Safety Norms: Programmatic Advertising Management

1Purpose, Scope, and Safety Principles

These norms set out how teams plan, buy, monitor, and optimize programmatic advertising in a way that protects people and keeps campaigns truthful and lawful. Programmatic advertising places ads through automated auctions, so one misconfigured setting can reach millions of impressions in hours. Harm spreads quickly, which makes preventive controls more valuable than cleanup after launch.

The scope covers campaign managers, trafficking specialists, analysts, creative reviewers, data partners, and vendors working in demand-side platforms, supply-side tools, data management platforms, verification services, and analytics dashboards. Where an organization policy is stricter than these norms, the stricter policy applies.

• Mandatory: no campaign launches without a documented objective, target audience definition, and approved creative.
• Mandatory: personal data is used only for purposes disclosed to people and permitted by applicable law and contracts.
• Recommended: default to the least intrusive targeting that still meets the campaign objective.

2Roles, Competence, and Accountability

The accountable owner for each campaign is the named account lead, who signs off on objectives, budgets, audience definitions, and risk acceptance. The trafficking specialist configures line items, creatives, and tracking tags. Compliance or legal reviewers confirm claims, disclosures, and data-use rules. Brand safety analysts maintain exclusion lists and verification settings.

Competence means trained use of each platform and tool, working knowledge of consent and privacy rules for the markets served, and the judgment to stop and escalate. Access to data management platforms, customer lists, and payment accounts is granted only after training and written approval, and removed promptly when roles change.

• Mandatory: each campaign records one named accountable lead in the campaign plan.
• Mandatory: access to audience data requires written approval from the data owner and the privacy reviewer.
• Recommended: pair new specialists with an experienced reviewer for their first few live campaigns.

3Activities, Hazards, and Risk Levels

Typical activities include building audience segments, setting bids and budgets, selecting inventory and publishers, approving creatives, running lookalike or retargeting models, and reporting results. Each activity carries specific hazards. Teams should score each campaign using the organization risk matrix, where risk combines likelihood and severity. A campaign that touches sensitive data or minors is automatically high risk and requires review before launch.

• High: exposure of sensitive groups, such as targeting based on health conditions, minors, or financial hardship. Example: a retargeting list built from visitors to a support page about a medical condition.
• High: ads carrying misleading prices, unlicensed images, or unverified claims shown to real people at scale.
• High: data leakage through shared spreadsheets, exposed API keys, or customer files sent to unapproved vendors.
• Medium: placement beside violent, hateful, or otherwise unsafe content, which harms audiences and the brand.
• Medium: ad fraud, invalid traffic, or spoofed inventory that wastes budget and distorts measurement.
• Medium: creative with rapid flashing that may trigger photosensitive reactions, or frightening imagery aimed at vulnerable people.
• Low to medium: eye strain, neck and wrist strain, and fatigue during long monitoring sessions.
• Low: frequency fatigue, where repeated exposure annoys people and erodes trust in the brand.

4Preventive Controls and Safe Campaign Practices

Preventive controls work before a campaign goes live. Every campaign passes a pre-launch checklist covering objective, audience, creative, placement exclusions, frequency caps, consent basis, and measurement. Post-launch changes that alter audience, creative, or placement categories follow the same review path as a new campaign.

Budget and bid changes are made only through approved accounts. Changes above a threshold set by the organization require a second reviewer. Stop-loss triggers should be agreed before launch and recorded, so that anyone on the team knows when to pause spending.

• Mandatory: audiences must not use or infer sensitive categories that platform policy or applicable law restricts, including health, religion, sexual orientation, political affiliation, and financial distress. Confirm restricted categories against current platform policy before build.
• Mandatory: children are excluded from behavioral targeting and data collection unless counsel has approved a documented compliant approach for the market.
• Mandatory: every line item has a frequency cap set by the campaign lead and recorded in the plan.
• Mandatory: all creative is reviewed for accurate claims, required disclosures, trademarks, and licensed imagery before upload.
• Mandatory: creative containing flashing or rapid transitions is flagged for photosensitivity review before approval.
• Mandatory: brand safety exclusion lists are applied and verified with approved third-party tools where available.
• Recommended: review live placements weekly by viewing the actual pages, not only dashboard reports.
• Recommended: use contextual targeting first, and treat first-party data as an asset governed by its data owner.
• Recommended: agree a stop-loss trigger, such as an anomaly in spend or invalid traffic, and name who may pause the campaign.

5Audience Privacy, Consent, and Creator Rights

Programmatic campaigns often rely on cookies, device identifiers, mobile advertising IDs, hashed email addresses, and clean room matching. Hashing does not make data anonymous. Treat hashed identifiers as personal data unless privacy counsel concludes otherwise. Consent requirements, opt-out rights, and data transfer rules differ by jurisdiction and must be verified with qualified privacy professionals.

Creators, photographers, influencers, and models hold rights in their work and their likeness. Teams must not repurpose user-generated content, creator footage, or talent images without written permission that covers the channel, duration, and territory. Creator safety matters too. Do not publish creator home locations, private messages, or personal contact details, and never ask a creator to perform a stunt or unsafe act for ad content.

