Metod

Single source of truth

All tax rates and treaty provisions are stored in a canonical, structured datastore. We never hardcode rates in page components or article copy. When a domestic rule changes, every affected result updates automatically — no rewriting individual pages.

Verified, not invented

Every published material rate and legal statement is linked to a primary legal source. A missing numerical rate means unknown, never zero. We never generate live rates with an LLM. If a scenario is incomplete, we show an honest "not yet verified" result.

How we compose results

We do not maintain a separate manually written article for every country pair, income type, and language. Instead, we compose case-specific explanations from reviewed, reusable modules: source-rule intro + applicable treaty provision + residence-rule intro + relief mechanism + pair-specific exceptions + optional refund explanation + sources.

Numeric rates are rendered from structured fields at viewing time, not copied as stale percentages in prose text. We publish indexable pair/subtype pages only where facts and substantively useful text have been reviewed.

Source strategy

• OECD treaty-based WHT data (imported as unverified observations, not automatic individual rates)
• EC Taxes in Europe Database and bilateral treaty directory
• OECD BEPS MLI Matching Database
• National tax authorities and treaty registers

We never scrape or republish paid proprietary databases without permission. Raw source tables are segregated from approved rule tables.

Review process

Each rule records its verification status, reviewer, review date, and primary sources. AI-generated draft scenario texts follow a deterministic workflow: verified facts → explanatory text draft → human legal/factual approval → locale translation review → publication. AI never bypasses permissions or directly publishes financial claims.