Guardrails: brand safety, exclusions and policy compliance

0:00 Guardrails

Automation is brilliant at finding conversions, and completely indifferent to where it finds them. It'll put your ad next to content you'd never approve, match you to search terms that embarrass you, and target people you're legally not allowed to target, if nothing stops it. In this lecture you'll learn the full guardrail inventory, the difference between brand safety and brand suitability, the regional rules you need to know, and a script that catches runaway spend before it becomes a disaster.

0:35 Why it matters

Why does this matter? Because one bad placement or one embarrassing search term can cost more in reputation than a month of efficient ads earns. Here's an analogy. Automated campaigns are like a very fast delivery driver who's never been to your city. They'll find the quickest route, but unless you mark the roads that are closed, the neighborhoods you don't serve, and the speed limits, they'll drive straight through them. Guardrails are those road closures and speed limits. You set them once, check them regularly, and let the driver move fast inside them.

1:16 Inventory, part one

Let's walk the inventory. Negative keywords and negative keyword lists stop irrelevant or harmful queries, and on Google they now apply to Search, AI Max and Performance Max. Brand exclusions stop automated campaigns cannibalizing your own brand. URL exclusions stop ads landing on your careers page or blog posts that don't sell. Content suitability settings and inventory filters keep you away from unsuitable content. Placement exclusions remove low-quality apps and sites.

1:47 Inventory, part two

Part two. Customer lists and new-customer goals stop you paying prospecting prices for existing customers. Special ad categories on Meta, for housing, employment, credit and social issues, restrict targeting by law and policy. Sector certifications, like financial services verification, are required in some markets. And budget caps, alerts and automated rules stop runaway spend. Each guardrail has a scope, account or campaign, so check where it applies.

2:16 Safety vs suitability

Now an important distinction. Brand safety is the floor, avoiding content that's clearly harmful to any brand. Brand suitability is your own tolerance above that floor. A gaming brand might be fine next to mature game content. A children's education brand must avoid it. So write your suitability tiers down: always avoid, avoid for this campaign, acceptable. Then map each tier to each platform's settings. Independent verification partners can provide reporting on some platforms.

2:48 Simple example: UK kids' education app (illustrative)

Here's a simple worked example. A children's education app in the UK launches Demand Gen and Performance Max. In week one, the placement report shows impressions on mobile games aimed at adults and on channels with mature content. The team sets account-level content suitability to exclude mature and sensitive content, adds an exclusion list of app categories, and checks the brand safety tier matches their written suitability policy. They also add negative keywords like free hack and cheat. In week two the placements look clean, and cost per sign-up barely changes, because those placements weren't converting anyway.

3:30 Regional rules (check current)

Regional rules matter, and they change, so always check current guidance. In the UK, the CAP Code enforced by the ASA restricts targeting for age-restricted products. In the EU, the Digital Services Act bans ads targeted using special-category data like health or religion, and profiling-based ads to minors. In the US, Meta's special ad categories limit targeting for housing, employment and credit. In the UAE, the Media Council regulates advertising content and licensing. In Saudi Arabia, GCAM. In Pakistan, PEMRA and sector regulators like DRAP for health products.

4:08 Example: Karachi fintech (illustrative)

Here's an illustrative example. A Karachi digital wallet launches on Google and Meta. They complete financial services verification where required. They build a negative list covering scam, hack, job-seeking and complaint terms. They exclude sensitive content categories and low-quality app placements. Every ad is checked against a claims register: no guaranteed returns, fees stated clearly. And an automated rule pauses any campaign spending more than one and a half times its daily plan, and alerts the team.

4:41 Automated overspend alerts

The lesson includes a Google Ads Script that emails you when any enabled campaign spends more than a threshold multiple of its daily budget today. One caution: Google can legitimately spend above the average daily budget on some days within its monthly limits, so tune the threshold to avoid false alarms. The same idea works with automated rules on Meta and TikTok.

5:08 AI for guardrails

AI can make guardrails stronger. Use an LLM to cluster search-term and placement reports and propose exclusions. Screen ad copy against your claims register before submission. Summarize policy changes weekly. But always keep a human approving changes that affect delivery. And watch the opposite failure: excluding so aggressively that the automation starves and can't find customers at all.

5:33 Mistakes + try this now

Common guardrail mistakes. Assuming the platform handles brand safety by default. Applying exclusions at campaign level and forgetting new campaigns don't inherit them. Over-excluding until the campaign can't spend. Ignoring special ad categories for housing, employment and credit. And never reviewing placement reports. Try this now: write your three suitability tiers on one page, always avoid, avoid for some campaigns, and acceptable, then check whether each item is actually set in your ad accounts, and at which level.

6:07 Watch me do it: school admissions guardrails (illustrative)

Watch me do it. Let's set up guardrails for an illustrative Abu Dhabi private school advertising admissions. Step one, the suitability policy on one page: always avoid mature content, gambling, violence and political controversy; avoid for this campaign, gaming content; acceptable, family, education and news. Step two, Google: account-level content suitability set to exclude the relevant sensitive categories, an account-level placement exclusion list for low-quality apps, and a negative keyword list with jobs, salary, vacancy and teacher recruitment, because those searchers want to work there, not enroll. Step three, brand: brand exclusions on PMax, with a separate brand Search campaign. Step four, Meta: inventory filter set to the more limited option, and a block list of publishers. Step five, alerts: an automated rule to pause any campaign spending more than one and a half times its daily plan. Step six, the first review after one week: I open the placement report and find a children's cartoon channel. Fine for suitability, but our audience is parents, so it's not converting; I exclude it for performance, not safety. That distinction matters in reporting.

7:26 Recap and next step

Recap. Guardrails are the hard constraints that keep automation safe. Know the full inventory and each guardrail's scope. Safety is the floor, suitability is yours to define. Know your regional rules. Automate alerts and let AI propose, while humans approve. Your next step: write your brand suitability tiers, always avoid, avoid for this campaign, and acceptable, and map them to settings on two platforms you use.