Competency-based performance appraisal
Appraise each person against the role they actually do.
Each person is graded by themselves, their peers and their manager on the competencies their role calls for, weighted for their level. Every score shows how it was reached.
In English and Bulgarian. Runs on your own servers or your own Cloudflare account.
- 540
- competencies, written for every stage
- 26
- role families and 70 specializations
- 3
- sources: self, peers and manager
- 100
- means Regular throughout, out of 130
01Why
One appraisal form for everyone measures nobody well.
A junior and a senior get the same list.
One list of questions for every role and level asks a new hire and a principal engineer the same thing, and weighs it the same.
Scores arrive without their working.
A number on a form tells a person where they landed, not why. The conversation that follows starts from doubt.
One view stands in for everyone’s.
When only the manager grades, the result says as much about the manager as about the person. Colleagues who see the work daily are not asked.
02The model
Three things decide what each person is graded on.
The competencies on an evaluation come from the person’s role family and specialization. How much each one counts comes from their stage-level.
Role family
The kind of work. Twenty-six standard roles of an ICT company ship with the product, from Software Engineering to Security Governance, Risk & Compliance.
- SE
- QE
- BA
- PM
- XD
- DA
- IO
- MC
- PD
- TC
- HR
- BD
- TS
- HI
- CI
- SM
- SD
- SA
- DP
- SC
- SO
- TI
- IR
- PT
- CE
- GR
Specialization
The focus within a family, such as Backend in Software Engineering. The family’s baseline holds what every member does, and a specialization adds what that role’s work calls for. Seventy ship, and administrators add their own.
Stage-level
Where the person stands, in 13 steps from Intern to Expert or Manager. A relevancy curve sets how much each competency counts at each step.
- N1Intern
- J1J2J3Junior
- R1R2R3Specialist
- S1S2S3Senior
- X1Expert
- T1T2Manager
Competency
Something a person’s work calls for: what they know and can do, how they work, or how they behave with others, from knowing a programming language to estimating a task or mentoring a colleague.
Every competency is audited against four principles.
- Observable
- A competency describes something a person does that their manager can point to, not a trait or an impression.
- Concrete
- A competency says what it looks like at each of the six stages, written so the person can answer “give me an example from your own work”.
- Unique
- A competency appears once in the dictionary and is shared by every role that needs it, so no two competencies grade the same thing.
- Individual
- A competency is graded only if it describes the person’s own work. Everyone in a role family is graded on what the whole family does, and their specialization adds individual scope.
540 predefined competencies across 26 standard roles of an ICT company. Each is written for every stage, so the person sees the version for their level, and administrators can rewrite any of them in both languages.
Three competency categories, nine subcategories
Every competency belongs to one of three categories. Together they span the hard skills of a role, the methods of work it relies on and the soft skills it calls for.
Expertise
Professional competence
- Theoretical knowledge
- Applied skills
- Practical experience
Insight
Methodological competence
- Processes
- Planning
- Estimation
Commitment
Social and personal competence
- Responsibility
- Communication
- Mentorship
03Scores
A final score you can follow back to its grades.
Every grade carries a weight. Competencies count by their relevancy at the person’s level, the three sources are blended, and the result is a score out of 130, where Regular throughout makes 100.
- Unsatisfactory× 0.6
- Regular× 1.0
- Superior× 1.3
- Not utilized× 0
- Self, peers and manager
- Blended at 20%, 30% and 50%. A source that took no part is left out and the others share its weight, so a missing round never pulls a score down.
- Peers stay anonymous
- The person sees one cumulative peer grade for each competency, never a colleague’s own.
- The working is shown
- Every final score opens to how it was calculated, beside a profile by subcategory against the one expected at the person’s level.
- Five bands
- Weak, Insufficient, Expected, Good and Outstanding. Expected runs from 90 to 105.
And what the cycle says about the organization.
Cycle analytics shows how far each evaluation has got and how scores spread at each level. It sets every subcategory against the expected profile by role family, and each person’s self-assessment against their manager’s. Managers get the same view of their own units, and Trends follows the bands from one cycle to the next.
04The cycle
One cycle, from the plan to the closing interview.
Each individual appraisal covers up to 32 competencies: the ones the person’s role family and specialization call for, weighted for their stage-level.
- Up to 32
- competencies in each individual appraisal
- 3–5
- peers chosen for each person
- 14
- days for the self-evaluation and peer feedback
- 30
- minutes for the closing interview
-
Plan
Set the dates, an optional focus KPI and the competency set for each role family and specialization. Lock the cycle once every set is complete.
-
Open evaluations
A manager opens each person’s evaluation and picks peers, seeing how many reviews each colleague already has.
-
Self and peers
The person grades themselves and the peers give their feedback.
-
Manager review
Once both rounds are in, the manager grades and writes the review. While the evaluation is open, they see neither round.
