Brand Safety: How to Control Where Your Ads Actually Appear

Brand safety is the least glamorous part of media buying and the one that produces the most awkward screenshots. Your ad, rendered faithfully, next to content you would never sponsor. Nobody chose it — an auction did — but the screenshot has your logo in it, not the auction's.

The good news: this is a solvable, mechanical problem. This guide covers the levers that actually control placement, how to use them without strangling your own reach, and how to build a list from data instead of anxiety.

What you are actually controlling

An ad network matches your campaign to available inventory. Left unconstrained, "available" means everything that fits your geo and format. Brand safety is the set of rules that narrows "everything" down to "places I am willing to be." On Adgora that comes down to four levers, and each does a different job:

• Site categories — exclude or include whole classes of content.
• Site blacklist — specific domains you never want.
• Site whitelist — only these domains, nothing else.
• Contextual keywords — aim at the topic of the page itself.

Used together they are precise. Used reflexively they quietly kill campaigns. The difference is whether each rule is answering a real question.

Categories: the blunt instrument, used first

Category targeting is the fastest way to remove whole neighbourhoods. Adgora classifies sites across categories like news and media, technology, finance and crypto, games and esports, entertainment, online stores, health, education, lifestyle, sports, travel, auto — and the ones most brands think about first: gambling and betting, and 18+ / adult.

If your brand cannot appear next to gambling content, exclude the category and the question is closed for the whole campaign — you do not need to discover and blacklist a thousand casino domains one at a time. Equally, if your offer is gambling, targeting that category puts you exactly where the audience already is.

Categories are a scalpel for classes, not for individuals. They cannot know that one particular site in an acceptable category is rubbish. That is what blacklists are for.

Blacklists: built from data, not from fear

A blacklist is a list of domains you never want. The temptation is to write one before launching, from imagination. Resist — you will block sites that would have worked and miss the ones that actually waste your money, because you cannot guess a placement report.

The workflow that works:

1. Launch broad with your category exclusions in place.
2. Read the placement report after real volume has accumulated.
3. Cut two kinds of placement: the ones that embarrass you, and the ones that spend without converting. They are usually different lists, and both are worth cutting.
4. Repeat. A blacklist is a living document; it should get better every week the campaign runs.

That second category — high spend, zero conversions — is where brand safety and performance quietly become the same discipline. A placement delivering volume and no revenue is often also a placement you would not want your logo on. Cutting it protects the brand and the budget in one move. It is also where fraud filtering pulls in the same direction: invalid traffic is de-duplicated in the tracking path before billing, so the placements that survive your report are ones with real humans behind them.

Whitelists: earn them, do not assume them

A whitelist restricts you to a named set of domains and nothing else. It is the strongest control available and the easiest to misuse.

Used on day one it is a cap on your own ceiling — you are limiting the campaign to what you already knew before you learned anything, which is the opposite of how discovery works. Used after a month of data, it is a performance tool: these are the fifteen placements that reliably clear my cost per action, so put every dollar there.

The rule of thumb: blacklist early, whitelist late. One removes known harm; the other locks in proven value. Doing them in the wrong order is how campaigns get safe and useless at the same time.

Contextual keywords: the positive control

Everything above is subtractive. Contextual keyword targeting is the additive version — instead of listing where you refuse to be, you describe the pages you want. Show this ad on pages about portfolio tracking; show it on pages about forex.

This is brand safety by presence rather than absence, and it is the natural partner for native and teaser formats, where the page's own topic does half the persuading. It also sidesteps the endless-list problem: you cannot enumerate every bad page on the internet, but you can describe the good one. The full lever set sits in the ad targeting guide.

The over-filtering trap

Here is the failure mode nobody warns about. A nervous buyer excludes eight categories, imports a blacklist of two thousand domains, whitelists forty sites, narrows to one geo and one device — and then wonders why the campaign has no volume, no data, and a high clearing price.

Every rule shrinks your pool. A thinner pool means less signal, slower learning, and often a worse price, because you are bidding on a narrow slice everyone else also wants. Brand safety has a cost, and the goal is to pay it only where it buys something real:

• Exclude what genuinely damages you — the categories your brand truly cannot sit beside.
• Cut what demonstrably wastes money — from the report, not from fear.
• Let the rest run and judge it on cost per action.

What the network owes you

Controls are only as good as the transparency behind them. If you cannot see which placements your ads ran on, you cannot blacklist anything — you are trusting an adjective. That visibility is the real precondition for brand safety, and it is one of the questions worth asking before you spend, alongside the rest of the list in how to choose an ad network.

On the supply side, Adgora moderates sites before they can carry ads and moderates creatives before they can serve, so the pool you are bidding into has been through a review rather than assembled automatically.

Create an Adgora account to set categories, lists, and contextual keywords on your first campaign — or see the advertiser overview for how the controls fit together.

Key takeaways

• Four levers: site categories (classes), blacklists (known harm), whitelists (proven value), contextual keywords (positive targeting).
• Blacklist early, whitelist late — a day-one whitelist caps your ceiling before you have learned anything.
• Build lists from the placement report, not from imagination.
• Cut both the embarrassing placements and the ones that spend without converting — often the same fix.
• Over-filtering starves campaigns: every rule costs volume, signal, and often price.
• Without placement transparency, brand safety controls are decoration.