly is not mapping
This distinction is the whole argument, so it is worth being precise.
A mapping takes two structures written independently and draws lines between them. It produces a matrix that looks impressive in a steering committee and quietly stops being true within a year, because neither structure has any reason to move when the other one does.
An assembly derives each layer from the layer above it. The capabilities come from the strategic bets. The competencies come from the capabilities. The skills come from the competencies. Nothing was written independently, so there is nothing to reconcile and nothing to drift apart.
The test is simple. If a strategic bet changes, does anything below it move on its own? In a mapped architecture, nothing moves, because the connections are annotations. In an assembled one, everything below it has a derivation that now has to be revisited.
Why "we need a capability uplift" is usually the wrong diagnosis
The reflex when this becomes visible is to declare a capability problem and commission an uplift. I have been on the receiving end of that brief many times.
It is almost always the wrong diagnosis. The organisation does not have a shortage of capability material. It has a surplus of it, sitting in incompatible formats, none of it derived from a bet. Adding another framework to the pile increases the surplus and changes nothing about the direction it was built in. It is the failure mode that kills most capability frameworks, and commissioning a new one is the most reliable way of repeating it.
The assembled line has to be clear and accepted
An assembly produces a line. The line is only worth something if it has two properties.
Clear. Walkable in both directions. Point at any skill and name the bet it serves. Point at any bet and name the work that delivers it, in which roles, at what level.
Accepted. Leadership owns the bets as the actual bets. The people in the roles accept that the line genuinely runs through their work. An assembly nobody believes is a diagram, and a diagram does not survive a reorganisation.
The language has to hold still for assembly to work
You cannot derive layers you cannot tell apart.
Capability gets used to mean competency. Skill gets used to mean capability. Framework, model and matrix get treated as interchangeable when they describe three different things. Once the terms blur, the layers merge, and a merged structure cannot roll one layer up into the next.
This is why I keep insisting the conflation of these constructs is structural rather than pedantic. A system that cannot distinguish a competency from a skill cannot evidence one with the other, which means the roll-up stops working at the second step.
The precision exists elsewhere. The European Commission's ESCO definitions treat a skill as the applied use of knowledge to complete a task, narrower than competence. The SFIA Foundation levels skills against responsibility and reserves the language of competence for demonstrated practice. It is available. It is not being used.
Why assembly pays twice
Here is what makes assembly worth the effort rather than merely tidy.
Three separate obligations already ask you to hold versions of the same material. Clause 7.2 of ISO 9001 requires you to determine the competence necessary for work affecting quality, act on gaps, and retain documented evidence of competence. Lack of role clarity is a named psychosocial hazard in Safe Work Australia's model Code of Practice, and a clear, levelled role standard is a documentable control for it. Fair process in a capacity matter turns on whether a person was held to a clear expectation and given a real opportunity to improve.
All three want the same thing: a clear standard for the role, evidence of competence against it, and a record of the development offered to close the gap.
In a mapped architecture, each of those is assembled by hand, separately, every time. In an assembled one, all three fall out of the ordinary operation of the system. You build and run it so the work is architected to the strategy, and the evidence comes as a by-product rather than as three separate compliance projects.
The system informs those regimes. It does not replace a quality management system, it does not run a safety programme, and it does not conduct dismissals. That boundary should always be stated.
The honest status of the claim
Each component is individually credible, from McClelland on competence, through the skills-framework tradition, to the peer-reviewed evidence linking role ambiguity and engagement to performance.
A set of individually validated parts does not entail a validated whole. The interaction between the layers, which is exactly where the assembly claim lives, is precisely what the component evidence does not reach. So the assembled system is a proposition rather than a result.
The central test is whether the assembled system produces a better capability-to-performance outcome than the same components deployed separately. That is what the whole thing should stand or fall on, and it has not been run yet.
FAQ
What does "the problem is assembly, not capability" mean?Most organisations already hold the parts of a capability system. What they lack is an assembly that derives those parts from the strategy and connects them, so the missing thing is architecture rather than a component.
How is assembly different from mapping?A mapping draws lines between structures written independently, and it decays because neither moves when the other does. An assembly derives each layer from the one above, so nothing drifts apart.
If the parts already exist, why does anything need building?Because the parts were derived from the work rather than from the strategy. Building is the act of re-deriving them downward from the bets so they connect. It is design work, done once, then maintained.
Does assembly require new software?No. It requires the artefacts to live somewhere queryable rather than in slide decks, which the HR system you already run can usually do.
Is the assembly claim proven?No. The components are individually well evidenced. The assembled composite is a proposition, and the working paper sets out its propositions specifically so they can be tested or falsified.