Methodology

Last updated: 2026-08-14

How content is built

Every guide follows one template: a quick answer, why it matters, numbered steps, a worked dollar example, common mistakes, what to do next, related tools, and named sources. Editorial content is drafted with AI assistance and human editorial review, checked against the primary sources listed on each page, and stamped with a Last Updated date. Anything time-sensitive — program rules, rates, limits — links to its official source rather than hard-coding a number that will rot.

How the calculators work

Each calculator is one pure mathematical function that serves both the page you see and the public API, so the two can never disagree. Standing assumptions, stated here once and on tools where they matter: interest compounds monthly; savings deposits are treated as end-of-month; debt simulations accrue monthly interest at APR/12, pay stated minimums, and roll finished payments into the next target; results are estimates for comparison, not quotes. When inputs produce impossible situations (payments below interest), tools say so plainly instead of showing nonsense.

How program information is maintained

Assistance-program summaries carry a Last Reviewed date and link only to official government sites or long-established nonprofits. A weekly automated job re-checks every outbound program link; sites that merely block robots are NOT flagged (bot walls are common on .gov sites), while genuinely dead links (404/410) are surfaced for repair. Eligibility language is deliberately conditional everywhere: programs depend on income, household size, disability status, age, location and year.

How news works

The News tab aggregates from hand-verified RSS feeds of named mainstream and official publishers, refreshed three times daily. Each item shows its source and links to the original article. We write none of the headlines, and an empty news page is always preferred to a fabricated one.

How the AI features are constrained

Ask the Data answers only from this site's own guides, FAQ, programs and tools, shows the source pages behind each answer, and refuses individualized advice. The AI Agents page produces prompts for YOUR assistant, with verification instructions embedded. AI can be wrong; primary sources win, always.