A competency framework is one of the most foundational investments an organization can make in its people strategy. Without one, performance reviews are subjective, career conversations are vague, and promotion decisions depend more on visibility than on merit. With a well-built framework, every manager in the organization is working from the same definition of what "great" looks like — which makes ratings consistent, development conversations concrete, and advancement predictable.

This guide covers the full process from scoping to rollout. It is designed for HR leaders and people managers who are building a framework for the first time, though many of the steps apply equally to organizations revisiting an existing framework that has grown stale.

Step 1: Define your job families and levels

Before writing a single competency, you need to establish the structure your framework will sit inside. This means two things: defining job families and defining career levels.

Job families are groups of related roles that share a common progression structure. Typical job families include Engineering, Product, Design, Sales, Marketing, Operations, Finance, and People. Within each family, roles may differ significantly in scope or specialization, but they advance through the same level structure and are evaluated against the same core criteria.

Career levels are the discrete stages of progression within each job family. Common level structures include:

L1 (entry/associate) → L2 (mid) → L3 (senior) → L4 (staff/lead) → L5 (principal/director)
IC1–IC5 for individual contributors with a parallel M1–M4 track for people managers
Associate → Senior → Lead → Principal → Fellow for technical roles

The specific labels matter less than the consistency: every employee should be able to look at the framework and immediately identify where they sit and what the next level requires. If you have a career ladder, this step is largely already done — you are building the competency detail layer on top of an existing level structure.

Step 2: Identify core and functional competencies

Competencies fall into two categories, and both need to be represented in your framework:

Core competencies apply to every employee regardless of role or level. They represent the foundational behaviors the organization expects from everyone: communication, ownership, collaboration, adaptability, and organizational values. Keep this list to 3–5 — if everything is a core competency, nothing is.
Functional competencies are specific to a job family or track. Engineering functional competencies might include technical execution, system design, and code quality. Sales functional competencies might include discovery, negotiation, and pipeline management. Aim for 4–6 functional competencies per job family.

A total of 6–10 competencies per job family is the practical ceiling for a usable framework. Beyond that, evaluation fatigue sets in, managers start rushing through competencies, and the quality of the ratings drops.

Step 3: Write behavioral indicators per level

This is the hardest and most important step. Behavioral indicators are the specific, observable descriptions of what each competency looks like at each career level. They are what transforms a framework from a list of admirable traits into a tool that can drive consistent, defensible ratings.

Good behavioral indicators follow three rules:

They describe actions, not traits. Not "is a strong communicator" but "documents technical decisions and shares them with affected stakeholders before implementation begins."
They specify scope and impact. The difference between levels is often about scope — individual scope at junior levels, team scope at senior levels, organizational scope at staff and above. Make this boundary explicit in each indicator.
They are falsifiable. Two reasonable managers should be able to look at the same employee evidence and agree on whether the behavior occurred. If they cannot, the indicator needs to be rewritten.

A useful test: can you name a specific real person (anonymized) who clearly meets this criterion? Can you name one who clearly does not? If the answer to both is yes, the indicator is well-written. If it is too vague to generate those examples, revise it.

Step 4: Validate with managers and senior ICs

No framework written in isolation by HR will fully capture the nuance of what great performance actually looks like in each function. Before finalizing anything, run a structured validation process with the people who know the work best.

Assemble a small working group for each job family: 2–3 people managers with direct experience managing the role, and 2–3 senior individual contributors who are widely recognized as high performers. Give them the draft framework and ask:

Is the level boundary between L2 and L3 in the right place? What evidence would you expect to see?
Are there behaviors at L4 that a strong L3 routinely demonstrates? If so, the bar may be set too low.
Are there important behaviors missing entirely — things that clearly distinguish your best performers but are not captured in the framework?
Can you think of a current or recent employee who sits clearly at each level? Do the criteria fit them accurately?

Run the framework through a mock calibration using real (anonymized) past cases. This stress-test surfaces ambiguities that seemed clear on paper but collapse under actual decision-making conditions.

Step 5: Map existing employees to levels

Once the framework is validated, apply it to your current workforce. This step has two purposes: it creates a baseline from which to measure growth, and it surfaces misalignments between official job titles and actual performance levels.

Work with each manager to assess their team against the framework criteria. Expect this to surface some uncomfortable conversations — employees who have been in a role for years but are performing below their official level, and employees who are clearly operating above their current designation.

Document the current mapping and share it with HR leadership. Cases where there is a significant gap between official level and assessed level should have a plan: a development timeline if the employee is below level, or an accelerated promotion conversation if they are operating above it. A structured Individual Development Plan is the right vehicle for the former.

Step 6: Integrate with your review cycle

A competency framework that is not embedded in the performance review cycle will not be used. The integration step is what transforms a document into a living system.

For integration to work, the framework must be the actual evaluation rubric in your review tool — not a PDF attached to a generic form. Managers should be rating employees competency-by-competency against the behavioral indicators for their level, not on a generic 1–5 scale. The system should surface the indicators at the moment of rating so managers cannot easily skip them.

Use the framework as the anchor for performance calibration sessions. When managers discuss outlier ratings, the question should always be: "Does the evidence support a consistent demonstration of the L3 criteria or not?" — not "what is your general impression of this person?"

Step 7: Publish and communicate to employees

A competency framework that employees cannot see does not drive the behaviors it describes. Transparency is essential — not just for fairness, but for effectiveness.

When you publish the framework, communicate:

Where employees can find it and how to navigate it.
How it will be used in the upcoming review cycle (which competencies will be rated, how self-assessments work).
How it connects to promotion and level decisions — what an employee needs to demonstrate consistently to be considered for the next level.
How and when it will be updated — setting expectations that this is a living document, not a permanent decree.

Encourage managers to reference the framework in 1:1s, not just during formal review cycles. The behavioral indicators are most useful when they inform day-to-day feedback conversations — "Here is what I observed, and here is how it connects to what we are trying to develop toward the L4 criteria" — rather than showing up only twice a year during review season.