Core Competencies: How to Define Your Own, Not Copy a List
What core competencies are, how to derive your own from your strong performers, write behavioral indicators at three levels, and use them in hiring.
Core Competencies
A competency framework only works when it describes your business rather than a generic one: what separates a competency from a skill and from a company value, how to derive yours by comparing your own strong performers against your adequate ones, how to write behavioral indicators at three levels so two managers reading the same evidence pick the same rating, how many a small company can actually sustain, and how to run identical wording through job descriptions, interviews and reviews so the framework never becomes a file nobody opens
My first competency framework took an afternoon, because I did not build it. I found a list of nine competencies with names like Drives Results and Champions Change, changed two words, saved it to the shared drive, and told the managers we had a framework. Six months later I asked two of them to rate the same salesperson on Communication. One said strong, the other said it was the main problem. Same person, same quarter, same evidence.
That is what a borrowed list buys you. The words were not wrong. They were true of every company on earth, which meant they carried no information about ours, and a competency that carries no information turns into a mirror: every rater sees their own preferences in it and calls the reflection a rating. The version that eventually worked was shorter, uglier, and written in sentences naming things people had actually done.
So this is the employer-side version for a business with no HR department: what a core competency is and what it is not, how to derive yours from your own people rather than a template, how to write behavioral indicators at three levels so two managers reading the same evidence land on the same number, how many a small company can carry without the whole thing collapsing, and how to run identical wording through job descriptions, interviews and reviews. I build the people and records tooling for companies without an HR department at FirstHR. General guidance, not legal advice.
What a Core Competency Actually Is
A core competency is a named pattern of behavior your business needs people to repeat, described by what it looks like when somebody does it well and when they do not. The name is a filing label. The sentence underneath is the entire product.
The phrase carries two meanings and they get mixed constantly. In business strategy, core competence describes what an organization as a whole is uniquely good at, a term from the Prahalad and Hamel article in Harvard Business Review published in 1990. In hiring and performance work, a core competency describes an individual and what you expect them to do.
For a small employer the individual meaning is the useful one, because it is the version that has to appear on a form somebody fills in. If you go looking for advice and land in the strategy literature instead, you will get several thousand words about sustainable advantage and nothing you can put on an interview scorecard.
Names alone are the trap. Ownership, Excellence, Collaboration: each is a word two people will define differently while both believe they agree. I now write the definition sentence before choosing the name, and about half the time the sentence turns out to describe something the original word did not cover at all.
The test for whether something is core is simple. Would you be uncomfortable if a new hire in any department still could not do this by month six? If the answer is yes, it belongs in the company framework. If the answer is only for the bookkeeper, it belongs with the bookkeeper and lives in the job description, not the shared list.
Competency vs Skill vs Value, and Why Mixing Them Breaks Everything
A skill is something a person can do, a competency is how they behave while doing it when conditions are not ideal, and a value is a statement about what the company cares about. Put all three in one list and the framework becomes unratable, because each of the three is evidenced in a completely different way.
The failure is specific and it happens fast. When a value sits in a rating grid, managers rate agreement rather than behavior, and the person who pushes back in meetings scores low on a category nobody will define out loud. When a skill sits in a competency framework, the level descriptions collapse into can and cannot, and you lose the middle band where almost everybody actually sits.
One question sorts most items in seconds. Ask what you would need to see before giving somebody a high rating on it. If the answer is a demonstration, it is a skill. If the answer is three or four episodes across a quarter, it is a competency. If there is no answer at all, it is a value, and it does not belong on a form with a rating box.
Keep the three in three separate homes. Values belong in your company values statement and in conversations about how the work gets done, never in a score. Skills belong in role requirements and in a skills gap analysis, where the question is proficiency against a level you set in advance.
Competencies sit between the two, and that middle position is exactly why they are worth the trouble. They are the part of performance that survives a change of tools, a change of software and a change of process, which is also the reason the distinction between hard and soft skills keeps producing arguments that a behavioral definition settles in one sentence.
