Brand safety after the YouTube adpocalypse: a practical plan

Programmatic and YouTube buying works by targeting people, not pages. Your ad follows an audience onto whatever video or site they are watching, and in an open auction nobody at your company sees the placement before it runs. In February, UK press reports documented ads from government agencies, banks and consumer brands running on videos from hate preachers and extremist groups, with the uploaders earning a share of the revenue. The story crossed to the US in March, and the pullout followed within days. Analysts estimated the short-term cost to Google in the hundreds of millions of dollars, and procurement teams at large advertisers started asking agencies the questions they should have asked years earlier.

The underlying problem is not new. Open-exchange display has carried the same risk for years, and fraud and viewability complaints from Procter & Gamble's chief brand officer in January had already put the whole supply chain on notice. The YouTube episode simply made it visible.

Google published a response on March 21 and has layered on changes since. The ones that matter to an advertiser:

• Tougher content policies on what is eligible for monetization, with more aggressive demonetization of borderline channels.
• Safer defaults. Campaigns now exclude more potentially objectionable content unless you opt in, and the sensitive-category settings in AdWords were expanded.
• Easier account-level exclusions so a list of excluded channels and sites can apply across campaigns.
• Third-party verification access, with Google saying it will work with vendors like Integral Ad Science, DoubleVerify and comScore to measure YouTube brand safety independently.
• A higher bar for the YouTube Partner Program. Since April a channel needs 10,000 lifetime views before it can run ads at all, which removes the long tail of throwaway uploads.

These help, but none of them is a substitute for deciding what safe means for your brand. A gun retailer and a children's hospital have different lists, and the platform default is neither.

Spend an hour and produce a one-page brand safety policy before touching a campaign. It should answer four questions.

1. Which categories are always excluded? Start from the platform sensitive categories: tragedy and conflict, sensitive social issues, profanity, sexually suggestive content, sensational and shocking content. Add industry-specific ones.
2. Which content types are excluded? Live streaming, embedded video on third-party sites, games, apps for children, parked domains, content not yet labeled.
3. Which sites and channels are excluded by name? Maintain a list and date each addition.
4. Where is an allowlist required? For brand campaigns and anything with sensitive creative, consider buying only from a curated list of channels or publishers, accepting the lower reach.

Have the policy signed off by whoever would have to answer the journalist's email. That is usually the owner or the head of marketing, not the media buyer.

On YouTube specifically, review the placement report weekly for the first two months. Sort by impressions, scan channel names, and exclude anything you would not want in a screenshot. For brand campaigns, build a channel allowlist of a few hundred channels in your category and accept that reach drops; the cost per view usually rises 20 to 50 percent, which is the price of knowing where you ran.

On open-exchange display, prefer private marketplace deals with named publishers for the bulk of spend, keep open auction to a tested exclusion list, and apply pre-bid blocking from your verification vendor. Also watch the new IAB ads.txt proposal, which lets publishers list their authorized sellers; it is early, but it is the first real tool against spoofed inventory.

On Facebook, the Audience Network is where the same issue lives. Review the publisher list it now provides, apply block lists, and keep placement reporting on.

Independent measurement is now table stakes for any brand with a reputation to lose. Integral Ad Science, DoubleVerify and Moat all measure brand safety, viewability and invalid traffic, and most will block before the bid on display. Pricing is usually a CPM on measured impressions, which for most mid-sized budgets works out to a low single-digit percentage of media. Measurement on YouTube is still being connected as Google opens access; ask your vendor what they can see today and what is on their roadmap, and do not accept a vague answer.

Set thresholds, not just reports: a brand-safety incident rate above 1 percent of impressions, or viewability below 60 percent on a placement, triggers exclusion. Reports nobody acts on are the way this happens again.

The adpocalypse is a trust problem as much as a tooling one. Three conversations are worth having this quarter.

With your agency or buyer. Ask to see the placement reports, the exclusion lists and the verification contract. If any of those does not exist, that is your answer.

With your own team. Agree what happens when an incident appears: who pauses, who excludes, who responds publicly if asked.

With your customers, if it matters. A short statement of what you will not fund is a brand position some audiences notice. Most businesses do not need to say anything; they need to not end up in the screenshot.

We treat brand safety as part of how a brand is built and how media is bought, not a checkbox at the end.

Discover. Audit the last 90 days of placements across YouTube, display and social, map the exclusions and verification currently in place, and identify what the brand would be unwilling to appear beside.

Plan. A one-page policy, a category and placement exclusion set per channel, an allowlist strategy for brand campaigns, verification vendor selection and incident response steps.

Build. Configure shared exclusion lists, campaign settings, private marketplace deals and verification tags; set up weekly placement reporting with thresholds.

Improve. Monthly placement review, quarterly policy review, and updates as the platforms continue to change controls through the year.

A representative engagement is our performance marketing work, where a brand with a growing video and display budget moved from default settings to a written policy, channel allowlists for its brand campaigns, pre-bid blocking on display and a monthly placement review. Reach fell modestly at first; the quality of the inventory, and the owner's confidence in answering questions about it, rose considerably.