Published methodology

The Rankix Score methodology

A locked, published formula - the same weights apply to every profile, every time. No hidden factors, no per-recruiter tuning.

Current weights

Scale
0-1000
Factors
6
Bands
5
Version
v1.1
In effect since 14 Sep 2026 · see the changelog

Five steps

From evidence to opportunity

Your score is built from real profile data in five steps - no step is skipped, and none is a black box.

  1. 01

    Evidence & signals

    Skills, experience, projects, certifications, and education you add to your profile.

  2. 02

    Verification

    Each claim carries a verification state - self-reported, partial, certified, or verified - that directly shapes its weight. Verification isn't live yet, so today every claim is self-reported and scores are in beta.

  3. 03

    Weighted formula

    Six published weights combine your component scores into one number - the same formula for every profile.

  4. 04

    Rankix Score

    A 0-1000 score and band, recalculated the moment your profile or verification status changes.

  5. 05

    Opportunities

    Higher scores and stronger verification lift your recruiter visibility and job match quality.

The formula

Six weighted components

Your internal score (0-100) is a weighted sum of published components, then displayed on a 0-1000 scale for readability.

Skills

30%

Your top skills (up to 10), each counted at its confidence scaled by verification state - verified 100%, certified 90%, partial 60%, self-reported 40% - so a focused specialist isn't penalized against a generalist with more, lower-confidence skills.

Experience

20%

Weighs total years, whether you hold a current role, and the share of your work history that's independently verified.

Projects

15%

Rewards active, non-archived projects - counting how many you have and what fraction are verified. Once you verify a project's GitHub repository, its real star count counts too.

Certifications

15%

Combines certification tier (foundational through expert), its verification state, and a freshness decay as a certification approaches or passes its expiry date.

Credibility & Verification

10%

The share of all your score-relevant signals - skills, certifications, projects, education - that carry independent verification rather than being self-reported. This is a cross-cutting factor: it measures verification quality, while each of the other factors measures the substance of that area.

Education

10%

Education entries, weighted toward having more than one entry and toward verified entries over self-reported ones.

Display scale

From internal score to display score

The internal 0-100 score is multiplied by 10 to produce the public 0-1000 score you and recruiters see.

Display score = internal score × 10

Internal score (0-100)Display score (0-1000)
85850
70700
55550
35350
15150

Bands

Score bands

Each display score falls into one named band.

  • Exceptional

    Deep, verified evidence across skills, experience, and delivery.

    850 - 1000
  • Strong

    Well-rounded, verified profile with consistent professional signal.

    700 - 849
  • Competitive

    Solid foundation with clear, achievable improvement opportunities.

    550 - 699
  • Developing

    Growing profile - evidence and verification will lift the score.

    350 - 549
  • Building Foundation

    Early stage. Adding verified experience and skills builds signal fast.

    0 - 349

Changelog

Methodology changelog

The formula changes only through a published version. Each entry says what changed, when, and what it did to scores. Current version: v1.1, in effect since 14 Sep 2026.

  1. v1.1

    14 Sep 2026

    Current

    Verification now counts for skills

    Self-reported skills count for less than verified ones, and only a Rankix check can mark a claim verified or record a project's GitHub stars.

    • Each skill now counts at its confidence multiplied by its verification state: verified 100%, certified 90%, partially verified 60%, self-reported 40%. Before, a skill counted at its self-rated confidence whatever its state, so typed-in confidence alone could max out the heaviest factor.
    • Only a Rankix review can change a claim's verification state, and editing a reviewed claim returns it to self-reported.
    • A project's GitHub stars can only be recorded by a Rankix check of a repository the candidate has shown they control. Stars recorded before this change stay until the project is checked.
    • The six factor weights are unchanged.

    Effect on scores: Every existing score was recalculated when this version took effect. Scores built mostly on self-reported skills went down; verified skills count exactly as before.

  2. v1.0

    7 Jul 2026

    Six-factor formula

    The first locked formula: six weighted factors, shown on a 0-1000 scale.

    • Skills 30%, experience 20%, projects 15%, certifications 15%, credibility & verification 10%, education 10%.
    • Certifications counted at their verification-state multiplier. Skills counted at their self-rated confidence, whatever their verification state.

    Effect on scores: The first version, so every score was computed with it until v1.1.

Fairness

Fairness commitments

Transparency isn't a policy page here - it's built into the score itself.

Published and locked

The formula and its weights are fixed and public - no per-recruiter tuning, no hidden factors applied to your profile behind the scenes. Every change ships as a dated version in the methodology changelog.

No protected-attribute inputs

Only skills, experience, projects, certifications, credibility, and education feed the formula - never demographic data.

AI explains, it doesn't decide your score

AI features like gap analysis and career coaching read from your real score and evidence to explain and act on it - they never compute or alter the number itself, and every AI output carries a visible reason and confidence.

Blind-hiring mode for recruiters

Recruiters can search with name, photo, and other demographic signals hidden, so evidence and score drive the shortlist first.

Think a factor is wrong? Dispute it.

Every component links back to real evidence in your profile. If something looks off, you can raise it directly from your score page - no support ticket required.

Go to your score
  1. 01

    Open your score page

    Sign in and go to your Rankix Score - every component and its evidence is listed there.

  2. 02

    Click "Dispute a factor"

    Describe which component looks wrong and why, referencing the specific claim or evidence.

  3. 03

    An admin reviews your evidence

    An admin takes ownership, checks your profile data and answers in writing, aiming to reply within 5 days. Disputes are never re-scored automatically.

  4. 04

    Your score updates if the dispute is upheld

    Any correction flows through the same recalculation pipeline and appears in your score history.

Your score

See where you land

Build your profile and your first Rankix Score is calculated from the evidence you add.