Brand safety in programmatic advertising is the set of controls that keeps automated ad buys away from harmful content, fake traffic, and misrepresented inventory, in every channel your ads run in.
The standard brand safety playbook (blocklists, keyword filters, verification tags) was built for web pages. But more and more programmatic budget now runs on connected TV, digital billboards, podcasts, and in-store screens. There, “unsafe” can mean a fake app, a screen that isn’t where the seller says it is, or a podcast episode nobody checked.
In this guide, we explain what brand safety means in each channel, what it costs when it fails, and the questions to ask any partner before you spend.
Key takeaways
- For years, brand safety was treated as a reputation issue. But today it is a major lever on what you actually pay for media.
- Each channel needs its own safety check: page context for web, app and device checks for CTV, screen and venue checks for DOOH, and show-level context for audio.
- Measure success in brand safety rate, viewability, invalid traffic, and cost per qualified impression, not in how many impressions you blocked.
What Brand Safety Means in Programmatic Advertising
Brand safety protects a brand from appearing in places that could harm it. In programmatic advertising, that protection has to happen automatically, before the bid, because no person reviews each placement.
A programmatic bid is decided in milliseconds. The impression may pass through several exchanges and resellers before it reaches your demand-side platform. Each step is a chance for inventory to be mislabeled, padded with bots, or resold without your knowledge. That is why brand safety cannot be a monthly report. It has to be built into the buying logic itself.
The three layers of brand safety risk are:
- Content risk: what your ad appears next to (the page, video, show, or episode).
- Supply risk: whether the inventory is real and honestly described (a real person, a real app, a real screen in the stated location).
- Creative risk: whether the ad itself is safe, which matters more now that AI can write scripts, voices, and images on the fly.
Who Sets the Rules: Brand Safety Standards After GARM
There is no longer one shared rulebook for brand suitability. That puts the job of defining your own standards back on you. GARM is gone, but its framework lives on
The Global Alliance for Responsible Media (GARM) was set up by the World Federation of Advertisers in 2019. It created the Brand Safety Floor and Suitability Framework that many tools still use. In August 2024, the WFA shut GARM down days after X filed an antitrust lawsuit against it, as reported by Legal Dive.
GARM’s content categories and risk tiers still appear inside platform and vendor settings. But nobody maintains them as a shared standard anymore. Each platform and vendor now updates its own version.
The standards that still apply
| Standard | What it Covers | Why it Matters |
|---|---|---|
| TAG certifications | Fraud, malvertising, brand safety, and transparency | Industry-recognized proof that a company follows anti-fraud and safety rules |
| ads.txt / app-ads.txt | Who is allowed to sell a publisher’s inventory | Stops unauthorized resellers and domain spoofing |
| sellers.json | Who each seller in the chain really is | Makes the supply path traceable |
| MRC accreditation | Whether measurement and verification tools are audited | Tells you if a vendor’s brand safety numbers can be trusted |
| ANA Programmatic Transparency Benchmark | Cost and quality across the supply chain | Lets you compare your waste with your peers |
| OAAA Programmatic OOH Transparency Pledge | Honesty in programmatic out-of-home | Bans hidden fees, inventory misrepresentation and auction tricks |
Where Perion Stands in Programmatic Brand Safety
Perion is TAG certified, one of the leading companies worldwide recognized for meeting the industry’s global standard for safety and transparency in digital advertising (read more).
Perion was also one of the platforms that launched the OAAA Programmatic OOH Transparency Pledge. The pledge commits signatories to honest inventory data, fair auctions, clear fees, and transaction-level reporting.
For example, for France’s ecological transition agency ADEME, Perion built quality into the buying itself rather than checking it afterwards. Perion’s custom optimization algorithm used Integral Ad Science brand safety and quality data to adjust bids in real time and filter out unnecessary middlemen in the supply path.
Across 8.15 million impressions, the campaign reached 96% brand safety and 88.5% IAS viewability. By cutting wasted hops, it also lowered carbon emissions per 1,000 visible impressions by 66%. Safer, more efficient, and lower-carbon came from the same decision: buy better supply.
Learn more about why media sustainability is becoming a performance standard.
How Brand Safety Works in a Programmatic Buy
Brand safety in a programmatic buy works in two stages. Before the bid, your policy, supply checks, and pre-bid filters decide which impressions are safe to buy. After the bid, independent verification confirms where the ad really ran and feeds that back into the next bids, so every purchase gets safer and less wasteful than the last.
Before the Bid: Filters and Suitability
Pre-bid targeting checks each impression before you pay for it. A good setup combines several signals:
- Context: what the page, video, or show is actually about, not just which words it contains.
- Invalid traffic: bots, hidden ads, and other fake impressions.
- MFA and low-quality signals: ad density, traffic sources, and real engagement.
- Your suitability tiers: the rules from your own written policy.
The Supply Path: Fewer, Better Sources
The ANA found that top-performing advertisers bought from more concentrated supply. Every extra reseller in the chain is another place for fees, fraud, or mislabeled inventory to hide.
Practical steps include checking ads.txt and sellers.json, using private marketplace deals and curated packages for sensitive campaigns, and choosing partners that show you the full path.
See how contextual AI is changing audience targeting.
After the Bid: Verification that Feeds Back
Ad verification from an independent, MRC-accredited vendor confirms where your ad really ran, whether it was viewable, and whether a real person saw it.
