Why suitability needs a written policy and who should own it
A demand-side platform (DSPSoftware advertisers use to buy programmatic inventory.) applies whatever rules it is given and has no view of its own about the brand. When nobody writes the rules down, they are set by whoever builds the line item, copied to the next one, loosened when a deal will not spend and tightened when someone forwards a screenshot. After a year the settings reflect a series of moments, not a decision, and nobody can say whether a given placement was inside the brand's tolerance or not. The programmatic advertising audit treats this as a gating weakness: it scores safety and suitability settings against the brand's written tolerance, and a missing tolerance scores as missing however careful the settings look.
A written policy fixes three things. It makes the judgement the brand's, decided once by the people who own the brand, rather than the trader's guess. It makes the settings explainable, because each one can be traced to a line of the policy. And it makes change deliberate: a looser setting for one deal becomes an exception someone approved, not a tick box someone cleared.
Ownership follows from that. The brand or marketing owner decides the tolerance and signs the policy; the trader or agency translates it into settings and says what each choice costs in reach and price; whoever handles press and complaints is consulted on what the brand could not defend in public; finance and legal are informed where the product is regulated. A RACI matrix records those roles in a few lines, and the guide to brand identity and voice is the natural companion document, because where a brand will appear says as much about it as how it speaks. A social media policy is a different document with a different job, governing what staff post rather than where paid media runs, though the habit of writing a stance down, naming an owner and reviewing it carries over.
The safety floor and the tiers above it
Brand safetyControls that keep ads away from content that would harm the advertiser's reputation. and brand suitability are separate questions. Safety is the floor: illegal material, terrorism, explicit content, incitement and the other content no advertiser should appear beside. The brand safety guide notes that the industry's shared vocabulary for the floor and the risk tiers above it came from the brand safety floor and suitability framework published by the GARM initiative; the initiative has closed, but verification vendors and platforms still use its categories and tiers as a common language. The policy states the floor in a sentence, says it is on by default for every buy, and says it is not open to exception.
Invalid traffic belongs beside the floor rather than in the tiers. It is not a matter of taste: the glossary entry for invalid traffic describes how industry guidance separates general invalid traffic, which routine filtering removes, from sophisticated invalid traffic, which is built to look human, and the Media Rating Council's standards are the reference for both. The policy says that IVTAd impressions and clicks not caused by genuine human interest: bots, fraud and accidents. filtering is on for every buy and that verification is used where the plan requires independent measurement.
Suitability is everything above the floor. A news report on a war is safe in the industry sense and unsuitable for some brands, essential context for others and neutral for most. The brand safety guide describes suitability as set per brand, usually in tiers of risk tolerance. A policy can use the tiers its verification vendor or platform offers as a shared scale, but it should not simply adopt the vendor's default tier: it should say, topic by topic, where this brand stands, and why. The tiers are a vocabulary for the decision, not the decision itself.
Topics, news and user-generated content: setting the stance
The core of the policy is a short table that takes each sensitive topic and states the brand's stance. Three stances are enough for most brands: excluded everywhere, allowed only on named publishers, or allowed under the default settings. The reasons column matters as much as the stance, because the next person to read the policy will want to know whether a choice came from the product, the audience or an event.
| Topic or content type | Typical stances to choose between | What drives the choice |
|---|---|---|
| Hard news, including conflict, crime and disasters | Allowed on trusted publishers regardless of topic; allowed under default labels; excluded | Whether the audience reads news and whether the brand can stand beside difficult stories on reputable sites |
| Tragedy and accidents | Excluded everywhere; excluded only near related subjects; allowed on trusted publishers | Whether the product could be read as exploiting the event, such as travel beside a crash |
| Alcohol, gambling and dating | Excluded; allowed | Whether the brand sells one of these, competes with them or simply does not want to be near them |
| Political and social debate | Excluded; allowed on trusted publishers | Whether appearing beside a side of a debate could be read as taking it |
| Crude humour and mature themes | Excluded; allowed with mature content labels | The age and expectations of the audience |
| Content made for or popular with children | Excluded; allowed | Whether the product or the law makes children an audience the brand must avoid |
| User-generated pages: forums, comments, social feeds | Excluded; allowed on named communities | Moderation quality, which varies far more than on edited sites |
| Unlabelled or unclassified content | Excluded; allowed on listed publishers | How much of the inventory the plan needs is unlabelled, as is common in deals |
News deserves a deliberate line of its own, because the commonest suitability mistake is to treat it as unsafe. The brand safety guide warns that keyword exclusions and broad category settings remove a large share of quality journalism because it reports on difficult subjects, pushing budget towards inventory that reports on nothing. A better stance keeps a list of trusted news publishers that may be bought regardless of topic, relies on page-level classification rather than keywords elsewhere, and accepts the occasional difficult adjacency on a reputable site as a smaller risk than funding content farms. The tutorial on building a programmatic inclusion list judges every candidate publisher against this table, so the policy should be finished before that list is built.
Contextual targetingChoosing where ads appear from the content of the page, app or video being viewed rather than from data about the viewer. is the positive side of the same document. Where the policy names topics the brand wants to appear beside, because they suit the product, the planner can use them to target as well as the trader to exclude, and both can read the same table.
Regulated products need more than a stance. Alcohol, gambling, financial products and medicines are subject to advertising rules that differ by market, and the policy should record that legal or compliance staff have reviewed it rather than attempt to state the rules itself.
