Nearly every organization of any size has a leadership competency model. Far fewer have one that a working manager could recite, and fewer still have one that changed who got promoted last year. The document exists, it was expensive, and it sits in a folder.
The waste is not trivial. Training magazine's 2025 Training Industry Report put total US training expenditure at $102.8 billion, with management and supervisory training taking 13 percent of budgets. That is the largest single allocation alongside compliance. A competency model is supposed to be the thing that directs that spend at something specific.
The reason to bother is that the target genuinely matters. Gallup analyzed 27 million employees across more than 2.5 million work units. It concluded that managers account for at least 70 percent of the variance in employee engagement scores across business units. If leader behavior carries that much of the difference, defining which behavior you want is a reasonable use of executive time. Most models fail at execution, not at the premise.
What the model is for, and what it is not for
A leadership competency model is a shared description of the behavior your organization needs from people who lead. It has to be specific enough that two managers watching the same person reach the same judgment.
That is its only job. It is not a development curriculum, not a values statement, not a performance rating scale and not a description of your current leaders. Models drift when they are asked to be all five, because each audience pulls the language in a different direction. HR wants it comprehensive. Learning wants it teachable. Executives want it aspirational. What emerges satisfies everybody and directs nothing.
The clarifying question to hold throughout: would this model help a hiring manager choose between two credible candidates? If the answer is no, the model will not affect any other decision either.
Start from the strategy, not from a library
The usual starting point is a vendor library of eighty or so competencies, from which a committee selects. This process reliably produces a model that would suit any company in any industry, because the library was written to suit any company in any industry.
Start instead with two questions about your own situation. What does this organization have to become good at over the next three years that it is not good at now? And where did leadership behavior cost us something real in the last eighteen months?
The second question is the productive one. Go through the specific incidents: the decision that took seven months, the two functions that stopped speaking, the acquisition that lost its best people. Behind each is a leadership behavior that was absent or an unhelpful one that went unchecked. Those are your competencies, and they arrive already attached to evidence.
Use a library afterwards, to sharpen the wording and check you have not invented a construct that already has a better name. Do not use it to generate the list.
Short enough to be remembered
Length is where most models die, and it fails before the content does. A model nobody can recall cannot be applied in the moment a decision is being made, and decisions are the only place a model has any effect.
The test is blunt. Ask five executives to name the competencies without looking. If they cannot, the model does not exist in any meaningful sense, whatever the intranet says. In practice that means a single digit number, clustered into three or four groups if you need more granularity beneath them. Clustering works because people remember three groups of four far more easily than a list of twelve.
Models balloon for an understandable reason: every executive consulted adds the one thing they care about, and nobody has the standing to remove anything. The discipline required is subtractive and it has to be enforced by whoever owns the model. A competency that is genuinely important to one function belongs in that function's own requirements, not in the enterprise model.
Write behavior, not adjectives
“Strategic thinking” is not a competency. It is a category heading that two reasonable people will apply to entirely different conduct.
Each competency needs behavioral descriptions at the levels where it is used, written so that a reader could confirm or deny whether they occurred. Compare “demonstrates strategic thinking” with “when presenting a proposal, states the option that was rejected and the condition under which the decision would be revisited.” The second is observable, teachable and assessable. The first is a Rorschach test.
Three or four proficiency levels are usually enough, and the differences between them should be qualitative rather than adverbial. If level three is level two with the word “consistently” in front of it, the levels are not describing different behavior and raters will ignore them. This is the same discipline that decides whether a talent review produces a real argument or a filled in grid. Both rest on whether the definition is specific enough for two people to apply it identically.
Wire it into three decisions
A model survives only where it changes a decision that has consequences for someone. Pick three and build the artifacts properly, rather than attaching the model loosely to eleven processes.
- Hiring and internal selection. Interview guides with behavioral questions per competency, a scoring rubric, and a debrief structure that requires evidence per competency rather than an overall impression.
