Core rule: A provider cannot buy a positive review, an editorial ranking, placement in a team brief, or the removal of relevant criticism.
Sources and verification
We prefer first-party provider pages, official certification documentation, regulator or accreditation records, and current program catalogs. Secondary sources may help discovery but do not replace an authoritative source for material facts.
What we review
- Provider identity and official program URL.
- Target learner, level, delivery format, and credential type.
- Published curriculum, duration, prerequisites, and price when available.
- Material limitations, expiry dates, retirement dates, and regional restrictions.
- Whether a commercial relationship exists.
Ordering and recommendations
Default ordering is editorial and may consider relevance, completeness, recency, and usefulness. Sponsored placements must be visually labeled and must not be presented as independent rankings. A Team AI Decision Brief request, waitlist registration, provider relationship, or payment never changes editorial order.
Team AI Decision Brief separation
The free fit assessment determines whether a team request matches the written service. It does not provide a paid placement, guaranteed recommendation, provider introduction, admission result, or training outcome. Any later paid brief would require a separate written scope and approval. Its comparison must use the same source standards and disclose evidence gaps; it cannot alter the public directory.
Program Change Monitor evidence
The beta may describe material changes only within a published, limited coverage list and only after a baseline source record exists. Each published item must identify the program, changed field, prior value when recorded, current value or “not stated,” evidence type, official source, checked date, reviewer, publication status, and inclusion reason. Search snippets and aggregators may help discovery but are not sources of material truth.
A second person or QA reviewer must review a sample of material changes, evidence labels, links, and dates before a digest is sent. If that review cannot happen, the digest stops; the gap is not replaced by a guess or a claim of complete coverage.
Affiliate relationships
Some links may generate a commission after a user makes an eligible purchase. Affiliate eligibility does not guarantee inclusion and does not control editorial order. A commercial link is disclosed close to the relevant action. Some provider links are affiliate links; they are disclosed on the relevant page and have no influence on editorial assessment.
How comparison verdicts are computed
The compact and full verdict use the same calculation and reader profile. Scores are a decision aid for that profile, not a universal ranking or evidence of employment outcomes.
Reader profile and editorial judgment
Every comparison declares a reader profile. Editorial scope classifications direct, adjacent or outside must reference the exact registry level, provider audience sentence and topics for each program, with a rationale. The category maps to points; no score or winner is authored by hand.
Calculation
Sum of eligible criterion score × weight / (2 × sum of eligible weights) × 10. Results are displayed to one decimal.
If either program cannot be scored on a criterion, exclude that criterion from both programs' numerator and denominator. Never assign replacement points. With no eligible weight, report no score.
Scope match to the reader profile Weight ×2
Direct fit scores 2, adjacent fit 1, and outside the stated profile 0. Inspect registry level, provider audience and topics or modules. A developer's direct fit requires explicit technical build/deploy coverage; business/governance-only coverage is adjacent or outside, never direct. An explicitly missing provider level is omitted without inventing a label; sourced audience and topics can still support fit. The editorial classification is disclosed, not based on word count or syllabus size.
Credential features Weight ×2
Use documented issuer, exam requirement, sharing and renewal policy. An exam credential scores 2; a non-exam credential scores 1; an explicit statement that no credential is awarded scores 0. Incomplete evidence omits the criterion for both programs. No issuer-prestige, employment-outcome or longer-validity bonus is applied.
Cost for the stated payment scope Weight ×1.5
Compare accepted regular prices in the same currency and explicit comparable payment/billing scope. Score = 2 × (lower price / this price). Differences strictly under 10% of the lower price tie at 2; exactly 10% is proportional, not a tie. Equal prices, including both free, tie at 2; a positive price against free scores 0. Same-platform accepted equivalent subscriptions tie at 2. Never invent currency conversion, preparation fees, tax, retakes or subscription months. A recorded regular minimum-plan monthly subscription enters cost. A month of access may be compared with an exam fee only with both payment scopes explicitly stated in the Why line; neither is a completion-cost estimate.
Official preparation or completion hours Weight ×1
For two exams, use accepted official preparation or learning-path hours in the preparation-hours record. Never use exam sitting time. For two courses, use explicitly sourced completion hours on the same basis. If either side lacks comparable hours, omit the criterion for both. Within 5% ties at 1; otherwise shorter scores 2 and longer 0. No elapsed-calendar or ease inference is made.
