How do you run brand safety monitoring on a live campaign?
Part of Brand suitability
The short answer
Brand safety monitoring is the work of checking a live campaign for suitability drift: approved articles that change, new sections on approved sites, and settings that stop covering delivery. A team sets a check cadence by risk, samples placements from delivery reports with a written method, acts on named pause triggers, and gives one person ownership of each check.
Key takeaways
- An approved shortlist is a dated snapshot; the pages behind it keep changing.
- No standard sets one monitoring frequency, so cadence should follow risk: launch, breaking news and settings changes call for extra checks.
- A hand drawn placement sample is judgmental, and the MRC says such results cannot be generalized across a campaign.
- Pause triggers work when written before launch, with the evidence each needs and the person allowed to act.
What is suitability drift in a live campaign?
Definition
- Suitability drift
- The gap between the content a brand approved before launch and the content its ads appear beside once the campaign runs, caused by edited pages, new site sections, aging labels or changed settings.
Planning approves pages as they looked on the day someone read them; delivery happens later, on pages that keep living. The MRC's 2018 supplement on content level brand safety (opens in a new tab) says fresh data "is absolutely necessary for accurate ad verification" and that content "such as news content" may be dynamic at first, then more static.
| Source of drift | What changes | Where it shows up | First check |
|---|---|---|---|
| Edited article | A story approved as a feature gains new reporting, images or a headline | The same URL keeps delivering | Reread the page against its review date |
| New site section | An approved domain adds a forum, sponsored hub or subdomain | New paths in the placement report | Decide whether the domain approval covers it |
| Aging classification | A label reflects an older version of the site | Labels assigned weeks earlier | Check the platform's stated update cycle |
| Settings change | A deal is exempted from suitability settings, or a line item is copied without them | An unreported entry in event history | Compare active settings with approved ones |
| Delivery off the list | Ads reach placements removed from targeting but never excluded | Unfamiliar domains in delivery | Confirm the placement was excluded |
Google's help page on excluding webpages and videos (opens in a new tab) explains the last row: "Excluding placements is different from removing them," because other targeting, such as keywords or topics, can still bring an ad to a placement that was only removed.
What happens when an article changes after the shortlist is approved?
The approval stays attached to the URL while the content under it moves. A local story about a school budget can gain a paragraph about an assault at a school, and nothing in the buying platform records that the approved version is gone.
Classification carries its own delay. Display & Video 360's help page on digital content labels (opens in a new tab) says "Our brand safety classification is updated monthly to reflect any changes and updates to a site," so a label on a page edited this week can describe last month's version. The MRC supplement recommends "a time stamp" or clear disclosure of data life, so users know the age of the latest assessment. Ask vendors for that date, and keep your own.
Five steps when an approved article has changed
- Capture the page as it is now. Save a dated screenshot, the full URL and the ad that appeared.
- Compare it with the review record. Find when the page or section was approved and what the reviewer saw.
- Judge the new version against the brand's tolerance. Use the category and tier language of the original brand suitability approval.
- Choose the narrowest fix. In Google Ads a page URL excludes only that page, a directory path that directory, and a domain the domain with its subdomains.
- Log the outcome. Record the decision, date and decider, so the next reviewer knows the page was seen twice.
How often should a live campaign be checked?
No industry standard gives one monitoring frequency for live placements. The MRC asks verification services to state their own refresh policies, based on empirical support that is periodically analyzed. TAG's Brand Safety Certified Guidelines (opens in a new tab), version 5.0 of July 2026, require certified buyers and intermediaries to review inclusion and exclusion lists quarterly: a governance interval for lists, not a schedule for watching delivery.
A practical cadence follows risk. Launch days show whether targeting behaves, a breaking story changes what approved news sites publish within hours, and a settings change can open inventory the client never saw. The template below is a starting point to agree with the client, not a standard.
| Moment | What to check | Starting interval |
|---|---|---|
| Launch | Delivery stays on the list; active settings match the approved ones | Daily for the first several days |
| Steady flight | A fresh placement sample; new paths on approved domains | Weekly |
| Breaking news in the brand's categories | Approved news sections and pages near the story | Same day, then daily until coverage settles |
| After a settings or deal change | Event history and active settings | Within one working day |
| List governance | Whole list and the reason for each exclusion | Quarterly, matching TAG's list review |
How do you sample placements for review?
Monitoring runs on a sample. The IAB and MRC Guidelines for the Conduct of Ad Verification (opens in a new tab) of February 2012 separate a statistical sample, reported with size, period and confidence limits, from a judgmental one, reported with "Selection method, sample size, time-period sampled out of total campaign run". Judgmental observations "cannot be generalized across a campaign". A team reading pages by hand runs a judgmental sample and should say so.
The frame has gaps too. Google says its Where ads showed report (opens in a new tab) should not be used as "an exhaustive list of placements", that placements shown only a couple of times may be missing, and that it "is not regularly updated and may include invalid traffic." Low volume pages can sit outside the frame entirely.
A weekly placement sample in six steps
- Fix the window. Pull the report for the same span of days and campaigns each week, and note the pull date.
- Take the top rows. Read the placements with the largest delivery since the last check.
- Take everything new. Read each domain, path or channel absent from last week's report.
- Take the unreviewed sections. Read approved sections with no review date in the log.
