Why skills visibility fails at 300 people
Every function can describe its roles. Almost none can describe its skills. The break happens at a fairly predictable size.
Future skills2 min
July 2, 2026
The question
Ask an executive team how many people they employ and you get an exact number. Ask them what those people can do and you get anecdote.
This is not carelessness. It is a structural break, and it happens at a fairly predictable size.
What is actually going on
Below roughly a hundred people, skills visibility is solved by memory. Somebody knows who can do what, and that person is reachable in a morning. The system is a human being and it works well.
Somewhere around three hundred, that stops working, and it stops working suddenly rather than gradually. The person who held the map has moved on, or the organization now spans enough functions that no single person covers it. Nothing is announced. The map simply stops being accurate, and because it used to be reliable, people keep trusting it.
The first visible symptom is almost never described as a skills problem.
It shows up as hiring: roles posted externally that could have been filled internally, because nobody could see the internal candidate. It shows up as redeployment: when a role changes, the organization cannot tell whether the person in it could move somewhere else, so it defaults to the expensive answer. And it shows up as duplicated development: two functions independently paying to build the same capability.
Each of these gets diagnosed as a recruitment problem, a change problem or a budget problem. They are the same problem wearing three coats.
What this means for you
The test is quick. Pick a capability that genuinely matters to your strategy. Ask how many people in the organization have it and at what level.
If the answer takes more than a day, or arrives as a list of names somebody assembled from memory, you have found the break. Note that the answer being slow matters more than the answer being wrong: slow means there is no system, only recall.
The instinct at this point is to buy something. Resist it for one quarter. A platform makes an existing vocabulary searchable, and if you do not have one yet it will store your inconsistency very efficiently.
Where I can help
What actually fixes this is a shared vocabulary for capability, applied consistently enough that two managers describing the same person would use the same words. That sounds soft and is the hardest part.
I have built these inside organizations from 1,200 to 13,000 people, and connected them to succession, performance and hiring so they stay in use. On one enterprise build, succession coverage went from 20 executive roles to 200 leaders, with 90 percent of critical roles supported. That was not a technology outcome. It came from finally being able to describe what people could do.
Questions on this
Do we need a skills platform to fix this?
Not first. A platform makes an existing vocabulary searchable. If two managers describe the same person in different words, the platform will faithfully store both and you will have paid for a more expensive version of the problem.
How granular should a skills taxonomy be?
Coarse enough that a manager will actually use it. Twenty to thirty capabilities across the organization is usually workable. Several hundred is a research project that quietly dies.
How long does this take?
Building a shared vocabulary and getting it used consistently is a few months of real work, not a workshop. Most of the effort is in the adoption, not the design.
Read next
The jobs that changed quietly
Most roles are not being replaced. They are being re-weighted, and the re-weighting almost never arrives with an announcement.
Article / 2 minWorkforce planningThe onboarding audit, on one page
Twelve questions that will tell you where new people are losing their first ninety days. Free to take and use inside your own organization.
Tool / 2 minRecognise this in your own organization?
Thirty minutes, no material to prepare. If the answer is that you do not need me, that is a useful outcome too.