Deriving Competencies From Your Own Strong Performers
Derive your competencies by putting two or three of your strongest performers next to your merely adequate ones and writing down, in verbs, what the strong ones do that the others do not. That contrast is the only source that produces a competency meaning anything at your specific company.
Run it in one sitting with whoever else manages people. Use real names in both columns, not types. Ask what the strong one did last quarter that the adequate one would not have done, and keep asking until the answers stop being adjectives. Somebody will say she is proactive. The follow-up question is what she actually did.
Usually the answer arrives as a story: she rewrote the delivery note after the second complaint without anybody asking her to. That is already a competency sentence. Strip the name, keep the shape, and you have something two managers can recognize when they see it again in an interview.
Expect the session to produce twelve or fifteen candidate behaviors, and expect most of them to overlap. Group them without mercy. Two sentences that would always be rated the same way are one competency with a naming disagreement attached, and merging them now saves you the same argument in every review for the next three years.
One caution, because this method has a known failure mode. Derive everything from a single favorite employee and you will encode that person: their communication style, their working hours, their background, their taste. Force yourself to name at least two strong performers who are not similar to each other, then cut anything describing style rather than outcome.
That discipline is the same one that keeps interview criteria defensible, and it is worth borrowing the checklist from reducing bias in the hiring process before you finalize the list. A competency you cannot evidence is a preference wearing a job-related costume.
The formal name for what you are doing is job analysis. The US Office of Personnel Management describes it as examining the tasks performed in a job, the competencies required to perform those tasks, and the connection between the two, and lists legal defensibility of assessment and selection procedures among its purposes (OPM, job analysis). Built for federal hiring at enormous scale, but the logic transfers exactly: competencies are derived from work, never chosen from a menu.
Writing Behavioral Indicators at Three Levels
Write three levels for every competency, each a single sentence describing behavior somebody could have watched, and rate by matching evidence to the closest sentence. Three is the smallest number that lets two managers agree while still saying something useful about the difference between people.
Five-point scales fail predictably. Everyone lands on three and four, the difference between them is undefendable, and the first time an employee asks why they received a three you discover you cannot answer without inventing a reason. Three levels force a real decision: does this person need help, do they do it alone to standard, or do they set how it is done here.
| Level | Indicator sentence | What counts as evidence |
|---|---|---|
| Level 1: needs support | Handles a customer problem once the steps are given, and checks in before deviating from them | A ticket log showing the same question asked more than once. A manager who was consulted before each non-standard decision |
| Level 2: does it alone | Owns a customer problem from first contact to resolution, including the follow-up, without the manager being told first | Three closed cases nobody else touched. A complaint that never became an escalation. The customer emailed the person, not you |
| Level 3: sets the standard | Fixes the process that produced the problem, and the next person handles the same case without asking | A written change to how the case gets handled. A colleague who now resolves the same case unaided. Repeat volume dropping |
| Not enough evidence | Do not rate. Say so, and name what you would need to see next cycle | Fewer than two or three observed episodes. New in role. Work not visible to the rater. This is a legitimate outcome, not a cop-out |
Four rules make the sentences work. Use a verb and an object, never an adjective. Write the middle level first, since it is your standard for the role, then work outwards. Make each level absorb the one below it. And ban the words that hide judgment: consistently, effectively, appropriately, as needed.
Those four words are worth dwelling on. Each of them lets two managers mean entirely different things while using identical language, which is precisely the failure you built the framework to prevent. Where you can, name the artifact or the event, because artifacts are countable and impressions are not.
Then run the test that actually matters. Give two managers the same three pieces of real evidence about one real employee and ask them to pick a level separately, without discussing it. If they disagree, the indicator is wrong, not the manager. I have never had a framework survive that test on the first pass, and the rewriting afterwards is where all the value is.
Repeat the calibration exercise whenever a new manager starts using the framework. Rating consistency is a trained habit, not a property of the document, which is the same reason interviewers need practice on your scorecard before they use it on a candidate.
