How we rank.

Define who the list is for

Each ranking must state its audience, use case, comparison group, eligibility requirements, and relevant region. We will separate different services and budgets when an overall order would be misleading.

Directory membership is not a recommendation. Candidate profiles appear alphabetically while a list is in research.

Check the evidence

Use primary sources for prices, policies, features, and terms, with direct links and check dates. Third-party reports, provider claims, and ATN observations must be identified separately.

Hands-on testing may be claimed only when it has actually happened, with the scope, date, and limitations recorded. A source-checked profile is not a completed review or a guarantee of safety.

Choose the right basis for the order

Source comparisons explain task fit, access, limits, and unresolved evidence without assigning positions. They can be published when a useful comparison is complete but differing roles, usage units, or policy gaps prevent a fair numerical order. Their researched-entry counts remain separate from ranked-entry counts.

Metric rankings compare one clearly defined fact, such as a published payout minimum. Each must specify the audience, region, currency, payment method or equivalent product, comparison group, sorting direction, and dated primary evidence. Equal values share a position. A missing or unconfirmed value stays outside the order. Supporting facts do not secretly act as tiebreakers.

A metric ranking can be published once that comparison has been evaluated, even while the broader review is incomplete. It must be labeled by its metric and cannot be presented as an overall quality verdict.

Overall quality scores use the category’s displayed criteria and percentage weights, totaling 100%. After a complete editorial evaluation, each criterion can receive a score from 0 to 5 supported by a written rationale.

0–1

Does not meet, or poorly meets, the stated criterion.

2–3

Partly meets, or adequately meets, the criterion.

4–5

Strongly meets, or excels against, the defined comparison group.

The planned overall score is the sum of each criterion’s score divided by 5 and multiplied by its percentage weight, shown out of 100. Equal rounded scores share a position.

Missing evidence is not zero. A candidate stays unrated until all required criteria can be assessed. We will not turn unknowns into invented scores or quietly redistribute their weights.

Explain the result

A published ranking must include a reason for each position, who each option suits, meaningful drawbacks, the evidence behind the position or score, its testing status, and an editorial review date.

Policy checks, price checks, and hands-on tests can have different dates. Those dates should stay separate. Rankings express an editorial assessment of a defined set of options, not every product on the market.

Keep money out of the score

ATN may earn commissions from qualifying purchases when affiliate links are introduced. Commission rates, affiliate acceptance, and payment for advertising must not affect editorial scores or positions.

Sponsored placements must be labeled and displayed separately. Any relevant commercial relationship, free access, or review sample must be disclosed alongside the evaluation.

Recheck, correct, and withdraw

Material changes to terms, pricing, ownership, eligibility, or evidence require reassessment. If the evidence no longer supports a position, mark it for recheck or withdraw the ranking until it can be supported.

Record what changed and when. Never refresh a verification date without actually rechecking the underlying information.