Editorial principles
Everything published here answers to four principles, in this order of priority: factual accuracy traceable to a document, independence from the platforms that pay this site, transparency about how the money works, and dating of every figure that can go stale. The order matters, because when two of them pull in opposite directions the earlier one wins. Content on this site influences spending decisions, and the cost of an error is not evenly spread. A reader who trusts a deposit floor and meets a different number inside the account has lost an afternoon. A reader who trusts a payout threshold and builds a monetization plan on it has lost a quarter. The process below is heavier than a comparison page normally needs because of the second case.
Source hierarchy
Sources are ranked, and a lower rank never overrides a higher one.
First, platform-controlled documents. Terms and conditions, advertiser and publisher help centers, FAQ pages, and dated company announcements. Deposit floors, payout thresholds, hold periods, payment methods and supported pricing models come from this tier and nowhere else.
Second, direct observation of a platform's own interface. What a dashboard exposes, which filters are self-serve, what a calculator shows before funding. This is labelled as observation when it appears, because an interface can change without an announcement.
Third, measurement published by named research organisations. Always with the method and the year attached, because survey panels and server-side pageview counts produce materially different numbers for the same question. A percentage without its method is not usable here.
Fourth, documented user testimony. Attributed, dated, and described as testimony, never as fact.
Competitor comparison posts sit outside the hierarchy entirely. They are read to map what the market believes and are never a source for a figure. A number existing only in a directory does not exist on this site.
How a page is built
Step one, intent. A page starts from a single reader question and the sub-questions that follow it. Sections that do not close part of that intent are cut before writing, however well they would rank.
Step two, document pass. Every platform in scope is read directly and each figure is logged with its source and the date of the check. This is the slowest part of the process and the only one that cannot be shortened.
Step three, conflict pass. Figures from the document pass are compared against what circulates publicly. Where they diverge, the divergence goes into the text with both numbers named. The main page on Popunder Ad Networks carries a worked example: one platform's published deposit floor had been superseded for a year while comparison lists continued to repeat the old figure.
Step four, gap pass. Points where a platform publishes nothing are marked as gaps and stay in the finished text as gaps. This applies to the recommended platform as strictly as to the others.
Step five, adversarial read. The draft is read against one question: which sentence would a competitor use to prove this page wrong? Anything that cannot survive that question is either sourced properly or removed. Claims of uniqueness are the usual casualty, because they are the easiest statement in ad tech to disprove.
Step six, publication and review. Pages carry an author, a verification date and a named source for figures that need one. They are reviewed when a platform announces a change, and on a scheduled basis regardless.
Verification and dating
Deposit floors, payout thresholds and payment schedules change without notice and usually without a press release. A page that cannot state when its figures were checked is not finished, so the verification date sits in the byline of every page, where it is read, and not in a footer, where it is not.
Undated accuracy is the most common failure in this niche, and it is worth being precise about why.
Most published figures in ad tech comparisons are correct at the moment they are written. They become wrong silently, and nothing in the page tells a reader that the moment has passed.
Editorial independence
This site earns money when a reader opens an advertising account through one of its links. That arrangement determines which platform is recommended and where the buttons sit. It does not determine what the tables say.
What that looks like in practice is easier to check than to promise.
The recommended platform's weaknesses are published on the main page, including the highest publisher payout threshold in the comparison set and the complete absence of an automated conversion goal. Competing platforms' verified advantages appear as advantages, in their own terms, with no hedging attached. No figure is softened, rounded or dropped because it is inconvenient to the commercial relationship.
A request to remove a documented weakness, to add a platform in exchange for payment, or to publish supplied text under the author's name is declined. The commercial arrangement itself is described on the about page.
Corrections
Reports of an error are the highest-value message this site receives, and they are welcome from anyone, including the platforms being written about. Send three things to the editorial inbox: the page address, the figure you are challenging, and the document that contradicts it. A link to a help center or a terms page settles most of these in a single exchange.
Confirmed errors are corrected in the page and the verification date is updated. Where a correction changes a conclusion and not just a detail, the change is stated in the text; silent edits are not made. Where a platform's terms have changed since publication, the page is updated, and the superseded figure is retained in the text when the change itself is worth knowing about. Pages are not deleted to make an error disappear.
Requirements for anyone writing here
Every page is signed with a real name linked to a public profile; anonymous content is not published. The person who writes a page is the person who read the documents behind it, because a text assembled from someone else's summary carries none of the judgement that makes it worth reading. Invented measurements are the one unforgivable error. A fabricated payout time or conversion figure is checkable by the reader, and a single one of them costs more trust than the rest of the site can rebuild. Desk research is labelled as desk research, as it is on the author page, and the limits of the site's claims are set out in the terms of use. Every route for reaching the person responsible is listed on the contact page, and correspondence arriving through any of them is handled as described in the privacy policy.