How Many Competencies a Small Company Can Actually Sustain
Five or six company wide, plus at most two that are specific to a role. Past eight, the framework stops being used and starts being complied with, which looks identical on paper and produces nothing.
Count the writing cost first. Six competencies at three levels is eighteen sentences you have to draft, argue about, calibrate and maintain. Now count the using cost, which is the one that kills frameworks: every review requires six separate judgments per person, every interview loop requires six scored questions spread across your interviewers, and every job description carries six lines.
There is a third cost most people miss. Every competency you add is a competency your managers must be able to explain to somebody who has just received a low rating on it, in a conversation nobody enjoys. Six explanations is a difficult afternoon. Twelve is a training program you are not going to build.
Large frameworks exist for a reason that does not apply to you. An organization with thousands of employees needs a shared vocabulary across divisions that never meet, and a structure a promotion committee can defend to someone who has never met the candidate. You have neither problem. Your managers eat lunch together.
Role-specific competencies are the one extension worth allowing, capped at two per job. A first-line manager needs one about giving feedback that changes behavior, which is the point where this work meets leadership development. A salesperson needs one about qualifying out early. Write them in the same three-level form and store them alongside the job responsibilities, not in the company list.
If you cannot decide what to cut, use this question: which of these would you actually act on if somebody sat at level one for two cycles running? Keep those. The rest are things you admire, and admiration does not need a rating scale.
Putting Competencies Into Job Descriptions
Put the competency name, its definition sentence and the target level directly in the job description, using the exact words that will appear on the interview scorecard and the review form. Different wording in the three places is functionally the same as having no framework at all.
Target level at hire is usually level two for most competencies and level one for one or two you are willing to develop. Writing that down does two useful things. It stops the search drifting toward somebody overqualified and expensive, and it tells the hiring manager in advance which gaps are acceptable.
Set the target levels before the search opens rather than after you have met people. Levels chosen once a shortlist exists bend toward whoever is standing in front of you, and the description quietly turns into a portrait of the strongest applicant instead of a statement about what the role requires.
Keep competencies visually separate from duties. Duties describe the work; competencies describe how the work has to be done. A description that mixes them produces a bullet list where reconcile the bank account monthly sits next to strong communicator, and no candidate can tell which of the two is being tested.
The public posting can carry a shortened version in plain language, two or three competencies at most, because a posting is marketing and a job description is a working document. Keep the full set in the internal version the interview panel reads, alongside the rest of your hiring process materials.
Running the Same Competencies Through Interviews
Turn each competency into one past-behavior question, assign competencies across the panel so no two interviewers cover the same one, and score every answer against the identical three levels used everywhere else. The scorecard becomes a level per competency plus one line of evidence, not a rating out of ten.
The question form matters less than the follow-ups. Candidates arrive with a rehearsed story for every headline competency, and the level only becomes visible two questions deeper: what did you do first, who did you tell, what happened when it did not work. The STAR method is the standard way to structure that, and it fits inside any structured interview format you already run.
Competencies do not replace a work sample. A competency interview tells you how somebody behaved; a work sample tells you what they can produce. Use both, which is the practical core of skills-based hiring, and if you are weighing a commercial test on top, the trade-offs sit in hiring assessments.
Use the same competencies for internal candidates. Promoting somebody you already manage feels like it should skip the scorecard, and skipping it is exactly how a promotion ends up resting on visibility rather than behavior. Run the same questions, and take the evidence from a year of observed work instead of from the stories you remember best.
One habit pays for itself immediately. Keep the completed scorecards. They are the record that the same standard was applied to every candidate, they make the eventual interview feedback conversation take four minutes instead of twenty, and they give the new hire a starting level on day one instead of a blank review form at month six.
Carrying Competencies Into Performance Reviews
In the review, the competency section answers exactly one question per competency: which level does the evidence from this cycle support? Not a score out of five, not a paragraph of adjectives, and not a comparison against the rest of the team.
