Twitch’s approach to brand safety, monetization, and brand suitability controls

Safety is the foundation of our Twitch community. Our goal is to enable creators to express themselves and provide a welcoming, entertaining experience for viewers, free of illegal, harmful, or negative interactions.

We take a collaborative approach to safety, with Twitch and the community working together globally to promote real-time reporting and enforcement. This starts with our Community Guidelines, which balance user expression with community safety and set clear expectations for behavior across all content on our service.

Twitch's policies and guidelines are designed for a live environment, with machine learning-powered detection and user reporting.

Twitch is a community-based service where creators and their communities interact live. Most content on Twitch is live and ephemeral, disappearing as it's created and consumed. Our policies and safety operations are tailored for this live, community-based environment.

We use machine learning and user reporting to detect policy violations. Amazon and Twitch leverage machine learning to continuously scan live video and chat, identifying and blocking unsafe behavior. User reporting provides an important additional signal for Twitch's community-based service, adding nuance and control.

Global teams of moderation professionals review violations and act swiftly to maintain safety.

Our Trust & Safety operations team works across the globe to quickly review content and accounts flagged by users and by our automated detection tools. If a violation is detected, they issue immediate enforcement action. Different violations carry different enforcement actions based on our policies. Repeat violations in the same category result in escalating consequences starting with the next violation in the same category.

Twitch’s moderation and enforcement model is designed to align levels of suspensions with severity of violations. If a policy violation is confirmed, our moderation team issues immediate enforcement action. Our enforcement system has some similarities to a driver's license points system. With the driver's license points system, different violations carry different enforcement actions. Over time, and with no other violations, these violations expire and do not remain on the driver's record. Higher severity violations are worth more points and remain on the record longer, up to the loss of the driver's license. In the same way, within the Twitch system, lower severity violations will expire sooner than higher severity violations, which count against an account for a year or more. The most severe violations of our policies never expire, leading to a permanent suspension of that account from Twitch.

Ad monetization on Twitch is a privilege that creators must earn and maintain and can be revoked in instances of policy violations. New creators cannot have ads on their channels. To earn the privilege of having ads, creators must demonstrate passion and respect for the community, as shown through their streaming activity, earned viewership, and adherence to Twitch’s Community Guidelines. Viewership and discovery on Twitch are earned solely through live streaming hours, not offline video uploads. This results in a much higher bar for monetization compared to industry standards.

Twitch's monetization criteria focus on broadcast recency, unique broadcasts, concurrent viewer count, number of followers, and adherence to our policies which are in line with industry best practice for user-generated content (UGC) monetization.

When it comes to pre-recorded and uploaded content prevalent on social media platforms, the video itself and the creator’s voiceover and conduct are stored as a single file. This is different in livestreaming content on Twitch, where the underlying content and streamer voiceover and conduct during broadcast are two different components of a livestream experience.

Knowing just the underlying content of the stream, such as a game, is only one element. So, it is important that both parts of the livestreaming content experience are assessed individually, consistently, and correctly.

When going live, creators provide information in a dashboard, which Twitch uses to automatically add content signals. For games, our database contains global titles and Entertainment Software Rating Board (ESRB) game ratings. Game ratings are signals that give us insight into underlying content of the stream. For non-gaming "lifestyle" content, creators select the relevant Twitch section, such as Just Chatting, Art, or Music.

Creators are required to indicate in their dashboard when they are about to stream content that is sensitive, based on Twitch’s policies. These labels are called Content Classification Labels (CCLs), and they are mandatory and specific to content that is permitted but sensitive. While such content is not prevalent on Twitch, it is important that it is clearly identified and labeled.

The application of the mandatory labels is monitored by Twitch through proactive moderation, AI-based detection, and user-reporting.

The signals from these mandatory labels feed into tools that allow viewers to opt out of certain content, and also inform Twitch's brand suitability tools for advertisers

For contextual brand suitability we offer first-party (1P) tools to help you control where your ads appear:

1. The first is Inventory Tiers, to help exclude sensitive and mature content based on advertiser’s suitability preferences. Customers can choose from three tiers of inventory—Expanded, Standard, and Restrictive—with each tier excluding progressively more content.
2. The second is Contextual Collections, to target or exclude stream content type, such as gaming or lifestyle, or hone in on a specific area of Twitch like Family Friendly games or Sports.

These two tiers of solutions work together to help brands achieve a desired level of suitability and balance suitability with reach that is right for them. Our tools were created to be effective specifically on Twitch, but they were built to the industry standard to give you clarity, consistency, and peace of mind.

We offer full suitability transparency through content-level reporting, which provides impression-level detail of the content your ads appeared in. The report is available on demand in the Amazon DSP, and it is part of the standard reporting offering available across Amazon O&Os to ensure full transparency.

We partner with industry organizations, experts, and leading technology partners to inform and evolve our approach to safety and suitability.

DoubleVerify (DV) and Integral Ad Science (IAS) are approved on Twitch to measure Invalid Traffic, Viewability, and domain/app-level brand safety.

Twitch is part of the Verified by TAG program and holds the Brand Safety Certified seal.