Our Methodology

By Certalyze Research Desk Automated Research & Source Audit

Reviewed by Certalyze Source Audit Updated 2026-04-26

This page explains exactly how Certalyze collects, validates, and presents IT certification data, who maintains it, what is automated and what is not, and what the visible disclaimers actually mean. Where the model is weaker than ideal, we say so explicitly rather than dressing it up.

Who maintains this site

Certalyze is published by Certalyze Research Desk. The Research Desk is not a panel of certified industry experts who personally take every exam we cover — pretending otherwise would be silent inflation of authority that the IT certification space already has too much of. The Research Desk is a transparent pipeline: schemas, source records, automated validation, multi-AI adversarial review, and a public corrections log.

The pipeline is as follows:

Each numeric or factual claim on a page is bound to a FieldSource record with a tier, URL, and checkedAt date.
Schema validation runs in CI on every push and pull request to main; the build is blocked if a critical field has no source or relies on editorial-tier sources alone.
Editorial decisions and methodological changes are run through external AI auditors (DeepSeek, Gemini, Groq, ChatGPT) configured to push back rather than validate by inertia. Findings that two or more auditors converge on are treated as real.
Material changes to published data are recorded in /corrections/, forward-only since launch (we do not retroactively backfill the log).
A scheduled weekly audit checks broken internal references, source freshness (any checkedAt older than 180 days is flagged), and schema regression. A daily backup keeps a versioned snapshot of content.
Verification of a source against its actual content (rather than just URL availability) is the part that benefits most from a human reading; we do not silently treat HTTP 200 as verification, and pages without that human contrast keep their disclaimer.

Per-field source verification

As of April 2026, every certification page carries a Data Sources table that lists each numeric or factual claim, the exact source we used, and when we last verified it. Previously we cited data sources at the page level (which site we used for salaries, for pass rates, etc.). That wasn't good enough — it told you the category of source but not which specific page supports which specific number. The per-field table closes that gap.

Every row in the table lands in one of four tiers, ordered by reliability. Official vendor or government sources are preferred; community and editorial estimates are flagged clearly so you know exactly how much certainty to assign to each claim.

Official

Vendor or government primary source

Examples: AWS official exam registration page for price, Bureau of Labor Statistics OEWS for salary medians, Cisco press releases for active-cert counts. What the issuing body publishes itself. Always preferred when available.

Aggregate

Job boards and salary aggregators

Examples: LinkedIn / Indeed job-listing counts, Glassdoor and Levels.fyi salary distributions, Payscale. Aggregated self-reported data — directionally useful but noisy. We cross-check across at least two aggregators when feasible.

Community

Reddit, TechExams, certification forums

Examples: pass-rate estimates (most vendors don't publish these), study-time anecdotes, exam difficulty reports. Valuable for what officials don't say — but vulnerable to survivor bias. Annotated as self-reported.

Editorial

Our own extrapolation

When we combine multiple weaker sources into a single number (e.g. a representative salary range from three aggregators), we label it editorial. Weakest tier; flagged clearly so you know we calculated it, not looked it up.

Freshness and stale-source detection

Each source entry includes a checkedAt date. The Data Sources table shows how many days ago that check happened. Checks older than 180 days are flagged as stale in red — a visible signal that the page is due for review. When we refresh a data point, we record the outcome (unchanged, updated, or failed) so readers can see the verification history over time.

Data Collection Philosophy

Certalyze aims to provide actionable, data-backed guidance for IT professionals evaluating certifications. We do not accept sponsorships, vendor partnerships, or affiliate incentives that influence our analysis or recommendations. Every data point on this site is collected independently and presented with its source context.

When data is unavailable or unreliable, we say so explicitly. We prefer accurate "N/A" labels over misleading estimates.

Salary Data

Salary figures represent US-based median annual compensation for professionals who hold the specific certification. Our process:

Primary SourcesJob postings from major US job boards (Indeed, LinkedIn, Glassdoor, ZipRecruiter)

Secondary SourcesIndustry salary surveys, professional association reports, self-reported data

MethodologyMedian of aggregated ranges, weighted by job board volume

ScopeUS market only (for consistency and comparability)

IncludesBase salary + bonus where available (total compensation)

Update FrequencyQuarterly review, updated when significant changes detected

Limitations: Salary data reflects correlation, not causation. Holding a certification does not guarantee a specific salary — experience, location, employer, and role all matter significantly. Our figures are best used for relative comparison between certifications, not as absolute expectations.

Job Demand & Market Data

Job listing counts represent active postings across major US job boards that explicitly mention the certification by name or code.

Tracking MethodActive postings mentioning certification by name, code, or abbreviation

Trend AnalysisMonth-over-month and year-over-year demand change

Demand Labels"Growing", "Stable", "Declining" based on 12-month trend

Limitations: Job listing counts are a proxy for demand, not a perfect measure. Some certifications are mentioned less frequently in job postings despite being highly valued. Counts may include duplicate postings across boards.

Pass Rates

Pass rate data is one of the most challenging metrics to report accurately. Here's why and how we handle it:

Official DataVery few vendors publish pass rates (CompTIA is a notable exception)

Our SourcesCommunity surveys, training provider data, forum reports, self-reported results

LabelingAlways labeled as "community-reported estimates" — never presented as official

Limitations: Self-reported pass rates have inherent survivorship bias — people who pass are more likely to share their results. Our estimates may be higher than actual rates. When we have no reliable data, we show "N/A" rather than guess.