The mistake is treating verification as a monthly report. The value comes when the results flow back into bidding, so the next impression is bought with what the last one taught you. That is the approach Perion took with ADEME, where IAS data shaped bids in real time.
Why Blocklists Are Not the Right Strategy
A blocklist is a useful safety net, but as your main defense, it does more harm than good. Keyword blocking fails because:
- It can’t read context. A report on a flood and a page glorifying violence may share the same words.
- It punishes good journalism. News sites are often blocked for covering hard topics responsibly, which pushes budget toward cheaper, riskier sites.
- It misses the real threats. MFA and AI-generated sites can be written to avoid risky words, so they pass straight through.
- It goes stale. New sites, apps, and slang appear every week.
What to use instead:
| Approach | Best for | Strengths | Weaknesses |
|---|---|---|---|
| Blocklist | A safety net under every campaign | Fast, simple, blocks known bad actors | Context-blind, over-blocks, ages quickly |
| Inclusion list (allowlist) | Highly sensitive or regulated campaigns | Maximum control | Limits scale, can raise CPMs |
| Contextual suitability | Most always-on campaigns | Reads meaning and tone, keeps safe news in reach | Needs good tools and regular tuning |
| Curated supply deals | Brands that want quality without full direct buys | Vetted sellers and shorter paths | Depends on the curator’s standards |
How Brand Safety Works by Channel
The same question, “Is this a safe place for my ad?” has a different answer in every channel. On a website, you check the page. On a connected TV, you check the app and the device. On a billboard, you check that the screen exists and played your ad.
| Channel | What “Unsafe” Looks Like | What to Check | Main Control | Perion Example |
|---|---|---|---|---|
| Web and in-app | Harmful pages, MFA and AI-generated sites, spoofed domains | Page context, ads.txt and app-ads.txt, invalid traffic | Contextual pre-bid filters, curated supply, verification | ADEME: 96% brand safety with IAS data inside the bid |
| Connected TV (CTV) | Fake apps, spoofed devices, unknown shows | App bundle ID, device and household signals, show-level data | Verified publishers, transparent reporting | Webroot: real-time reporting, 89% lower cost per action |
| Digital out-of-home (DOOH) | Screens misdescribed by location or venue, ads that never played, hidden fees | Screen ID, venue type, proof of play | Transparency pledge principles, curated deals | OAAA pledge launch partner; Green Curated Deals |
| Audio and podcasts | An ad inside a controversial episode, vague show lists | Show and episode data, genre | Show-level inclusion, genre matching | WAVE matches ads to podcast or playlist genre |
| In-store retail media | Screens that don’t play as reported, poor placement | Store, aisle and screen data | Retailer-run networks, first-party data | Best Buy Canada: Perion powers its retail DOOH network |
| Social and walled gardens | Placements controlled by the platform | Platform suitability settings, third-party checks | Platform tools plus independent optimization | Outmax AI agent on Meta, TikTok, and YouTube |
How to Measure Brand Safety
Measure brand safety by what reached real people in suitable places, and what that cost, not by how many impressions you blocked.
- Brand safety rate – Share of impressions in safe and suitable environments
- Viewability – Share of impressions that could actually be seen
- Invalid traffic (IVT) rate – The percentage of ad requests or impressions generated by non-human, fraudulent, or malicious activity (bots, crawlers, automated scripts, domain spoofing, or hidden/stacked ad units).
- MFA share – Share of spend on made-for-advertising sites
- Supply path length – Number of resellers between you and the publisher
- Over-blocking cost – Safe reach lost to blocklists, and its effect on CPM
- TrueCPM (ANA Benchmark term) – Effective cost per 1,000 impressions that were measurable, viewable, fraud-free, and non-MFA
The Bottom Line
Brand safety in programmatic advertising is no longer just about avoiding a bad headline. It decides how much of your budget reaches real people in the right places, and that shows up directly in what you pay.
Start with three steps: write your own suitability policy, choose the safety checks that fit each channel you buy, and report results as cost per qualified impression.
If you want to see how Perion One builds quality into every campaign across CTV, DOOH, audio, in-store retail media, web, and social, request a demo.
FAQs
What is the difference between brand safety and brand suitability?
Brand safety is the floor no advertiser should go below, such as hate speech, terrorism, or fraud. Brand suitability is the layer each brand sets for itself, based on its values, audience, and tone. A news story about crime can be safe for one brand but unsuitable for another, such as a baby-care brand.
Does brand safety reduce reach?
It can, if it relies on broad keyword blocklists that remove safe news and premium content. Context-based suitability, curated supply, and regular reviews of what you block protect the brand while keeping more quality reach. Done well, brand safety lowers your cost per qualified impression instead of raising it.
What is the difference between pre-bid and post-bid brand safety?
Pre-bid brand safety checks each impression before you buy it, using context, fraud, and suitability filters. Post-bid brand safety verifies where the ad actually ran after it was served. The best setups use both and feed post-bid results back into bidding, so each new purchase is safer than the last.
How do you measure brand safety?
Track brand safety rate, viewability, invalid traffic rate, MFA share, and supply path length. Then combine them into one cost figure: cost per thousand qualified impressions, or TrueCPM, the ANA Benchmark’s version of this measure. It shows what you paid for impressions that were measurable, viewable, fraud-free, and non-MFA, which is the number finance teams care about.