Every rating gets one line of evidence, and the line has to be an event. Took ownership of the Henderson account escalation in March and closed it without involving me is a rating. Shows real ownership is an impression, and impressions are what make a review argue with itself when somebody disagrees six months later.
| Where it appears | What it says | What it decides |
|---|---|---|
| Job description | The competency, its definition sentence, and the target level for the role | Whether the search is aimed at somebody who already operates at level 2 or somebody you will develop |
| Job posting | Two or three competencies in plain language, no levels | Whether the right people recognize themselves in the ad and apply |
| Interview scorecard | The three level sentences verbatim, one competency per interviewer, with space for evidence | The hire decision, and the level the new person starts at |
| Onboarding and first review | The starting level from the scorecard, plus what level two looks like here | What the first ninety days are actually for, rather than a generic checklist |
| Performance review | One level per competency plus one line of evidence per rating | Development priorities, and the honest input to any pay or promotion conversation |
| Development plan | The single competency where moving up a level changes the most | Where the year of effort goes, instead of spreading it across six weak areas at once |
Keep competency ratings separate from goal results in the same review. Goals answer whether the outputs arrived; competencies answer how the person operated. Collapsing them means a great quarter hides a behavior problem and a bad quarter buries a genuinely strong performer, which is the most common complaint about any performance review process.
Where competencies earn their cost is development. Pick the one competency where moving from level one to level two changes the most about the person’s year, write that into an individual development plan, and check it in the regular one-on-one rather than waiting for the next cycle.
Rate competencies once a year rather than every quarter. Behavior patterns do not move fast enough for a quarterly level change to carry information, and asking managers to re-rate six competencies four times a year produces copying rather than thinking. Quarterly conversations are for evidence accumulating; the level itself gets revisited annually.
Promotion also gets shorter. Moving up is moving from level two to level three on the competencies the next role requires, which is a claim you can evidence rather than a feeling about readiness. That is also the sentence to use when you decline a promotion request, and it holds up far better than the alternatives, as anyone who has written a performance review under pressure will recognize.
Once a Competency Decides Who Gets Hired, It Is a Selection Procedure
Any competency rating that influences who gets hired, promoted or let go functions as a selection procedure under the federal Uniform Guidelines on Employee Selection Procedures, which apply to procedures used as a basis for employment decisions including hiring, promotion and retention. What triggers the duty to validate one is adverse impact on a race, sex or ethnic group rather than the mere existence of a rating, and Title VII, the statute behind the guidelines, reaches employers at fifteen employees. That does not mean a small employer needs a validation study. It means the competency has to come from the work.
The guidelines set out three ways to demonstrate that a procedure is job related: criterion-related validity, which requires data showing the procedure predicts job performance; content validity, which requires showing the procedure represents important aspects of performance on the job; and construct validity, which requires evidence that it measures characteristics determined to be important in successful performance (29 CFR 1607.5).
Read that as a design instruction rather than a compliance chore. A competency derived by comparing your own strong and adequate performers, written in observable behavior, and applied identically to every candidate for a role, is already built along the lines the guidance describes. A competency copied from a template because it sounded impressive is not.
Two habits do most of the protective work. Use the identical competencies and identical level sentences for every candidate for a given role, and keep the completed scorecards with the rest of the hiring file. Records showing one standard applied consistently are worth more than any document titled framework.
