Share on LinkedIn
Share on XIAS didn’t improve keyword avoidance. We made it obsolete.
Keyword-based brand safety is a relic. While others still rely on primitive keyword or URL scans, we’ve built a more sophisticated approach — one grounded in AI-powered, full-page contextual analysis. At IAS, keywords now account for just 0.25% of blocked or flagged impressions on buy-side open web traffic, based on IAS post-bid measurement from the past 30 days.
That’s not just low — it’s proof that AI-powered context is here to stay at IAS.
Keywords don’t get context. That’s the problem.
- Overblocking: Keywords can’t understand nuance. “Shoot” could mean a camera or a crime scene. “Apple” could refer to a tech product or a healthy snack. Without context, you end up blocking premium, brand-safe inventory.
- Underblocking: Keywords often miss real risks buried in the content. If it’s not in the URL string, it’s invisible to legacy systems.
IAS doesn’t just “check for words.” We deeply understand context.
Core capabilities powering IAS Context Control
Our AI-powered classification system analyzes full-page content, not just snippets or metadata, and processes approximately 15 billion unique URLs every 24 hours, using the following capabilities:
- AI: IAS utilizes AI capabilities. Today, our natural language models interpret true meaning, sentiment, and intent — not just terms. We continue to invest to deliver even deeper content intelligence.
- Proprietary Knowledge Graph: Understands word relationships, industry nuance, and context across categories, tuned specifically for advertising.
- Semantic Analysis: Distinguishes “Jaguar” the animal from the car brand in milliseconds.
- 90+ Emotional & Sentiment Signals: Pinpoints tone — love, fear, outrage — to assess content suitability more precisely.
- Entity Recognition: Detects people, places, and brands to match advertiser relevance.
- 52+ Languages: Delivers global coverage with local nuance.
Built for precision. Powered by scale.
Not keywords. Not sampling. Full-page intelligence.
IAS doesn’t guess. We analyze every page at the moment of the first ad request — no shortcuts, no sampling, no extrapolation. Since launching Context Control in 2020, we’ve stuck to a simple principle: precision beats approximation.
That means better outcomes for advertisers and fairer representation for publishers, who benefit from fewer false blocks and more accurate classification of quality content.
Here’s how we deliver real contextual intelligence, at global scale:
- Daily Score Refreshes: Every page we classify gets re-scored daily. That adds up to billions of refreshed context signals per week across the open web, ensuring your ad decisions reflect current content — not stale scans.
- Dynamic Crawling: Pages that change more often (like news homepages) are crawled more frequently. Static pages? Less so. Our system adapts automatically.
- Instant Decisioning: Advertising customers can choose to avoid unscored pages entirely. When a new page is encountered, it’s labeled “unscored” for just a few seconds before it’s fully classified and distributed across our systems and partners. No guessing. No delay.
Context Control is built for advertisers who want scale and control. We don’t cut corners. We classify everything.
Custom protection that fits your brand. At scale.
Advertisers don’t want black-box brand safety. They want clarity, control, and actionability. That’s why IAS built Context Control to deliver on the things marketers care about most: transparency and flexibility.
Here’s how we deliver:
- Full Control with Infinite Flexibility
Build your own brand suitability strategy with self-service tools that support everything from broad protection to hyper-specific needs. Choose from 280+ pre-built contextual segments — or create custom segments that reflect your brand’s unique risk tolerance, tone preferences, and industry nuances. - Granular Precision, Instantly Deployable
Move beyond “high, medium, low” with fine-tuned subsets of industry-standard risk levels, enhanced by content-type modifiers (e.g., distinguishing between video game violence vs. breaking news coverage). Launch new protection strategies in minutes, not days — because media moves fast. - Transparent Classification You Can See and Act On
Every customer gets URL-level reporting by default, so you know exactly which content is being flagged and why. No hidden logic or guesswork — just actionable visibility into the decisions our tech is making.
We go further with true URL detection, powered by our proprietary client-side script and fraud detection systems. That means we verify where an ad actually ran — not where it claims to. And when verification isn’t possible, IAS gives you proactive tools to avoid unclassified or invisible inventory altogether.
Contextual intelligence that goes beyond the open web
Context doesn’t stop at the article — and neither do we. IAS has expanded its contextual classification to social and video platforms, unlocking protection and performance across the content that dominates today’s media plans.
Our multimedia AI analyzes text, image, and audio at the frame level — delivering deep classification across social and video platforms like Meta, TikTok, and more. Whether it’s a trending video or a fast-moving social feed, IAS provides contextual understanding for multi-media.
Together, these capabilities provide advertisers with unified brand suitability definitions across both open web and walled gardens.
Innovating Further: Multi-Modal Models
IAS is pushing beyond today’s leading contextual capabilities with our patent-pending Multi-Modal Models, designed to analyze text, image, and video together, just like a human would. This unified, cross-format approach enables us to:
- Understand content contextually, regardless of medium — enabling a far deeper and more nuanced understanding of context.
- See how a brand is represented across all forms of media: which other brands it appears with, what types of content it’s featured in, and who’s talking about it.
- Surface emerging concepts, even if they’ve never been predefined.
- Build dynamic, on-demand contextual categories tailored to an advertiser’s needs.
IAS is creating a smarter, more adaptive way to understand and target digital content, pushing contextual classification into the next era of advertising.
We’ll dive deeper into our innovations in social and video classification in a future article, but know this: IAS is redefining suitability across every digital channel where ads run.
Protection meets performance
Brand safety shouldn’t come at the expense of scale — and with IAS, it doesn’t.
Our approach is data-driven, flexible, and built for results. By combining precise analysis with transparent control, we help advertisers protect their brand and hit their performance goals.
It’s not either-or. With IAS, it’s both.
