Evidence before opinion

How Comidoc reviews courses

A Comidoc Review is an evidence-based editorial assessment of a course's public learning proposition. It is not a claim that our editors completed the course, and it does not simply repeat the instructor's marketing copy.

Public evidence

We collect the course presentation, curriculum, requirements, target audience, aggregate metrics and a rating-balanced corpus of substantive public learner feedback.

Separate passes

Curriculum and learner signals are extracted independently before synthesis. This separation reduces the risk that one attractive claim dominates the assessment.

Quality safeguards

Structured validation, evidence references, scoring rules, comparison checks and targeted editorial repairs run before a review can be published.

The five-part Comidoc Score

The overall score is a weighted editorial index from 0 to 10. It measures the strength of the learning proposition supported by the evidence available to us—not popularity, sales or Udemy's own star rating. To make meaningful differences easier to read, the public display expands values around the midpoint while preserving the 0–10 scale and the original ranking. Udemy's rating remains separate in the course information.

Curriculum depth

30%

Breadth, progression and demonstrated depth of the published syllabus.

Applied learning

20%

Projects, exercises, practice and opportunities to use the material.

Clarity & experience

20%

Signals about explanation quality, pacing, navigation and learner support.

Currency & reliability

15%

Version relevance, maintenance signals and recurring reports of outdated material.

Audience fit

15%

How well the prerequisites, level and teaching approach fit the intended learner.

How learner feedback is used

We look for recurring, course-specific signals across rating bands, not a simple average of a few comments. Recent feedback normally carries more weight. Older criticism may remain relevant when a new update label alone does not demonstrate that the reported issue was fixed. A sampled theme is a diagnostic signal, not a measurement of prevalence.

What confidence means

Confidence reflects the quantity, recency and agreement of the supporting evidence. Limited evidence constrains how assertive the recommendation can be. A neutral dimension is explained rather than hidden, so readers can distinguish uncertainty from an average performance judgment.

The role of AI

AI assists with evidence extraction and structured synthesis in multiple focused passes. Deterministic rules validate references, schema, score consistency and common editorial defects. Targeted repair can correct a failed response, but it cannot create new source evidence or silently alter protected scoring data.

Limits and corrections

We do not independently verify every instructor claim, complete every exercise or access private course materials. Course pages and learner feedback can change. Reviews are versioned so a new candidate does not replace the published assessment until it passes validation and is approved. Material errors can be reported for correction.