The Failure Mode: A Framework Nobody Ever Opens
Competency frameworks die for one reason. They live in a document instead of in the forms people already have to complete, so using the framework requires somebody to remember a file name and go find it. That file gets opened twice and then never again.
| Symptom | What actually happened | The fix |
|---|---|---|
| Nobody can name the competencies | They exist only in a standalone document nobody has a reason to open | Paste the text into the job description template, the interview scorecard and the review form. Delete the standalone version from daily use |
| Two managers rate the same person differently | The levels are labels rather than behavior sentences, so each rater supplied their own definition | Rewrite each level as a verb with an object. Calibrate on one real employee with two raters before relaunching |
| Everyone scores in the middle | Five-point scale with no anchors, plus no consequence attached to the number | Three levels with anchor sentences, and one real decision tied to the levels |
| The list has grown to fourteen | Competencies were added after every bad hire and never removed | One owner, one revision a year, and a hard cap. Adding one requires removing one |
| Ratings never change anything | The framework was launched as a documentation exercise with no decision behind it | Attach promotion, pay band or the next stretch assignment to the levels. Arguments about ratings are the sign they matter |
| It contradicts the review form | The wording was edited in one place and not the others | A change is not real until all three forms carry the identical sentence. Make that the rule, in writing |
The fix is mechanical rather than cultural. The competency text gets copied into the templates, so nobody ever navigates to the framework, and the framework file becomes the place you edit wording rather than the place anyone reads it. Everything else about adoption follows from that one decision.
The second death is quieter. If the levels never change anything, people stop investing effort in them within two cycles, and you get a form filled in from memory the night before. Attach at least one real decision to the ratings, then expect arguments about them. Arguments are the evidence the framework is being used, and a rating system nobody ever contests is a rating system nobody is reading.
Finally, resist growth by accumulation. Every bad hire creates pressure to add a competency that would have caught it, and nobody ever proposes removing one. Give the framework a single owner, allow one revision a year, and make adding a competency require removing one. That constraint is what keeps a small company’s performance management honest rather than ornamental.
When a Small Business Genuinely Should Not Build One
Skip the framework if the founder is the only person who rates anybody, if the work is a single role with countable output, if you do not yet have a review cycle that reliably happens, or if the business is in the middle of changing what it sells. In all four cases the framework solves a problem you do not have yet.
The single-rater case is the clearest. Consistency between raters is the problem a competency framework exists to solve, and with one rater there is nothing to be consistent with. Writing it down at that stage documents your own preferences and calls them a standard. Write the competencies when a second manager starts rating people, which is the exact moment the vocabulary problem appears.
The countable-output case is next. Some businesses run on one job with a number attached: routes completed, units produced, tickets closed to standard, covers served. Where output is countable and the quality bar is objective, a competency framework adds ceremony to a question the numbers already answer, and the honest move is to invest in the measurement instead.
The third case is the most common and the least admitted. If reviews are not actually happening, a framework will not create them. It gives a meeting that does not occur a nicer agenda. Build the review cycle first, run it twice so you know it survives a busy quarter, then add competencies to a form people already complete.
Last, a business mid-pivot should wait. Competencies describe what good looks like in the work you do now, and if the work is about to change, you will write the framework twice and lose whatever credibility the first version had. Six months of stability is a reasonable bar before you start.
None of these exemptions is permanent. The single-rater company hires a second manager, the countable-output company adds a role whose output is not countable, and the pivot finishes. Build the framework at the moment the problem shows up and it takes an afternoon and produces something people use. Build it early, before anybody feels the pain it solves, and you have written the document that convinces your team frameworks do not work here.
Frequently Asked Questions
What are core competencies?
Core competencies are the small set of behaviors a business expects from every employee, written so that they can be observed and rated rather than argued about. Each one has a name, a one-sentence definition built around a verb, and level descriptions that say what the behavior looks like at different stages of mastery. The word core means company wide: a behavior that only matters for one job is a role requirement, not a core competency. Note that the phrase has a second, older meaning in business strategy, where core competence describes what the organization as a whole is uniquely good at. That version comes from a Harvard Business Review article by Prahalad and Hamel published in 1990 and describes the company, not the person. For hiring and performance work, the individual meaning is the one you need.
What is the difference between a competency and a skill?
