Every second HR team we meet has a competency framework. Very few can tell us where it is used in a decision. It sits in a PDF, gets referenced in an appraisal form, and is otherwise ignored. The behaviours in it are usually well written. That is not the problem.
The problem is arithmetic. A framework with forty competencies across five levels requires a manager to hold two hundred combinations in their head while conducting a performance conversation on a Thursday afternoon. Nobody does that. So the manager falls back on the thing they already trust, which is their impression of the person.
How frameworks grow
Frameworks are built by committee and grow by accretion. Every function adds its behaviours, every consultant adds a level, and every year adds a theme that seemed important at the time. Nobody ever removes anything, because removal requires someone to say a colleague’s work is not needed.
The result is comprehensive and unusable. It is also expensive to maintain, and it creates a second problem: because the document is authoritative, disagreements about a person’s performance get argued in its vocabulary rather than in plain language. The framework stops being a tool and becomes a legal text.
The test is simple. Can a line manager recall the three behaviours that matter most for their team without opening the document? If not, the framework will not change a single conversation.
What a usable framework looks like
- Six to eight competencies, not forty. Grouped by what the role actually produces, not by HR taxonomy.
- Three levels at most. Enough to distinguish, few enough to remember.
- Observable language. “Gives feedback within 48 hours of the event” beats “demonstrates feedback orientation”. One can be seen. The other has to be interpreted.
- One page per role family. If it needs a workshop to explain it, it will not be used in a live conversation.
The observable-language rule does most of the work. A competency written as an adjective forces the manager to translate it into behaviour on the spot. A competency written as a behaviour lets them simply answer yes or no.
A framework is a tool for decisions, not a document for audits.
Where it earns its keep
A framework is worth building only if it is used in at least two of three places:
Selection. Structured interview probes that map to specific competencies. “Tell me about a time you handled a customer escalation” is not a probe. “Describe the last time you told a client something they did not want to hear, and what you said in the first thirty seconds” is a probe, and it maps to one competency.
Development. A gap analysis that produces a plan with a date and a named reviewer. If the output of the analysis is a list of training courses, the framework has become a catalogue.
Measurement. The same anchors scored before and after an intervention, so the before-and-after is comparable. This is where a framework stops being paperwork and starts producing evidence.
If a competency is not used in two of those three, it is decoration. Most of the frameworks we rebuild start as a framework document that nobody has opened in a year, and end up as a consulting engagement to rebuild it. It is also a candidate for deletion, and deleting it is the cheapest improvement most frameworks can make.
That is the logic behind BolderBrain’s gap analysis: the framework becomes a set of scored behaviours, the score becomes the baseline, and the baseline becomes the before-and-after. The framework is not a document the platform stores. It is the unit the platform measures.
For what happens after the framework is built, and how to tell whether the behaviour actually moved, see four levels of evaluation.
A worked example
A financial services client had a framework of thirty-two competencies across four levels. It had taken eight months and two consultants to build. When we asked twelve line managers to name the competencies their team was assessed on, four could name one, and none could name three.
We cut it to seven competencies, three levels, one page per role family. Each competency got three behavioural anchors, written with the managers in a single afternoon, using their own language. Then we wired it into three things and nothing else: the interview guide, the half-yearly review form, and the development plan template.
Eighteen months later the same seven competencies were still in use, with one addition. The measure we care about is not satisfaction with the framework. It is whether a manager can open a review conversation by naming the behaviour, and whether the employee recognises the description. In the first cycle after the change, 68 per cent of reviewed employees said the feedback described something they recognised doing. That is not a high bar. It was a large improvement on the previous framework, where the equivalent number was not collected at all.
The honest limit
A short framework cannot capture everything that matters about a person. It will miss judgement, temperament, and the accumulated trust that makes someone effective in a specific team. That is fine. The framework’s job is to make the discussable parts discussable, not to describe a human being.
There is also a real cost to changing one. Managers who learned the old vocabulary have to learn a new one, and the first cycle after a change is worse than the last cycle before it. Budget two cycles before judging whether the new version works, and do not rebuild it again in between.
Writing an anchor that survives a manager
An anchor is the sentence a manager reads before a review. Most frameworks write it as an aspiration, which is why it gets ignored.
Weak: “Demonstrates strong stakeholder management.” It cannot be observed, so it gets interpreted as agreement with the manager’s own view.
Usable: “Before committing a date to a client, confirms the delivery owner has seen the scope in writing.” A manager can answer yes or no without a discussion about what the words mean.
The test for any anchor: could two managers who dislike each other reach the same verdict after watching the same week? If not, rewrite it until they could. That single test does more for consistency than a competency calibration workshop.
What to do on Monday
If you are rebuilding a framework this year, cut it in half first. Then cut it again. Seven competencies with three observable anchors each will do more for capability than forty competencies that pass review.
The version your managers can repeat from memory is the version that will show up in a real conversation. Everything else is a document.