- Add a random draw. Pick a set number of remaining rows at random, so the sample is not only the obvious places.
- Record the method. Note how rows were chosen, how many were read and the period covered.
Which changes should trigger a pause?
A pause trigger is a condition written before launch that lets the person on duty act without a meeting. It names the evidence required and the narrowest action, since pausing a line item to fix one page costs delivery on the good pages around it. In Display & Video 360, Google's brand suitability guide (opens in a new tab) describes an option to "bypass all advertiser-level brand suitability settings for specific deals"; it keeps Google's automatic protections and does not apply to YouTube or Demand Gen. A bypass the client never approved deserves its own trigger.
| Trigger | Evidence needed | Narrowest action | Who may act |
|---|---|---|---|
| Ad beside floor level content | Screenshot, URL, time and ad | Exclude the page now, or pause the placement | Anyone on the rota, reported the same day |
| Approved article edited above tolerance | Current capture and the review record | Exclude the page URL | Monitoring owner |
| New user generated section on an approved domain | The new path and sample posts | Exclude the path until reviewed | Monitoring owner |
| Deal exempted from settings without sign off | Event history entry | Restore settings or pause the line item | Campaign owner, client informed |
| Delivery on domains off the list | Placement rows with domain, deal and date | Pause the line item while tracing | Campaign owner |
Floor content should not wait for a second opinion. The other triggers need a reviewer, since an edited article that mentions a crime may still be responsible reporting. Full incidents follow the brand safety incident response plan.
Who owns monitoring during a campaign?
Ownership gaps are easy to miss. Section 4.10 of the TAG guidelines requires certified companies to define the roles responsible for responding to incidents of ad misplacement, document the external resources they rely on, and show "clear lines of communication across partners." Any team can borrow the structure: one named owner per campaign, a deputy, and a contact at each partner.
Display & Video 360's brand controls page (opens in a new tab) shows "active settings across all of your line items and YouTube ad groups" with a download of recent label and sensitive category changes, and its event history (opens in a new tab) records the time, date, description and user of each change, with an option to include system changes.
Monitoring ownership, agreed before launch
- A named monitoring owner and deputy cover the whole flight.
- Pause triggers, their evidence and who may act are written in the plan.
- The client contact for same day floor level reports is named.
- Each partner, including any verification service, has an incident contact.
- Active settings are compared with the approved settings after each change.
- The log records each check with date, reviewer, sample method and outcome.
How does PlacementGraph show what changed in a plan?
What does monitoring look like on a real campaign?
Fictional example
A child care chain watches its fall enrollment campaign in Iowa and Nebraska
An unnamed chain of early learning and child care centers runs a fall enrollment campaign in Des Moines, Omaha and Lincoln on an approved list of 40 parenting, family and local news sites. The client set a low tolerance for crime and harm involving children and a moderate one for general local news.
In week two the sample includes a local news article approved as a story about preschool funding. It now carries a report on an injury at an unrelated day care. The owner captures the page, checks the review record and excludes that one URL; the site stays, because its other sampled pages still fit.
In week four a parenting site shows an unreviewed path, a new community forum, and the owner excludes it pending review. Event history shows a teammate exempted one deal from suitability settings to fix slow delivery. The owner restores the settings, tells the client the same day and notes the reason on the change.
Frequently asked questions
Does monitoring a domain tell you that its pages are safe?
Not on its own. The MRC's October 18, 2025 policy separates property level ad verification, based on text, keywords and language, from content level brand safety that also covers images, video and audio. A domain level result describes the site, not the page an ad appeared on.
Should a page excluded during the flight ever come back?
It can. If the publisher corrects or removes the material, a reviewer can reread the page, record a new review date and lift the exclusion. Keeping the original capture and reason in the log lets anyone check the restore decision later.
What should monitoring do with placements that have no label?
Treat them as unknown, not safe. Google's Display & Video 360 help lists new sites, masked URLs and unsupported languages as reasons content goes unlabeled, and the MRC supplement says buyers should be able to prevent serving to unclassified content. Sample unlabeled rows on purpose.
Is a verification report enough to monitor a campaign?
It is one input. The 2012 IAB and MRC guidelines ask verification services to disclose their methods, sample sizes and limits, and to identify judgmental analysis as such. Read those disclosures before treating a verification rate as the full picture, and keep a separate log of what the team reviewed.
Sources
- MRC Supplement to IAB Guidelines for the Conduct of Ad Verification: Enhanced Content Level Context and Brand Safety (opens in a new tab)Media Rating Council, September 2018
- Guidelines for the Conduct of Ad Verification (opens in a new tab)IAB and Media Rating Council, February 14, 2012
- Policy for Property-Level Ad Verification Representations (opens in a new tab)Media Rating Council, October 18, 2025
- Brand Safety Certified Guidelines, version 5.0 (opens in a new tab)TAG (Trustworthy Accountability Group), July 2026
- Digital content labels in Display & Video 360 (opens in a new tab)Google Display & Video 360 Help
- About the Where ads showed report (opens in a new tab)Google Ads Help
- Exclude specific webpages and videos (opens in a new tab)Google Ads Help
- Brand suitability (opens in a new tab)Google Display & Video 360 Help
- View and edit brand controls for your advertiser (opens in a new tab)Google Display & Video 360 Help
- Event history (opens in a new tab)Google Display & Video 360 Help
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.