Exceptions, approvals and the people who decide
A policy without an exception route is ignored the first time a deal will not spend. Write the route down: who may ask for an exception, who approves it, what the request must state, and how long it lasts. A sensible request names the line item or deal, the setting to be loosened, the reason, the reach or delivery problem it solves and an end date; the approver is the policy owner or a named deputy, never the person asking. Approved exceptions are recorded in a log the policy points to, so an auditor can see that a looser setting was a decision rather than a drift. The academy suitability and incidents checkpoint tests who decides and approves exceptions, alongside the safety floor and the stance on news.
The same section names who decides when something goes wrong. When a screenshot arrives, someone must judge whether the placement breached the policy or was inside a tolerance the brand chose, and that someone should be the owner, not whoever happens to receive the email. The tutorial on responding to a brand safety incident works through that day step by step and ends by feeding its cause back into this policy.
Turning the policy into DV360 settings and vendor segments
The trader turns each line of the policy into a setting, and the DV360 inventory and deals guide describes what is available: digital content labels that classify content by suitability and let a line item exclude the stricter levels and unlabelled content; sensitive category exclusions; keyword exclusions; channels that include or exclude lists of URLs and apps; viewability targeting built on Active View; and pre-bid segmentsVerification applied before the bid: a vendor's segments, checked by the DSP, decline impressions classed as unsafe, unsuitable, unviewable or invalid. from the verification vendors DV360 lists. The DV360 account structure guide adds that brand-safety settings can be set as advertiser defaults, so every new line item starts inside the policy without anyone remembering to tick a box. The Help Centre is the reference for the current label and category names; the policy should describe the intent and let the settings record carry the names, because names change.
| Policy line | Where it becomes a setting | Notes |
|---|---|---|
| Safety floor on for every buy | Advertiser defaults: sensitive categories and the strictest content labels | Not overridden on any line item |
| Topic stances | Sensitive category exclusions, content label levels, vendor suitability segments | Keyword exclusions only for narrow, unambiguous terms |
| Trusted news and named publishers | An inclusion channel, or deals with those publishers | Content settings on those line items set to match the stance |
| User-generated pages | Content settings, vendor segments and channel exclusions | Named communities on an inclusion channel where allowed |
| Unlabelled content | The unlabelled content setting on each line item | Checked against each deal's inventory before launch |
| Viewability expectations | Viewability targeting or optimisation | A separate filter: ViewabilityShare of measured ad impressions that had a real chance to be seen. says nothing about suitability |
| Exceptions | Line item overrides with the log reference in the name or notes | Removed on the end date |
Two habits keep the translation honest. Record the settings per line item in the programmatic campaign build sheet, whose brand safety and verification section asks for exactly this, so the reviewer can check each line against the policy before the build. And put the policy, or a link to it, in the media brief under mandatories, so the plan is written inside it rather than discovering it at launch. Buyers who also run Google Ads display will find simpler versions of the same controls in the display targeting options guide, and the same policy should govern both.
Reviewing the policy against what actually ran
A policy is tested by the placements it produced. On a regular rhythm, read the site and app report and the verification report by spend, and ask three questions. Did anything run that breaches the policy? Did the settings remove inventory the policy allows, such as trusted news blocked by a keyword or a deal starved by an unlabelled content setting? And did anything happen in the world, or in the business, that the policy does not yet address?
The answers lead to different changes. A breach is a setting that does not match the policy, so the setting is fixed. Over-blocking is a setting stricter than the policy, which is also fixed, often recovering reach the plan was paying for. A gap is the policy's own fault: the owner decides the new stance and the trader translates it. Record each review with its date and outcome, and re-read the policy whenever the product range, the audience or the agency changes.
A craft brewery writes its suitability policy (illustrative)
Grenblatz Craft Brewery writes its suitability policy with its agency
Grenblatz Craft Brewery, an invented independent brewer selling through its own web shop and 400 pubs and bottle shops, plans £60,000 of programmatic display and video over a year. Its agency had been using the platform's standard tier for every line item, and the first placement report showed most national news sites blocked by a keyword list.
The marketing director owns the policy; the agency trader drafts the settings; the founder, who answers press questions, is consulted. They write the policy in a single meeting.
| Line | Stance |
|---|---|
| Floor | On by default for every buy; no exceptions |
| Children's and family content | Excluded everywhere, because the product is alcohol; reviewed by the brewery's compliance adviser |
| Hard news | Allowed on a named list of 24 national and regional publishers regardless of topic; keyword list removed |
| Tragedy | Excluded on all line items |
| User-generated pages | Excluded, except 6 named homebrewing and pub forums |
| Unlabelled content | Excluded on open-auction line items; allowed on the 2 regional news deals |
| Owner and exceptions | Marketing director approves; exceptions logged with an end date |
| Review | Quarterly, against the site and verification reports by spend |
The trader turns it into advertiser defaults for the floor, the children's and tragedy exclusions and unlabelled content, a vendor suitability segment on open-auction line items, and an inclusion channel holding the news publishers and forums. At the first quarterly review the news publishers take 31 per cent of spend, against almost none before, and no placement breaches the policy; one gap is found, a sports forum nobody had named, and the owner decides to exclude it.
Common mistakes
- Adopting the vendor's default tier as the policy. The tier is a vocabulary; the decision about this brand still has to be made and written.
- Letting the trader set the tolerance. The person who owns the brand decides; the trader translates and prices.
- Blocking news by keyword. It removes quality journalism and leaves the content that reports on nothing.
- Mixing safety and suitability. The floor is fixed for everyone; only the tiers above it are the brand's choice.
- No exception route. The first deal that will not spend gets loosened silently instead of approved and logged.
- Setting suitability on each line item. Advertiser defaults make the next line item start inside the policy.
- Treating viewability as suitability. A highly viewable placement can still be one the brand would never choose.
- Writing the policy once. Without a review against what ran, it drifts away from the settings and the world.
