Customer Database · Schema

Customer Identity Link

An assertion that an external identifier belongs to a customer, carrying the method and confidence that produced it. These are the edges of the identity graph. No open standard models identity resolution, which is why identity graphs are proprietary and do not port between vendors; this schema is a neutral shape for describing one.

AccountConsentContactsCRMCustomer DataCustomer Data PlatformCustomersDatabaseGDPRIdentityProfilesSchema

Properties

Name Type Description
id string
customerId string
namespace string The system or identifier space the value belongs to. Without a namespace an identifier is not resolvable.
identifier string The value itself. Hash rather than store raw identifiers where the namespace implies it.
method string How the link was established. deterministic = exact agreement on a strong key; probabilistic = weighted similarity across weak attributes; declared = the customer told us.
confidence number Match score. REQUIRED and meaningful for probabilistic links; meaningless for deterministic ones, where its presence usually signals a modeling error.
matchedOn array Which attributes contributed to the match. Makes a probabilistic link reviewable instead of opaque.
linkedAt string
expiresAt string Pseudonymous identifiers decay. A cookie link that never expires overstates what is known.
active boolean Set false to break a link without losing the record that it was once asserted.
View JSON Schema on GitHub

JSON Schema

customer-identity-link.json Raw ↑
{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://raw.githubusercontent.com/api-evangelist/customer-database/refs/heads/main/json-schema/customer-identity-link.json",
  "title": "Customer Identity Link",
  "description": "An assertion that an external identifier belongs to a customer, carrying the method and confidence that produced it. These are the edges of the identity graph. No open standard models identity resolution, which is why identity graphs are proprietary and do not port between vendors; this schema is a neutral shape for describing one.",
  "type": "object",
  "required": ["namespace", "identifier", "method"],
  "additionalProperties": false,
  "properties": {
    "id": { "type": "string" },
    "customerId": { "type": "string" },
    "namespace": {
      "type": "string",
      "description": "The system or identifier space the value belongs to. Without a namespace an identifier is not resolvable.",
      "examples": ["crm", "cookie", "device-idfa", "device-gaid", "email-sha256", "phone-e164", "loyalty", "billing"]
    },
    "identifier": { "type": "string", "description": "The value itself. Hash rather than store raw identifiers where the namespace implies it." },
    "method": {
      "type": "string",
      "enum": ["deterministic", "probabilistic", "declared"],
      "description": "How the link was established. deterministic = exact agreement on a strong key; probabilistic = weighted similarity across weak attributes; declared = the customer told us."
    },
    "confidence": {
      "type": "number",
      "minimum": 0,
      "maximum": 1,
      "description": "Match score. REQUIRED and meaningful for probabilistic links; meaningless for deterministic ones, where its presence usually signals a modeling error."
    },
    "matchedOn": {
      "type": "array",
      "description": "Which attributes contributed to the match. Makes a probabilistic link reviewable instead of opaque.",
      "items": { "type": "string" }
    },
    "linkedAt": { "type": "string", "format": "date-time" },
    "expiresAt": { "type": "string", "format": "date-time", "description": "Pseudonymous identifiers decay. A cookie link that never expires overstates what is known." },
    "active": { "type": "boolean", "default": true, "description": "Set false to break a link without losing the record that it was once asserted." }
  }
}