Sources and method

The catalogue is made of facts. A fact is a statement that has a document address and a date on which we read it there. Everything else on this site is navigation around the facts.

Where the information comes from

Only from the documents of the platforms and services themselves: terms of use, help centres, pricing pages, developer documentation, community rules, official blogs.

We do not take information from third-party reviews, "top 10" round-ups, agency blog posts or retellings in search results. If the only source for a number is someone else's article, the number will not be in the catalogue.

Once we had to apply this rule to ourselves: the thread collecting Twitch data honestly noted that it had taken the wording from search results, because the help centre serves an empty shell. Those rows were removed from the catalogue.

How a source page is read

robots.txt is read in full and first. This is not a formality: on 28 July 2026 we read the Hupu file down to the middle, decided crawling was allowed, and collected 27 facts. Ten days later we re-read the file in full — the last two lines forbade crawling entirely. All 27 facts were withdrawn, and the record of the mistake was left in the database.

Bot protection is not circumvented. We do not solve human-verification challenges, we do not forge tokens, and we do not request addresses closed off in robots.txt. If a page is closed, the catalogue carries no information from it.

The boundary runs here: requesting a different address on the same host, one that is explicitly allowed, is not circumvention. Where the HTML serves a bot check, an open help-centre JSON may be read, and we write down in the note exactly how we read it.

What we do not do

We do not compare quality. "The best scheduler" is a judgement, and everyone makes their own. We show the conditions and leave the choice to you.

We do not promise earnings. How much you will manage to earn we do not know, and we do not undertake to guess. Payout thresholds are the platform's conditions, not a forecast of your income.

We do not publish workarounds. Not for platform rules, not for bot protection, not for advertising disclosure, not for tax obligations. A platform's prohibition we do show — knowing about it is more useful before you build your work around the platform.

We do not compute derived numbers. No averages, no currency conversions, no "effective rates". A number in the catalogue is the number the platform itself published.

How the checking works

The site is built from the database with a single command, and the checks run together with the build: broken links and anchors, accessibility, Cyrillic inside foreign identifiers, schema completeness, the share of untranslated text, facts lost during the build.

A project rule: a check you have to remember to run will sooner or later be forgotten. That is why the checks are part of the build rather than run from memory.