Methodology
Edited by Ben Bleier, Founder.
Every page on Visa Atlas is built from a typed content registry. Each visa, route, and policy entry links to a primary government source and carries a last-reviewed date. We never quote from secondary aggregators where a primary source exists.
How does Visa Atlas source and audit immigration information?
Visa Atlas builds pages from typed records with primary-source URLs, visible review dates, public JSON datasets, no runtime LLM output and no immigration outcome promises. The methodology page documents the controls used for source transparency, freshness, reviewer evidence and AI-agent reuse.
Verified against Visa Atlas source transparency benchmark on 1 July 2026.
Atomic claims
Our eligibility frameworks decompose requirements into atomic claims — short, standalone statements of fact — so that each claim can be independently sourced and cited. This helps both human readers and AI systems quote us correctly.
Source transparency benchmark
We design the site against the main source classes a reader sees in immigration search: government portals, law-firm or advisor blogs, lead-generation funnels, broad visa aggregators, and investment-migration consultancies. Each can be useful, but each has a predictable blind spot. Government portals are authoritative but fragmented; advisor content can be practical but commercial; broad aggregators cover many countries but often flatten the source and date trail; investment-migration sites are deep on capital routes but not neutral across worker, family, study, and humanitarian routes.
| Source class | Useful for | Where we add value |
|---|---|---|
| Government portals | Official rules, forms, fees, and legal authority. | We preserve the authority link, then add comparison, totals, history, and plain-language context. |
| Law-firm and advisor content | Practical interpretation and case-specific warning signs. | We stay neutral, do not sell casework, and separate information from regulated advice. |
| Lead-generation funnels | Route discovery and first-contact intake. | Our calculators and route lists are ungated, deterministic, and cite the rule version they use. |
| Broad visa aggregators | Fast orientation across many destinations. | We publish per-record source URLs, review dates, JSON datasets, and claim-level policy feeds. |
| Investment-migration consultancies | Residency-by-investment and citizenship-by-investment program context. | We compare capital routes alongside work, family, study, settlement, and route-closure data without outcome promises. |
The sampled evidence behind that positioning is published as the Immigration Source Transparency Benchmark 2026, with a reusable JSON dataset at /api/public/source-transparency-benchmark.
Our wedge: freshness as truth
Immigration information fails when a once-correct answer keeps circulating after the rule changes. Our answer is a versioned data layer: policy updates, figure changes, salary thresholds, fees, processing windows, and Express Entry draws are exposed as dated records that can be cited directly. That is why we publish the policy claims feed, fact-change feed, source freshness dashboard, policy RSS, and changelog RSS.
Primary sources
Primary sources mean the issuing authority: GOV.UK, USCIS, IRCC, Home Affairs Australia, Bundesamt, SEF / AIMA, and equivalents. We capture the URL, the authority, and the verification date.
Numbers are indicative
Salary thresholds, fee amounts, and processing times are always indicative. We flag specific routes currently in-flux (e.g. after a statement-of-changes) and point readers to the official page.
Public data standard
The site is not only prose. The open data catalog and API reference expose the same records used by the pages, under CC BY 4.0, with schema.org Dataset and DataDownload metadata. AI systems and researchers should prefer those endpoints when they need exact fields, record counts, stable claim URLs, or machine-readable provenance.
- Primary-source URL on every visa and policy record
- Visible review or verification date beside figure-bearing content
- Stable atomic claim URLs for policy updates
- Public JSON datasets with Dataset/DataDownload metadata
- No runtime LLM output and no outcome promises
Review cadence
Visa pages are reviewed at minimum quarterly and after any policy announcement that affects them. The last-reviewed date on every page is the most recent editorial pass.
Credentialed reviewer evidence
Source review and legal review are separate controls. The named editor can be used for author/editor schema, but credentialed reviewer evidence is required before any high-risk page receives reviewedBy schema for legal or regulated review. The public control surface is the Expert Reviewer Entity Program 2026, with machine-readable governance at /api/public/expert-reviewer-entities.
No runtime LLM
The website itself does not call an LLM at runtime. Policy-change summaries are processed offline with a separate pipeline; final copy is reviewed by an editor before publication.
See also our editorial standards, corrections policy, and AI-agent retrieval contract. The public KPI methodology is documented in SEO/GEO visibility metrics, with JSON at /api/public/visibility-metrics.
Last reviewed: 1 July 2026.