Skip to main content

METHODOLOGY

How HireGPS actually scores you.

Four principles. Each one has a receipt: a real example you can inspect, not a marketing claim.

PRINCIPLE 01

Same CV, same score. Every time.

A generic AI coach reads your CV through a language model: every run uses sampling, temperature, and model-confidence blending, so the same CV can come back with two different scores on two different runs. The "feel" of the answer changes; the answer changes with it.

HireGPS does not score with a language model. The score is the weighted sum of eight named category scores, run through explicit numeric thresholds. Nothing samples. Nothing drifts. Run the same CV against the same target role twice, and the arithmetic produces the same number, every time.

  • Weighted sum across 8 named scoring categories: no temperature, no sampling.
  • Decision (Shortlist / Hold / Reject) comes from explicit numeric thresholds.
  • Every risk fires from a named rule with a Detection, a Relevance, and an Action.

Generic AI coach

Same CV. Same role. Two runs.

  1. Run 171"Generally a strong candidate."
  2. Run 264"Some gaps in leadership signal."

Different score, different read, every time.

HireGPS arithmetic

Same CV. Same role. Same math.

CategoryRawWeightContribution
Role Relevance8222%18.04
Seniority Match8817%14.96
Achievement Evidence6122%13.42
Career Stability7414%10.36
Education & Certification Fit709%6.30
Language & Market Fit687%4.76
Experience Context Signal725%3.60
CV Clarity854%3.40
Total (all 8 categories)74.84

Illustrative raw scores. The weights are the engine's own senior-level starting weights, before industry, function and region adjust them. The published score is this sum rounded to a whole number: 75.

PRINCIPLE 02

Your CV scored by the recruiter who hires for your role.

A recruiter hiring Senior Software Engineers in tech weighs Achievement Evidence differently than a recruiter hiring Hospitality Managers in a regional market, and both read Career Stability through different rules. Treating every CV with the same scoring curve flattens that judgment. HireGPS doesn't.

Inside the app, the eight category weights tilt up or down across four axes: industry, region, function, and seniority. The weights mirror the recruiter who would actually read you for the role you target. Persona adjustment applies at analysis time. There is no single set of weights: each of the engine's 7 seniority levels starts from its own curve. The chart below shows one of them, the senior-level curve, before industry, function and region adjust it.

Receipt: Senior-level starting weights8 scoring categories. Sum = 100%.
  1. Role Relevance22%
  2. Seniority Match17%
  3. Achievement Evidence22%
  4. Career Stability14%
  5. Education & Certification Fit9%
  6. Language & Market Fit7%
  7. Experience Context Signal5%
  8. CV Clarity4%

Senior-level starting weights shown. Industry, function and region adjust them inside the app.

PRINCIPLE 03

40 risks. Named. Tiered.

A recruiter reading a CV doesn't think "red flag keyword." They think objections: concrete, named, ordered by how much each one would slow a yes. HireGPS writes objections like these into a fixed catalog of 40 risks, each with a name, a severity, a trigger condition, and an explanation template.

The catalog is tiered by impact: 2 BLOCKER risks that stop a shortlist outright, 10 HIGH risks that land as direct recruiter objections, 21 MODERATE risks that slow the read, and 7 LOW risks that read as polish-grade noise. 40 rules. Reproducible every run.

A HIGH risk does not reject a CV on its own. Blocker Aggregation does: it fires on a combination of other active risks: for example, three or more HIGH risks at once.

BLOCKER

2

Stops a shortlist outright.

  • K20Language Blocker
  • K40Blocker Aggregation

HIGH

10

Lands as a recruiter objection.

  • K01Function Mismatch
  • K03Triple Career Change
  • K04Seniority Overreach
  • K06Leadership Gap
  • K08No Measurable Impact
  • + 5 more

MODERATE

21

Slows the read.

  • K02Industry Mismatch
  • K05Overqualification / Retention Risk
  • K07Title Inflation Concern
  • K09Weak Measurable Impact
  • K10Unanchored Achievement Claims
  • + 16 more

LOW

7

Polish-grade noise.

  • K24Weak Market Familiarity
  • K26High-Context Without Depth
  • K30Formatting Inconsistency
  • K31Keyword Stuffing / Vague Language
  • K33Generic Untargeted Positioning
  • + 2 more
Receipt: Expanded risk

K20Language BlockerBLOCKERLanguage & Market Fit

Detection
The mandatory language requirement for [region name] includes [mandatory languages], but this language is missing from the CV. Language context: strict.
Relevance
In [region name], [mandatory languages] proficiency is not a preference — it is a legal or operational requirement for [function label] roles. Candidates who cannot communicate in the required language cannot perform the role. This is a hard filter.
Action
If you speak [mandatory languages] and have not listed it, add it immediately with your proficiency level. If you are learning the language, state your current level and expected timeline. If you do not speak it, this role may not be eligible without relocation or remote arrangements.

The engine's own template for this risk, word for word. Bracketed slots fill in from your analysis.

PRINCIPLE 04

Every score traces back to a rule.

A score you can't audit isn't a score: it's a guess wearing a number. HireGPS treats explainability as a doctrine: every category contribution, every fired risk, and every ranked action ties back to a named rule the engine can show you, in plain language, written for the industry and level you are targeting.

The three columns below are not branding. They are the literal structure of every rule inside the catalog. Read one rule, and you can tell why it fired, why it matters for your role, and what to do about it.

  • No score without a rule. No rule without a name.
  • Every rule explains itself for your target industry and level.
  • Ranked actions are tied 1:1 to the rules that fired, never freeform.
  1. 01

    Detection

    The signal in your CV that fired the rule: named, not guessed.

  2. 02

    Relevance

    Why this signal matters to a recruiter hiring for your industry and seniority.

  3. 03

    Action

    The concrete change the rule asks for, written so you can apply it today.

Know your CV's risks before you send it.

Get the app. Run your first analysis. Decide what to fix before you apply.

Download on the App StoreGet it on Google Play

Coming soon on iOS and Android