How we score.

Six criteria, weighted twice — once for publishers, once for advertisers. Every number on this site is produced by the table below, and by nothing else.

What this page is

The weights below are read directly from the code that computes the scores. They are not a description of our process written by hand and left to rot — they are the process. Rubric v5.

The six criteria

Performance & delivery pub 25% · adv 20%

Fill rate, eCPM stability, latency, viewability. Measured across publisher accounts, not vendor decks.

Targeting capabilities pub 5% · adv 25%

Granularity, contextual and audience options, and whether the controls do what the docs claim.

Inventory quality pub 5% · adv 25%

Publisher vetting, invalid traffic controls, brand safety, MFA exposure.

Transparency & reporting pub 15% · adv 15%

Revenue share disclosure, log-level access, reporting API quality, placement visibility.

Pricing & payment terms pub 30% · adv 10%

Payment reliability, NET terms, minimum payout, minimum spend, hidden fees.

Support & reliability pub 20% · adv 5%

Response times, account management, uptime, how they behave when something breaks.

For publishers

Pricing & payment terms 30%
Performance & delivery 25%
Support & reliability 20%
Transparency & reporting 15%
Inventory quality 5%
Targeting capabilities 5%
total 100%

For advertisers

Inventory quality 25%
Targeting capabilities 25%
Performance & delivery 20%
Transparency & reporting 15%
Pricing & payment terms 10%
Support & reliability 5%
total 100%

Ties, and unscored networks

Two networks with the same weighted total are ordered alphabetically. That tiebreak is arbitrary, but it is fixed, published, and applied by code — so a sponsored network never floats above an identical one that pays us nothing.

A network we have not scored has no rank. It is listed, unranked, below everything that has been scored. We would rather show you a gap than invent a position.

What does not move a score

Referral fees. Press releases. Advertising bought anywhere on this site. Publisher reviews — which sit beside the score, not inside it, and which send us looking rather than doing the scoring themselves.

The editor who scores networks does not see the press release queue, and does not know which networks pay us until after the criteria are locked.

Vendor-stated numbers

Figures we have not independently confirmed carry a dagger and are labelled as claims. eCPM ranges usually stay vendor-stated until we have watched a full quarter across several publisher accounts. Fill rate, payment terms, and minimum payout are confirmed against publisher statements before they appear without one.

When a score moves

Every change is dated, attributed to a criterion, and given a reason on the network's page. The old review stays readable at its permalink. Networks scoring under 6 are rescored quarterly; everything else annually, or whenever evidence arrives that says we should.

Before any score change is published, the network sees the criteria, the totals, and the prose, and has seven days to correct facts. They cannot negotiate the score. If they decline to respond, we say so.