A skill is something a person can do; a competency is how they behave while doing it when conditions are not ideal. Reconciling a bank statement is a skill, and you can verify it in twenty minutes with a work sample. Raising a discrepancy the same day instead of hoping it resolves itself is a competency, and you can only see it across several episodes. The practical difference is evidence. Skills are demonstrated once and rated against a required proficiency level. Competencies are inferred from a pattern, which is why they need behavioral level descriptions rather than a pass or fail. Mixing the two into a single list is the most common reason a framework becomes unusable, because the level descriptions collapse into can and cannot and you lose the middle, where nearly everybody actually sits.
How many core competencies should a small business have?
Five or six that apply to everyone, plus no more than two role-specific ones per job. The constraint is not ambition, it is arithmetic. Six competencies at three levels each is eighteen sentences to write, argue about and maintain, and every review then requires six separate judgments per person while every interview loop requires six scored questions spread across your interviewers. That is sustainable for a team without a dedicated HR person. Frameworks with twenty or thirty entries exist for organizations that need a shared vocabulary across divisions which never meet and a defensible structure for promotion committees. A small company has neither problem. If you cannot decide what to cut, ask which competencies you would actually fire somebody for failing at level one, and keep those.
Can I just download a standard competency list and use it?
You can use one as raw material, but not as the framework itself. Published lists are written to be true of every organization, which means they carry no information about yours, and a competency that carries no information becomes a mirror in which each rater sees their own preferences. That is how two managers end up rating the same employee three levels apart on the same word. Use a published list the way you would use a menu: read it, see which items describe something real in your business, then rewrite every one of them in your own sentences using examples from your own people. The rewriting is not a formatting exercise. It is where you find out that what you actually value is narrower and stranger than the generic label suggested.
How do you write behavioral indicators?
Write one sentence per level, each describing something a person could have been watched doing, using a verb, an object and a condition. Start with the middle level, which is the standard for the role, then write the level below it as what needs support and the level above as what sets the standard for others. Ban the words that hide judgment: consistently, effectively, appropriately, as needed. Each of those lets two managers mean different things while using identical language. Where you can, name the artifact or the event, because artifacts are countable. Reworks the delivery note after a complaint without being asked is an indicator. Takes ownership is not. Test the result by giving two managers the same evidence about the same employee and asking them to rate separately.
Are core competencies the same as company values?
No, and treating them as the same is the single most damaging mistake in this area. A value is a statement of what the organization cares about, asserted rather than evidenced, and used to settle arguments about priorities. A competency is a behavior you can observe, describe at levels and rate from evidence. The moment a value appears as a line in a rating grid, managers begin scoring agreement rather than behavior, and the employee who challenges decisions in meetings quietly loses points on a category nobody can define or defend. Keep values in your values statement and in conversations about how work gets done. Keep competencies in the job description, the interview scorecard and the review form, where each rating has to be backed by an example.
How do you stop a competency framework becoming a document nobody opens?
Stop treating it as a document. Copy the competency text directly into the three forms people are already required to complete: the job description template, the interview scorecard and the performance review form. If using the framework requires anyone to remember a file name and navigate to it, it will be opened twice and then abandoned. The framework file becomes the place you edit wording, not the place anyone reads it. Two more habits matter. Give one person ownership and allow one revision a year, with the rule that a wording change is not real until all three forms are updated. And attach at least one decision to the levels, such as promotion or pay band, because ratings that never change anything stop attracting effort within a cycle or two.
Do competency ratings create legal risk?
They create exposure only in the way any selection criterion does, and the fix is the same. Once a competency rating influences who gets hired, promoted or let go, it functions as a selection procedure under the federal Uniform Guidelines on Employee Selection Procedures. Those guidelines require a procedure to be validated when it produces adverse impact on a race, sex or ethnic group, and they set out criterion-related, content and construct validity as the three ways to do that. Title VII, the statute behind them, reaches employers at fifteen employees. A small employer is not running a validation study, but the underlying logic still applies: derive competencies from the work rather than from taste, apply the identical criteria to every candidate for a role, and keep the completed scorecards. The competencies that create real risk are the ones describing style rather than work: presence, polish, fit and energy. Cut those. This is general guidance, not legal advice.