- Promotion. The promotion case explicitly evidences the competencies at the target level. This connection does more than the other two combined, because employees learn what is rewarded by watching who gets promoted, not by reading the framework.
- Development planning. Each competency maps to a small number of real work experiences rather than courses. The assignment that builds it, not the workshop that describes it.
Deliberately leave some processes alone at first. Attaching the model to performance ratings in year one is tempting and usually a mistake. It converts a development language into a pay conversation, and people start gaming the definitions immediately.
Keep it alive, and know when to retire parts of it
Models go stale quietly. The business model changes, the competency list does not, and within three years it is describing the leadership a former version of the company needed.
Set review triggers rather than a review calendar. A change of strategy, a significant acquisition or a shift in operating model all qualify. So does a pattern of strong performers failing at the next level, which says the model is describing the wrong thing. An annual review on principle usually produces cosmetic edits nobody asked for.
When you revise, remove something. A model that only ever grows is a model nobody is willing to own.
What is actually hard about this, and what is merely long
Drafting, consultation and formatting are laborious but straightforward. The genuine difficulty is elsewhere, and it is political rather than technical.
Naming a standard your current leaders would fail
An honest model describes the leadership the strategy requires, which will not match the leadership you currently have. Writing it down creates a visible gap at the top of the house. Most models quietly resolve this by describing existing leaders, which makes them safe and useless.
Choosing between the company you are and the company you are becoming
A model built for the current operating model will be obsolete on arrival. One built entirely for the future state will fail to describe anybody, which costs it credibility in year one. Getting the mix right is a judgment call, and it has to be made deliberately rather than averaged out by a committee.
Holding the line on a leader you cannot afford to lose
The model's credibility is decided the first time a commercially important leader clearly breaches it. Whatever happens next is what the organization will believe the model means. This is where a competency framework stops being an HR artifact and becomes an instance of actual culture change, with all the difficulty that carries.
Resisting the request to make it comprehensive
Every review cycle brings a reasonable case for one more competency. Each addition is individually defensible and collectively fatal. Saying no repeatedly, to senior people, on grounds that sound like minimalism, requires a mandate that most model owners are never given.
Gather this before you commission anything
Four pieces of evidence will tell you whether you need a new model, a repair, or neither.
First, the last twelve promotion decisions at the level in question, and the actual reasons recorded. If those reasons do not reference the current model, the problem is adoption rather than content, and a rewrite will change nothing.
Second, the recall test: five executives, no notes. Third, the specific incidents from the last eighteen months where leadership behavior cost money, time or people. Those tell you what the model would have to address to be worth having. Fourth, the exit interviews of leaders who left voluntarily in the last year, which frequently describe the behavior your model claims to prohibit.
Bring those four to the decision and it usually makes itself. Where the evidence points to a genuine redesign, our organizational development work starts from your own incidents rather than from a library. That is the difference between a model that gets used and one that gets filed.
Frequently Asked Questions (FAQs)
How many competencies should a leadership model contain?
Few enough that the executive team can name them without looking, which in practice means a single digit number. Where more detail is needed, cluster the competencies into three or four groups, because people recall groupings far better than lists. Models that run past fifteen or twenty are typically applied by nobody outside the HR team that wrote them.
What is the difference between a competency model and a skills framework?
A skills framework describes what someone can do, usually technical and relatively easy to verify. A competency model describes how someone behaves, particularly under pressure and ambiguity, which is what distinguishes leaders at senior levels. The two answer different questions and organizations generally need both, kept separate.
How do you write behavioral anchors for each proficiency level?
Take real examples of the behavior at different levels from people in your own organization, then strip the identifying detail. The anchor should describe an action in a situation, not a trait with an intensity adverb attached. If the only difference between two levels is a word such as consistently or effectively, the levels are not describing different behavior.
How often should a leadership competency model be updated?
On triggers rather than on a schedule: a strategy change, a significant acquisition, a shift in operating model, or a run of strong performers failing at the next level. Calendar driven reviews tend to produce cosmetic changes that cost credibility. When you do revise, remove at least as much as you add.