Every OnlineToolPot page is built around one task, checked against worked examples, and reviewed by a named person. We name our sources, state a tool's limits, show when the page was last reviewed, and correct errors in public. Where a result is probabilistic or an estimate, the page says so.
How we choose what to build
We start from a task people actually search for, not from a list of tools we could clone. A tool earns a page only if it does one job clearly, works without an account, and is at least as useful as the established option for that job. We deprioritise commodity widgets that a search engine can answer itself, and we do not publish a new page for every phrasing of the same query.
How a tool page is built
A writer defines the task, the inputs, the outputs and the limits before the interface is built. The page then follows a fixed shape: a direct answer, what the tool does, how it works, a worked example, limits, and questions people actually ask. The tool and the copy are checked against each other, so the page describes what the tool does, not what we intended it to do.
How we check the numbers
Where a formula exists, we use the standard published version and name the source. Hash tools follow the published NIST specifications, loan and mortgage examples follow standard amortisation arithmetic, and format guidance follows the body that owns it, such as IANA for media types. We compute worked examples independently and check the tool against them, including rounding, negative values, unit changes, time zones and recurrence. If a number cannot be checked, it does not go on the page.
How we test the interface
Each tool is tested on the normal path and on the cases that break tools: empty input, very large input, invalid units, unusual characters, mobile width and a second browser. We log defects found in review, fix them, and re-test the fixed version. A page is not treated as reviewed until the live tool matches the copy.
How we use AI
We use AI tools for drafting, research assistance and code help. We do not publish unreviewed AI output. A named author reviews every page for accuracy, checks the claims against sources, and signs off before publication. The author whose byline is on the page, Amna or Irfan Wadood, is accountable for it.
What we do not claim
We do not claim laboratory benchmark results we have not run, certification we do not hold, or a level of accuracy a tool cannot support. AI detectors are probabilistic and produce false positives; the pages say so. Calculators are estimates, not financial advice. Medical, legal and financial decisions need a qualified professional, not a tool page.
How we update and correct
Tool pages are reviewed when a tool changes and at least once a year. The review date is visible on the page. If you find an error, email [email protected] or use the contact page with the page and what is wrong. We check it, correct it, and update the review date. Corrections that change a conclusion are noted in the page rather than silently replaced.
How to check our work
The editorial policy explains who writes and how sources are chosen. The security page explains how tools process data. The About page explains who runs the site. If a claim on any page cannot be traced to a source, tell us and we will fix it or remove it.

