Key concepts

AdCP signal types (binary, categorical, numeric), signal sources (catalog vs agent), discovery via get_signals, activation via activate_signal, and authorization through adagents.json.

Key concepts

The Signals Protocol enables AI agents to discover, activate, and manage data signals for advertising campaigns. Signals represent targetable audiences, contextual categories, geographic regions, and other data attributes.

What are signals?

Signals are data segments used for targeting or measurement in advertising campaigns:

• Audience signals: User segments based on demographics, interests, or behaviors
• Contextual signals: Content categories or page contexts
• Geographic signals: Location-based targeting data
• Temporal signals: Time-based targeting patterns
• Multi-dimensional signals: Combined or custom signal types

Signal value types

Every signal has a value_type that determines how buyers construct targeting expressions:

Binary

User either matches or doesn’t. The most common type.

{
  "id": "likely_ev_buyers",
  "name": "Likely EV Buyers",
  "value_type": "binary",
  "tags": ["automotive", "purchase_intent"]
}

Targeting: Include or exclude users matching this signal.

Categorical

User has one of several possible values.

{
  "id": "vehicle_ownership",
  "name": "Current Vehicle Ownership",
  "value_type": "categorical",
  "allowed_values": ["luxury_ev", "luxury_non_ev", "mid_range", "economy", "none"]
}

Targeting: Target users with specific values (e.g., “users who own a luxury EV or luxury non-EV”).

Numeric

User has a score or measurement within a range.

{
  "id": "purchase_propensity",
  "name": "Auto Purchase Propensity",
  "value_type": "numeric",
  "range": { "min": 0, "max": 1, "unit": "score" }
}

Targeting: Target users within a value range (e.g., “propensity score > 0.7”).

Signal sources

Signal IDs use source as a discriminator:

Source	Fields	Verification
catalog	data_provider_domain + id	Verifiable via data provider’s adagents.json
agent	agent_url + id	Trust-based — buyer trusts the agent

Catalog signals come from external data providers who publish their offerings at /.well-known/adagents.json. Buyers can independently verify that a signal agent is authorized to resell them.

Agent-native signals are proprietary to the signal agent — custom models, first-party data, or composite segments the agent builds from multiple sources.

The two tasks

Task	Purpose
get_signals	Discover signals matching campaign criteria
activate_signal	Activate a signal for use in campaigns

Discovery with get_signals

Buyers describe what they need in natural language. The signal agent searches across all its data providers’ catalogs and its own proprietary signals:

{
  "tool": "get_signals",
  "arguments": {
    "signal_spec": "In-market auto buyers with high purchase propensity"
  }
}

The response includes matching signals with pricing, size estimates, and value type metadata — everything a buyer agent needs to make a targeting decision.

Activation with activate_signal

Once a buyer selects a signal, they activate it on their DSP or data platform:

{
  "tool": "activate_signal",
  "arguments": {
    "signal_agent_segment_id": "trident_likely_ev_buyers",
    "pricing_option_id": "po_trident_ev_cpm",
    "destinations": [
      {
        "type": "platform",
        "platform": "the-trade-desk",
        "account": "agency-seat-123"
      }
    ]
  }
}

The signal agent pushes segment membership to the specified platform. The buyer’s campaign can then target against it using the platform’s standard tools.

Agent integration

The Signals Protocol operates within the broader AdCP ecosystem. Signal agents integrate directly with decisioning platforms (DSPs, orchestration platforms), eliminating intermediary reporting and usage tracking. Signal agents advertise their available data providers via get_adcp_capabilities.

Once signals are activated on a platform, all usage reporting, billing, and campaign metrics are handled directly by that platform.

Authorization and trust

Data providers control who can resell their signals via the authorized_agents array in their adagents.json. Two patterns:

• Signal IDs: Authorize specific signals by ID — fine-grained control
• Signal tags: Authorize all signals with certain tags — scales as catalogs grow

Buyers can verify authorization by fetching the data provider’s adagents.json and checking whether the signal agent appears in authorized_agents.

Go deeper

• Data provider guide — how to publish a signal catalog
• Signals ecosystem — how each type of company participates
• Protocol specification — formal conformance requirements
• get_signals task reference — discovery API details
• activate_signal task reference — activation API details