A competency matrix from manager interviews: example and steps
Build a competency matrix from manager interviews using workplace situations, observable behaviors, levels and evidence. Includes a sample table and review template.
A competency is useful when you can see it at work
“Shows leadership” and “communicates well” sound convincing, but different managers interpret them differently. Ask for episodes: whom did the employee warn about a risk, which options did they propose, and how did they document the decision? Those actions are the basis of the matrix.
A matrix describes expectations for a particular role and context. Add the type of task, independence and authority to each indicator. One story can suggest a level, but it needs further checking before you apply it broadly.
Build levels from examples of work
Set the scope of the role
Record the tasks, decisions and working relationships the matrix will cover. Ask for two recent successes and two difficult cases. “What did the person do?” usually gets more useful detail than “What qualities did they have?” Keep timestamps and separate what happened from the manager’s evaluation.
Code observable behavior
Break the transcript into short actions: clarified the criterion, involved the right person, compared options, warned about risk. Group only similar actions. Tool knowledge, motivation and outcomes require different ways of checking them.
Build levels of difficulty
Describe how scope, uncertainty and independence change. Someone at a basic level may follow an established procedure. At the next level, they can choose an approach and explain why. At a higher level, they can set rules for recurring cases. Distinguish the levels by observable behavior.
Test with counterexamples
Give the draft to several managers and ask them to match it to real cases. Look for behavior that fits two levels and expectations people can’t meet with their current authority. Record the revision, date and open questions after making changes.
Risk management as a competency for a coordinator
This is a fictional training example. The names, circumstances and outcomes are included solely to explain the method.
A strong coordinator does more than report a delay. They show what decision needs to be made today and what will change for each team.
The fictional manager describes actions: notice a signal, assess its impact, say when a decision is needed and identify whom to tell. You can check those actions in a work sample.
At a basic level, it may be enough to flag the risk promptly through the usual channel. At the next, suggest options with consequences. A higher level might include setting up a recurring risk review. This is an example. Check it against the authority that comes with the role.
| Level | Observable behavior | Evidence |
|---|---|---|
| Basic | Reports risks through the usual process | Risk record and timestamp |
| Independent | Compares options and consequences | Review of a work case |
| System-level | Addresses a recurring cause | Changed process and outcome |
| Unconfirmed | Only a general assessment is available | Another example is needed |
Competency card for review
Use one card per group of behaviors. Keep broad terms for short labels.
Role and work task: Competency: Situation where it’s needed: Observable action: Basic level: Independent level: System-level: What this competency doesn’t cover: Interview example and timestamp: Counterexample: Who reviewed the wording: Version and date:
Check that people can apply the matrix consistently
- Each level uses verbs and observable actions.
- The context shows when the behavior is needed.
- One story hasn’t been made into a general rule.
- Managers agree on the boundaries between adjacent levels.
A closer look
How many competencies should we include?
As many as the role’s key tasks require. Start with a small set and remove overlapping definitions. There’s no universal number.
Can we use the matrix to assess staff right away?
Check whether people understand the indicators and agree on a common assessment process first. Interviews suggest expectations that still need to be tested.