Inventory buyers can trust, verified at every step.
Brand safety on SkyBoosts is a set of working mechanisms on both sides of the auction. Before the bid, invalid traffic is filtered, unauthorized supply paths are blocked and creatives are screened; after it, delivery is measured through Open Measurement and creatives are re-checked in flight. Buyers get controls per deal, publishers get controls per placement, and when something violates policy there is a written process with an enforcement ladder behind it. This page walks through each mechanism and where it applies.
Filtered before the bid · measured after it
Filtering happens before the bid, measurement after it.
Filtering
Before an impression reaches the auction, it passes invalid traffic signals, app-ads.txt authorization checks, category screens and declared-app validation. What fails never gets bid on and never gets counted: it is reported as filtered volume, so buyers and publishers can both see how much was removed and why. Clean inventory keeps the prices it deserves because it is never diluted by junk.
Measurement and takedowns
After serving, OMID-based measurement and verification partners evaluate delivery, and in-flight creative is re-checked against policy, destinations and payloads can change mid-flight, so re-checks catch post-review swaps. A creative that violates policy is pulled from rotation immediately; assessment follows. The incident process below takes over from there, with the outcome reported to both sides.
General and sophisticated invalid traffic, each filtered its own way.
Known patterns
Known data-center ranges, declared bots, malformed requests and duplicated device signatures are filtered by pattern matching at ingestion, using industry bot lists kept current on a schedule. This traffic is removed before the auction and reported as filtered volume. Buyers never pay for it because they never see it.
Behavioral signals
Sophisticated invalid traffic is scored from behavioral signals across the request lifecycle: timing distributions, session coherence, attribution consistency and cross-session device behavior. Scored impressions are filtered or downgraded before they can monetize, and each removal carries a reason code. When patterns repeat, the source is escalated rather than silently re-filtered every day.
Verification partners: integrations with third-party verification providers are available on request and scoped per deal. The current partner list is confirmed during onboarding.
Malware and creative threats: creatives and their destinations are screened for malware, deceptive downloads and redirect chains pre-flight and re-checked in rotation, alongside the policy review in the ad policy. Inventory quality bar: apps whose primary purpose is displaying ads rather than serving users, the app-store equivalent of made-for-advertising sites, do not pass store verification review and do not monetize on the network.
Buyers know exactly where every impression comes from.
Declared app identity
Every bid request declares bundle ID, store URL and app name where the platform provides them, and the declaration is checked against the actual store listing before the app monetizes. A mismatch fails review, the request cannot claim to be an app it is not. Buyers can audit the mapping themselves through the sellers.json file we publish and the app-ads.txt files each developer domain serves.
OM SDK / OMID
Supported mobile formats expose viewability through Open Measurement, giving verification partners the standard signal set with no bespoke integration work. What the measurement partner reports is what the buyer works from, the platform does not grade its own homework. TV support follows each platform's measurement capability and is scoped per platform during onboarding.
Content signals
Where the platform supplies content-level signals, CTV genre, content ratings, episode and stream metadata, they are passed through to buyers for targeting and exclusion. A brand that needs to avoid specific content categories can target or exclude on the signal itself, before the bid. Signal availability per platform is documented so buyers know exactly what travels.
Buyers set rules per deal, publishers set rules per placement.
Controls are only useful if they hold across every demand path, so both sets below are enforced by the platform, on auction bidders, waterfall networks and direct campaigns alike.
Buyer controls
Demand side- Block lists by app, publisher and domain
- Sensitive category controls per deal
- Deal-level rules: formats, platforms, geos
- Verification partner of choice via OMID
- Content-category exclusions where signals allow
Publisher controls
Supply side- Creative review before and during flight
- Category blocking per app and placement
- Competitor exclusion lists
- Ad load and frequency limits per placement
- Separation rules across a break or session
When a bad ad serves, here is exactly what happens.
Removal process
Report
Anyone can report a violating ad through the contact form or the ad policy page. A report needs three things: the creative, the app and the time it served. Every report is acknowledged and tracked to a written outcome.
Pull
Policy-violating creatives are removed from rotation while the report is assessed, because a suspended creative costs an advertiser minutes while a live violation costs trust.
Assess and act
Severity is assessed against the written ad policy. Outcomes range from creative correction to partner escalation and demand suspension, with the ladder published in the two policies.
Close the loop
The reporter and the publisher receive the outcome, including what was removed and under which policy section. Repeat offenses climb the enforcement ladder automatically across successive incidents.
Where the rules live
Brand safety mechanisms enforce two written policies: thepublisher policy, which governs inventory and placements, and thead policy, which governs demand categories and creative requirements. Enforcement follows the ladder in those documents.
Brand safety, in plain terms.
Filtered volume is reported alongside delivered volume, with reason codes per category, so you can see exactly what was removed from your supply and why. The figure belongs to your inventory and your reporting, and it is visible to you the same way delivery is.
Yes, wherever the vendor supports Open Measurement on the formats in question, the OMID signals are standard, so the vendor you already trust can measure the delivery without a bespoke integration. On CTV it depends on what the platform measurement interface exposes; the current matrix is confirmed per platform.
They do not monetize here. The check runs at onboarding and re-runs whenever an app materially changes, a new ad load, a different content mix or a rebuilt experience can push an app that started legitimate into MFA behavior.
Yes, the platform applies category blocks, competitor exclusions and placement rules itself, so no demand path can bypass them, direct included. A block list that a direct campaign could ignore would not be a control, it would be a suggestion.
Immediately upon a credible report or an in-flight detection, removal precedes assessment. The assessment then decides the severity outcome on the enforcement ladder, and both the reporter and the publisher hear the result.
Viewability measurement on CTV follows what each TV platform measurement interface exposes, per the support matrix confirmed during onboarding. Where measurement is supported, beacons are forwarded server-side so buyers get the numbers their verification partner produced.
Safety you can inspect.
Verification requirements, category controls and incident terms are set per deal. Bring your verification stack.
Pre-bid filtering · OMID measurement · incident process