Supply Descriptors

If machine-readable supply is the goal, the supply descriptor is the vehicle. A supply descriptor is a structured data record that describes a unit of advertising inventory: what the content is, what commercial context it creates, what constraints the publisher imposes, and who is attesting to those claims.

Think of it as the product manifest for an ad placement. A physical product's manifest specifies dimensions, weight, materials, and compliance certifications. A supply descriptor specifies content type, topics, brand suitability, freshness, provenance, and more. A buyer agent reads the manifest to determine whether the product meets its requirements — without having to open the box.

Anatomy of a Supply Descriptor

A supply descriptor is not a single standard with one fixed schema. Different protocols — AdCP and others — define their own schemas. But the category of information a supply descriptor contains is consistent across implementations. The following fields appear in most mature supply descriptor formats:

Content type. A classification of what kind of content this is: article, video, audio, live stream, listing. Content type is the broadest structural signal — it tells a buyer what kind of ad unit and what kind of editorial context to expect.

Topic taxonomy. A hierarchical set of topic labels that describe what the content covers at multiple levels of specificity. Topic taxonomy in a supply descriptor is more precise than a broad IAB category: it may indicate that an article covers "personal finance > retirement planning > IRA contribution limits" rather than just "Finance."

Brand suitability. An explicit assertion about what categories of advertising are appropriate for this inventory. Brand suitability is publisher-controlled — the publisher (or a system acting on the publisher's behalf) states what kinds of brands and messages are contextually appropriate, inappropriate, or acceptable within defined conditions.

Freshness. A timestamp indicating when the content was last meaningfully updated. Freshness is a significant signal for some campaign types: a news campaign values recent content; an evergreen information campaign may not.

Publisher identity. Who publishes this content: domain, name, and any available verification signals. This field enables a buyer agent to apply publisher-level policy (e.g., preferred publisher lists, reputation signals) in addition to content-level evaluation.

Provenance record. A chain of custody that indicates how the descriptor's claims were established — whether by the publisher directly, by an automated intelligence system, by a third-party audit, or some combination. Provenance is covered in detail in Unit 7.

What a Descriptor Describes

It is worth being precise about what a supply descriptor is actually describing. A supply descriptor does not describe the audience — it describes the supply. These are different things.

Audience data describes who is likely to be viewing the content: demographic estimates, inferred interests, behavioral signals. Audience data is a claim about people, and in most systems it involves some form of user-level tracking or inference.

A supply descriptor describes the context: what the content is, what it creates for a viewer, what commercial signals it carries. Supply data is a claim about content. It requires no user-level information. A publisher can produce an accurate supply descriptor for a page without knowing anything about the people who read it.

This distinction matters both ethically and practically. Ethically, supply-based buying does not require or involve user profiling. Practically, supply descriptors are stable and directly actionable: a buyer agent evaluating context does not need to wait for audience inference; it reads the descriptor.

How Descriptors Are Consumed

A buyer agent receives a supply descriptor — either in response to a query, through a feed, or as part of an exchange initiated by a seller agent — and evaluates it against the campaign brief.

This evaluation is not a keyword match. It is a semantic comparison: does the content's topic coverage align with the campaign's intent? Does the brand suitability declaration meet the campaign's thresholds? Is the freshness within the campaign's window? Is the publisher's provenance level sufficient for the buyer's trust requirements?

If the descriptor passes all applicable checks, the buyer agent determines that the opportunity is eligible and proceeds toward execution. If it fails on any dimension, the agent passes — and, in a well-designed system, records why.

The evaluation logic sits in the supply intelligence layer, which is the subject of the next unit. Supply intelligence is what transforms raw supply descriptors into scored, ranked opportunities that a buyer agent can act on with confidence.

The Descriptor as Contract

A supply descriptor is not merely informational. When a publisher produces a descriptor, they are making claims that a buyer agent will rely on. If the descriptor asserts brand suitability and the content is actually unsuitable, the transaction that results from that assertion fails. The publisher's credibility — and the reliability of the supply descriptor system overall — depends on accurate claims.

This accountability relationship is what distinguishes supply descriptors from the programmatic-era signals they replace. An inferred audience category is a probabilistic estimate with no accountable author. A supply descriptor is an assertion with a source. The provenance layer (Unit 7) is what makes that accountability visible and verifiable to buyer agents.

Unit 6 covers how supply intelligence systems evaluate descriptors to produce confidence scores. Unit 7 covers how provenance makes those scores trustworthy.