Entry requirements Weight ×0.5
Explicit no prerequisite scores 2; explicit basic knowledge 1; explicit advanced or formal qualifications 0. Ambiguous or unpublished requirements omit this criterion for both programs. This is not a rating of exam difficulty, disability access, language access or pass probability.
Both programs receive normalized numeric scores when at least three of the five criteria, including scope match, can be scored for both. Each number states how many criteria were scored. Otherwise both show Not enough source-checked facts to score and a profile-specific Depends outcome. Missing facts are never replaced with assumed points. A score of 10.0 describes the scored criteria, not proof that every possible fact is known.
Confidence in the facts
Deduplicate the actual program/provider plus registry field facts used in scores, explanations, choices and Best-for rows. A fact is fresh only when all its accepted checks are checked against official sources and no older than 60 days, inclusive. One older or due for rechecking fact means Medium; two or more mean Low. Omitted evidence does not lower confidence. Future check dates are invalid. No used facts means no confidence rating.
Winner and reader choices
A score difference strictly below 0.5 is a tie. Use the declared platform/profile phrase. Exactly 0.5 is not a tie; rounding to one decimal is for display only. Recompute without entry requirements: if their removal changes the winner and the resulting difference is below 0.5, report Depends on your background. Lead the explanation with scope or credential facts and disclose the small margin.
At least two distinct reader-use cases must name a compared program and cite accepted registry evidence. Custom platform or specialism labels are allowed. Do not invent career outcomes, easiest paths or filler rows.
Generate exactly three complete, natural factual lines. The compact explanation begins with the highest-weight eligible criterion whose scores differ, not the largest weighted score gap. Equal weights follow the declared criterion order: scope before credential. Keep the other explanations in descending weight order. If no criterion differs, use weight order throughout. Credential validity may be described as information but does not increase points or determine the leading explanation. Do not expose scoring jargon or manufacture unsupported lines.
Verdict is computed from source-checked facts and does not change with commercial relationships.
Method change log
- · 006-F5-1: Supersedes the earlier perfect-score display restriction: both sides show normalized numbers when at least three shared criteria including scope are scored, with the count disclosed. Otherwise both show the same insufficient-facts message. Recorded regular minimum monthly plans enter cost with explicit payment scope.
- · 006-F4-1: Cost is proportional: 2 × lower price / this price, with strict under-10% gap relative to the lower price tied at 2. Both sides must have comparable currencies and billing scope. A named winner is withheld against an incomplete score; existing symmetric platform ties are preserved. Source-read unstated credential terms remain unknown, never false. Regular access quotes and explicit payment scenarios remain distinct from completion cost.
- · 006-F3-1: Show all five weighted criteria and omission reasons. Require at least three scored criteria including scope for a number, and all five source-checked criteria at 2 for a perfect score. Validate technical coverage for developer direct fit and recompute without entry requirements to identify background-sensitive decisions. Omitted criteria remain excluded from both numerator and denominator.
- · 006-F1-1: Implemented F.1: disclosed reader profiles, shared-denominator omissions, fact-freshness confidence, official preparation hours, a strict 0.5 tie boundary and complete factual explanations.
- · 006-F2-1: Implemented F.2 explanation ordering: the highest-weight differing criterion leads, with scope before credential at equal weights. Scores, weights, reader profiles and scope classifications are unchanged; longer credential validity remains informational, not a scoring advantage.
- · 006-F2-1: Extended the existing entry mappings to the literal source phrases working knowledge and there are no prerequisites; criteria, weights, omissions, tie handling and existing classifications are unchanged.
Updates and corrections
High-traffic and time-sensitive listings are prioritized for review. We aim to recheck the core directory at least every 90 days. Providers and readers may submit corrections through the program submission form. Material corrections are applied without charge.
What we do not publish
We do not publish invented ratings, unsupported salary outcomes, fake scarcity, hidden paid rankings, unverified change claims, complete-market monitoring claims, or credentials that cannot be traced to an authoritative issuer.
Last updated: 5 September 2026.
← Back to the directory