-
Interview and close
A Supervisor books the interview in one of a manager’s available slots. The manager who holds it records the outcome and the next period’s goals and submits them, and a Supervisor approves them, which closes the evaluation.
05Who it’s for
For the people who answer for appraisal.
HR leaders
Run every cycle from one place in the Supervisor role, from planning to the closing interview, with nothing in spreadsheets.
- Follow every evaluation in the cycle, overdue first
- Act on a stuck one with the reason recorded
- Research-use consent, recorded within the process
Engineering and technology leaders
Appraise engineers on what their specialization calls for, at the depth their level calls for.
- 26 role families and 70 specializations ship
- Each competency written for every stage, from Intern to Manager
- Peers who see the work daily take part
CEOs of small and mid-sized companies
A structured appraisal without building one: the dictionary, the curves and the scoring ship ready, and the content can be fitted to the company.
- Analytics for the whole organization and for each manager’s units
- Scores that compare across teams
- Runs on your servers or your Cloudflare account
06Security and data protection
Appraisal data stays where you put it.
Appraisals hold some of the most personal data a company keeps. The application runs on infrastructure you choose and shares it with no one.
Built for the GDPR
A Supervisor can export everything held about one person in a single file, except their peers’ individual grades and notes, for access and portability requests (Articles 15 and 20). An administrator anonymizes a former employee when their data must go (Article 17). Research use rests on a recorded consent. The employee import’s review shows each row’s ID and name and, of the rest, only a change of employment status.
- EU data residency
- On Cloudflare, the data is kept in a D1 database in the EU jurisdiction. Run it on your own servers and it never leaves them.
- Work-account sign-in
- Microsoft Entra ID by default, or Google, each limited to the e-mail domains you allow. Every screen checks the role on the server.
- No third-party scripts
- The application loads nothing from another origin: no analytics, no trackers, and its fonts are served from your own installation.
- Anonymous peers
- A person sees one cumulative peer grade, never a colleague’s own. In analytics, a group of fewer than three people shows no figure.
07Configuration, hosting and license
Run it yourselves, on your terms.
The content and the organization are yours to change, each on a screen of its own, in English and Bulgarian.
Out of the box
Ready as it ships.
A complete default configuration comes with the product: the 540 competencies, standard competency sets to start each cycle from, the eight relevancy curves, the scoring and the process itself. You can run a full cycle on it as it is.
When you need it
Yours to change.
Change what the content says and how much it counts: every competency’s text in both languages, the role families and their specializations, the relevancy curves and which one each competency follows, and each cycle’s competency sets.
Hosting
One container image: Docker Compose with Redis, a standalone container, Kubernetes or your own Cloudflare account, where a Worker fronts the container.
Configuration
Role families, specializations, curves and every competency’s text, in both languages. Each change keeps its author, and any earlier version can be restored.
Upgrades that keep your edits
When a release ships new content, the Configuration drift panel shows how it differs from yours and replaces only the documents you tick. One you have edited is never ticked for you.
People and organization
The structure, work sites and employee records. The import shows what will happen to every row before anything is written.
License
AGPL-3.0-or-later. The source code is available to companies only, on request. Built on ti-engine, an Apache-2.0 framework.
Languages
English and Bulgarian throughout, including the user guide built into the app.
08Questions
Asked often.
Can we use our own competencies?
Yes, by making the ones that ship your own. Administrators rewrite any competency’s text in both languages and set the relevancy curve each one follows. Each such change keeps its author, and any earlier version can be restored. A Supervisor chooses which competencies each role family and specialization is graded on in every cycle, starting from the standard sets.
Can we start without configuring anything?
Almost. The product ships with a complete default configuration, from the competencies and their standard sets to the curves, the scoring and the process, and a cycle runs on it as it is. What it can’t know is your organization: you set up your organization chart and import your employee profiles. Adapt the content later, when you need to.
What happens to our edits when a new release ships?
They stay unless you replace them. The Configuration drift panel shows how the release’s content differs from yours, document by document, and replaces only the documents you tick. One you have edited is never ticked for you, and any replaced version can be restored.
Where is our data stored?
On your own servers, or in your own Cloudflare account, where it is kept in a D1 database in the EU jurisdiction. No appraisal data is sent to us or to anyone else.
Do people need a new account?
No. They sign in with their work account, through Microsoft Entra ID or Google, limited to the e-mail domains you allow.
Is it GDPR compliant?
It is built to support your GDPR obligations. A Supervisor can export one person’s data in a single file, an administrator anonymizes a former employee whose data must go, research use rests on a recorded consent, and the data stays on infrastructure you choose. As the controller of the data, your organization stays responsible for how it is used.
Can we get access to the source code?
Companies can ask for access at contact@competence.work, and each request is considered individually. The application is licensed under AGPL-3.0-or-later.
See it with your own roles.
Write to contact@competence.work for a demo. Companies can ask for the source code at the same address.