Editorial policy and content standards

Numbers come from primary sources or they don't come at all. Formulas are traced to the standard or agency that defines them — IRS publications for tax math, NIST SP 811 for unit conversion factors, the Mifflin-St Jeor paper for BMR, the actual JSON/CSV RFCs for data formats. When a source cannot be verified, we say so in the text instead of quietly rounding our uncertainty into false precision.

Every guide carries its sources. Long-form pages end with a "Sources and further reading" block linking the primary documents, standards and official documentation referenced in the article. Where we cite a regulation or standard inline, it is linked inline.

We prefer boring sources. Government agencies, standards bodies, peer-reviewed research and official documentation outrank listicles, forums and secondhand summaries — including our own older posts. If our own guidance conflicts with a primary source, the primary source wins and we fix the page.

Every article and blog post carries a named author with a real beat — our editorial team covers the topic areas they have professional history in, from newsroom editing to CFA-level portfolio analytics to clinical dietetics. Bylines link to the author's bio and credentials so you can evaluate the source, not just the prose.

Experience claims are real. When an article says "we have watched this mistake cost people months", that maps to a specific person's professional history — the same histories listed on their bio. We do not manufacture war stories.

Our computation is code and our structure is assisted, but our editorial judgment is human. Concretely: tool engines are real, tested source code (the same libraries the live tools run on); worked examples are computed by those engines, not typed by hand; drafting and structural organization use AI assistance the way a writer uses an outline and a spell-checker — acceleration, not substitution.

What we refuse to do: publish unverified AI-generated "facts", pad pages to word counts with repeated filler, invent statistics, or produce the confident-but-wrong content pattern people rightly hate. Every factual claim in a guide traces to either a cited primary source or a formula rendered from source code.

The test we apply: would this page survive a subject-matter expert reading it with a pen? If a guide would embarrass us in front of a practitioner, it isn't finished.

Dates are honest. Every guide displays its last-reviewed date. "Reviewed" means a human re-checked the formulas, examples and sources — not that a script touched the file.

Time-sensitive pages get priority. Anything tax, benefits, rate or deadline-adjacent is re-reviewed at its regulatory inflection points — new withholding tables in January, Medicare windows in October, rate changes when the Fed moves.

Bug reports are fixes, not suggestions. If a number is wrong, the page gets corrected and the correction is noted. Email hello@toolzep.com with the URL and what you found.

The site is funded by clearly-labelled advertising and affiliate commissions, and that funding has exactly zero influence on rankings, recommendations or tool output. Affiliate products earn placement only if the editors would recommend them unpaid — the full rules live in our affiliate disclosure. Ads never appear inside tool controls, and no advertiser sees editorial copy before publication.

Substantive errors — wrong numbers, broken formulas, misattributed sources — are fixed the same day they are verified, with the last-reviewed date updated. Typo-level fixes happen in the next scheduled pass. We don't silently memory-hole significant changes; if a recommendation or formula materially changed, the update is noted in the text.