Difficulty Ratings

Our 1-10 difficulty scale is a relative ranking within the IT certification space, not an absolute measure.

Factors ConsideredPrerequisite knowledge, exam length/format, content breadth, community feedback

Scale Reference1-3 = Entry/Foundational, 4-6 = Intermediate, 7-8 = Advanced, 9-10 = Expert

CalibrationRated against other IT certifications, not academic exams or professional licenses

Study Time Estimates

Study hour figures represent the median preparation time reported by successful candidates. These are estimates based on community data, not controlled studies.

Limitations: Individual study time varies enormously based on prior experience, study habits, and learning resources used. Use these as rough planning guides, not commitments.

Comparison Methodology

When comparing certifications side by side, we follow strict principles:

Same data points for each certification ensures fair comparison
"Choose X if..." recommendations are generated from the data, not editorial opinion
Verdicts consider the full picture — salary, demand, cost, difficulty, and target audience
We never recommend one certification over another based on vendor preference

Update cadence (the honest version)

Earlier versions of this methodology page promised quarterly review of every page and 2-week response to vendor price changes. Both required staffing that does not currently exist. The honest cadence is below; the language here is what we will be measured against.

Automated weekly auditEvery Monday: link rot, freshness flags (any source older than 180 days), schema regression. Results are exported as artifacts for review.

Daily backupVersioned snapshot of the JSON content, schema, and operational docs as a CI artifact, retained 90 days.

Source verificationDone as a deliberate sprint, not on a schedule. The pages with the highest impressions in Search Console are prioritized first.

Vendor price / version changesUpdated when we observe the change in the source or the freshness audit flags it. We do not promise a 2-week SLA we cannot keep.

Per-page last updateEach page shows the date of its last material edit. checkedAt on individual sources is the more granular signal.

Conflicts of interest and affiliate disclosure

Transparency on incentives is non-negotiable for a publisher that asks readers to trust its numbers. As of April 2026:

No vendor sponsorships or placement deals. We are not paid by AWS, Microsoft, CompTIA, PMI, Cisco, or any other issuer to feature, promote, or rank their certifications higher. Our editorial picks (including the "Editor's Pick" on every hub page) are selected on data, not relationships.
Affiliate links are disclosed inline. Some links to exam vouchers, official training, or third-party study materials are affiliate links that pay us a commission if you buy through them. When a link is an affiliate link, it is marked as such on the page that uses it. Affiliate revenue does not affect which certifications we recommend or how we rank them.
No exam brain-dumps or NDA-violating content. Any site that offers actual exam questions is breaching the vendor's non-disclosure agreement. We don't sell, reference, or link to that content.
Editorial independence from business team. If we ever add a business or sales function, the editorial team's rankings will remain independent of any revenue considerations. We will document any such separation publicly.

If you believe an editorial call on Certalyze looks influenced by an incentive we haven't disclosed, tell us at the corrections page — we will investigate and respond in public.

Current verification status

We are not pretending every source on this site is perfect. As of April 2026, across 62 certification guides:

Approximately 8% of certification guides have sources that have been individually contrasted against the cited URL. These pages do not carry the pattern-generated note.
The remaining ~92% have sources generated by vendor-URL convention as a starting point, explicitly flagged in the note field as pattern-generated, pending manual verification against the current exam guide. If you rely on a specific page, check the Data Sources table for that note and treat those rows as provisional until they are cleared.
Stale flag in red appears when any source check is older than 180 days. When you see it, treat the data point as due for refresh.

Verification work is prioritized by impressions in Google Search Console — the pages that real readers are reaching first are the ones we contrast first. We deliberately do not promote a page to fully verified status just because its URLs return HTTP 200; the disclaimer stays until a person has read the source and confirmed it actually supports the cited field.

We have published explicit operational tests for when this disclaimer drifts from honesty into alibi (for example, if fewer than 5 certifications are cleared per week for 4 weeks running, or if more than 3 of the top 10 pages by impressions still carry it). Those tests live in the project's operating document and are revisited quarterly.

What we will not do until specific gates are met

The site has explicit do-not-do gates that are tracked publicly:

No new certifications, comparisons, or career-path pages are created during the current phase. The effort goes into upgrading existing pages, not inflating page count. Adding pages is unlocked when a defined set of conditions is met (transparent authorship live on every Article template, ≥30% of certification pages cleared of the pattern-generated note weighted by traffic, top-10 pages by impressions with at most 3 still carrying the note, search CTR ≥2% on top queries, and clean technical health). Any exception requires explicit human approval and a multi-AI cross-check; it is not opened by default.
No advertising is placed on the site until a stricter version of those gates is met, and never via auto-generated content designed only to capture queries. This is operational policy, not aspirational language.
No human reviewer credit appears on a page unless there is a real, verifiable person behind it with a named profile and a recorded review date. The schema accepts an optional reviewer record; we leave it empty rather than fabricating one.

What We Don't Do

We don't accept payment for favorable reviews or rankings
We don't present estimates as official vendor data
We don't hide data limitations or sources
We don't recommend certifications based on affiliate commission rates
We don't fabricate data when real data is unavailable

Contact & Corrections

If you notice incorrect data, outdated information, or have suggestions for improving our methodology, please reach out. Accuracy is our priority. We maintain a public record of every correction at our corrections log.