• Mandatory: obtain and retain licenses for every image, music track, font, and video clip used in ads.
• Mandatory: do not use a person's name, image, or voice in an endorsement without a signed release.
• Mandatory: do not upload customer lists without a documented lawful basis and a data processing agreement with each vendor.
• Mandatory: honor opt-out and consent signals passed through platforms and consent management tools.
• Recommended: collect only the fields needed for matching, and delete raw uploads at the end of the agreed retention period.
• Recommended: provide creators a named contact for takedown or correction requests, and respond within the time set in the contract.

6Truthful Communication and Reputational Risk

Ads must be truthful, substantiated, and not misleading about price, availability, results, or the identity of the advertiser. Programmatic scale makes errors costly, since one wrong offer can appear across thousands of sites before anyone notices. Reputational harm also arises from placements beside extremist content, stereotyping imagery, and opaque data practices that people discover later.

Sensitive campaigns need a reputational plan before launch. The plan names the spokesperson, the approval path for any public statement, and the criteria for pausing the campaign.

• Mandatory: every quantitative claim has a source file that the claims reviewer can check before launch.
• Mandatory: paid partnerships and sponsored content are disclosed as required by platform rules and the law of each market. Confirm current rules with the compliance team.
• Mandatory: no fake reviews, invented testimonials, or interface imitations that suggest a platform control or official endorsement.
• Recommended: review creative for stereotypes and for imagery that could cause distress, and involve diverse reviewers where practical.

7Tools, Workspace, and Information Security

Advertising accounts, data management platforms, and reporting dashboards hold credentials that control spend and customer data. A compromised account can be used for fraudulent spending, data theft, or the publication of malicious ads. Workspace safety also covers the physical environment where analysts spend long hours at screens.

Security and ergonomics are both managed through the workspace owner, usually the office manager or occupational health contact, with the security team responsible for accounts.

• Mandatory: multi-factor authentication is enabled on all advertising, data, and payment accounts, and logins are never shared.
• Mandatory: API keys and passwords are stored only in the approved password manager or secrets vault, not in spreadsheets, chat messages, or code repositories.
• Mandatory: access is revoked promptly when someone leaves or changes role, and the security team confirms removal.
• Mandatory: suspicious login alerts, unexpected charges, or unknown users are reported immediately to the security contact.
• Recommended: set screens at or slightly below eye level, keep keyboards at elbow height, and take breaks at the intervals set by local occupational health guidance.
• Recommended: keep dashboards that display personal data off shared displays in open-plan areas.

8Incident Reporting, Emergency Response, and Escalation

An incident includes a data exposure, an unsafe or illegal ad live on a platform, a fraud event, a serious complaint, a creator dispute, or a near-miss where an error was caught before harm. Report every incident the same working day, even if the impact is unclear. Near-misses are reported because they reveal control gaps that could cause harm later.

Immediate actions are to pause the affected line item or campaign, preserve screenshots and logs, stop further data uploads, and notify the accountable lead and the security or privacy contact. Do not delete evidence, contact affected individuals without approval, or argue publicly with complainants. If anyone faces a risk of physical harm, contact local emergency services first and then the account lead.

• Mandatory: escalate any personal data exposure to the privacy lead and legal team within the time set in the incident procedure.
• Mandatory: record the root cause and corrective action for each incident in the incident log.
• Mandatory: keep platform support contacts and appeal routes on file for each account.
• Recommended: run a tabletop exercise on a data exposure and a harmful ad scenario at a set interval defined by the risk owner.

9Records, Training, Review, and Continuous Improvement

Keep campaign plans, approval records, creative files with license evidence, consent logs, data processing agreements, access reviews, and incident logs. Retention periods depend on law and contract and must be confirmed locally. Training covers this document, current platform policies, privacy basics, and the escalation path. Training completion is recorded before anyone receives live account access.

Monitoring is continuous. Teams review placements, complaints, fraud signals, and access lists, then update these norms after significant incidents or policy changes. Local procedures should be written from this framework and approved by the accountable leadership and the relevant specialists.

• Mandatory: access lists are reviewed at an interval defined by the security team, and unused accounts are removed.
• Mandatory: staff are retrained when platform policies or privacy rules change for a market they manage.
• Recommended: hold a short post-campaign review that records what worked, what failed, and which norm should change.
• Recommended: audit a sample of live placements and creative files each week.

Implementation checklist

• Confirm a named accountable lead and an approved objective are recorded before launch.
• Check every audience for restricted or sensitive categories and remove any that are not permitted.
• Verify that licenses, releases, and claim sources are on file for all creative.
• Confirm the consent basis, opt-out handling, and vendor data processing agreements for any data use.
• Apply frequency caps, brand safety exclusions, and a stop-loss trigger to every line item.
• Store credentials in the approved vault and confirm multi-factor authentication is active.
• Review live placements by viewing the pages, not only the reports, at least weekly.
• Report incidents and near-misses the same working day, and pause affected campaigns first.
• Log corrective actions and confirm training is current for everyone with account access.