Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Programmatic Advertising › Advanced

Handling a Brand Safety Incident for a Car Seat Maker

You run an invented car seat maker's brand safety incident from the parent's screenshot to the fixed cause, while the chief executive asks for all programmatic to stop and a colleague wants a broad keyword block. The exercise practises judging the placement against the written policy, tracing it to the line item that bought it, choosing a stop that removes the risk without removing the inventory the brand wants, and fixing the cause where it actually lives.
ExerciseAdvanced levelModule 2: Suitability, inclusion lists and incidents5 minUpdated

Why this matters

At the beginner level you set brand safety rules before a buy ran. Here the rules exist and something got past them, and the first hour is when the broadest controls get added. The tutorial on responding to a brand safety incident sets the order: evidence, records, route, narrow stop, bill, words agreed in advance, cause, fix and record. The guide to writing a brand suitability policy is the yardstick, because a placement is judged against what the brand decided, not the mood of the day. If verification and ad serving are not yet familiar, revisit the intermediate programmatic level.

The car seat maker and the screenshot

Ottlecrag Child Safety is an invented maker of children's car seats, buying in DV360 through an agency with a written suitability policy owned by its marketing director. On Monday morning a parent emails a screenshot of an Ottlecrag banner beside a regional news report of a road accident. The chief executive asks for programmatic to be paused. The brand manager proposes blocking the keywords crash, accident, collision and injured on every line item.

Illustrative example — not a real result

Ottlecrag Child Safety: policy, records and line items

Policy extract: floor on for every buy; hard news allowed on a named list of trusted publishers regardless of topic; tragedy and accidents excluded on every line item; user-generated pages excluded except 3 named parenting forums.

Line itemRouteTragedy exclusion setSpend last 7 days
Parenting publishersPrivate auction dealYes£2,100
Trusted newsOpen auction, inclusion channelYes£1,600
Child and road safety contextOpen auction, contextualYes£1,400
Basket abandonersOpen auction, remarketingNo£900
  • Ad server: 6 impressions of the "Rear-facing from birth" creative on that page on Sunday evening, all from the basket abandoners line item, through one exchange and a direct seller.
  • Verification vendor: traffic measured as human; the page classified as accidents and tragedy.
  • Basket abandoners, last 7 days, on pages the vendor classified as accidents and tragedy: 1,850 impressions on 23 domains, costing £14.80.
  • The change log: the basket abandoners line item was built three weeks ago by copying a line item from before the policy was written. Nobody checked its brand safety settings against the policy before launch.
  • The contextual line item's top pages last month were crash-test ratings and fitting guides; 38% of its impressions were on pages whose address contains "crash".

Narrow controls, on the record

The brand safety and viewability guide warns that controls added in reaction to a screenshot tend to be too broad, removing good journalism along with the problem. Ottlecrag's case is sharper: the brand manager's keyword list would remove the crash-test pages that its contextual line item was built to reach, as the guide to planning contextual targeting warns when it tells planners to note the words that also appear in unsuitable contexts. Every change goes in the change log.

Task

  1. Decide whether this is an incident or a poor placement, and who leads.
  2. List the evidence to keep before any setting changes.
  3. Find the impression in the records and trace how it was bought.
  4. Choose the narrowest stop, and say what you decline from the chief executive and the brand manager.
  5. Check what was billed and whether any claim can be made.
  6. Draft the message to the chief executive.
  7. Find the cause and fix it where it lives, then decide what happens to any temporary stop.

Worked answer

Incident or poor placement. An incident: the policy excludes tragedy and accidents on every line item, so the settings did not match the policy. The trader leads, and the marketing director, as policy owner, is told the same hour.

Evidence. The original screenshot and email, the page address, time and device from the parent, a fresh capture of the page, and a note of who reported it and who has been told. All before any setting changes.

Records and route. The ad server and verification records confirm the screenshot is genuine and pin the impressions to the basket abandoners line item, open auction, one exchange, a direct seller. The page sat on a trusted news domain, which the policy allows; the topic did not.

The narrowest stop. Apply the tragedy and accidents exclusion to the basket abandoners line item, and check every other line item for the same gap. Decline the pause: it stops the line items that did nothing wrong and says nothing about the cause. Decline the keyword block: it would cut the contextual line item's best inventory and the trusted news the policy allows. Keep the news domain on the inclusion channel.

The bill.

Illustrative example — not a real result

Ottlecrag's billing view

ItemFigure
The screenshot page6 impressions, a few pence
Same pattern across the line item, last 7 days1,850 impressions, £14.80
Basket abandoners spend in the same week£900, so the pattern was 14.80 ÷ 900 ≈ 1.6% of its spend
Spend the chief executive's pause would have stopped across all line items in a week£2,100 + £1,600 + £1,400 + £900 = £6,000

The traffic was human and the seller authorised, so there is nothing to claim from the exchange or the publisher. The setting was the buyer's own: check whether the agency agreement covers settings that depart from the written policy, but do not assume a refund.

Message to the chief executive. "Our ad ran beside an accident report on a regional news site on Sunday evening. A remarketing line item, built recently, was missing the accident exclusion our policy requires; the others had it. We have added it and checked every line item. The rest of programmatic keeps running because it was set correctly. The parent has had our agreed reply. The cause and the change to our launch checks will be in the incident record this week." The parent receives the holding statement, promising nothing about the platform or the publisher.

Cause and fix. Why did the ad appear? The line item allowed the topic. Why? It was copied from a pre-policy line item. Why was that missed? The launch check does not compare brand safety settings with the policy. Fix the setting on the line item and at advertiser level, add a brand safety row checked against the policy to the programmatic campaign build sheet review, and ask the policy owner nothing new: the policy was right. No temporary block was added, so none needs lifting.

Common mistakes

  • Pausing everything. Most of the spend was set correctly.
  • Keyword blocks after a screenshot. They remove crash-test reviews, the brand's most relevant context.
  • Removing the news domain. The policy allows trusted news; the topic was the gap.
  • Fixing the screenshot page only. The same gap ran across many domains.
  • Rewriting a policy that was right. The setting departed from it; the launch check is what failed.

The checkpoint that follows tests these judgements alongside inclusion lists.

Pages this lesson applies

Terms used on this page

Also related

Related