OneView automatically manages your Identity Graph as the data foundation of your workspace. No action required.
An Identity Graph is a database that connects all customer identifiers — anonymous clients, emails, phone numbers — without relying on a centralized or proprietary ID (e.g. like a CRM’s “master_id”). This allows you to incrementally build profiles, without having to know in advance to which (if any) customer(s) a given identifier will belong to. Instead of treating two email addresses as a separate user, an Identity Graph lets you distinguish when:
  • these belong to the same person (e.g. test account)
  • these are linked to multiple users (e.g. many accounts, same person)
Example:
When a visitor who browsed anonymously for weeks finally signs up, you instantly connect their entire history, giving you complete visibility into the customer journey and enabling accurate attribution, personalization, and analytics.

​
How Identities Connect

​
Connection Events

Identities link when users perform connecting actions:
1

Anonymous → Email

User signs up or enters email in form
// Before: Anonymous session
{ anonymous_id: "anon_123" }

// Action: User signs up
identify({ 
  anonymous_id: "anon_123",
  email: "user@example.com" 
})

// After: Connected identities
{ anonymous_id: "anon_123" } ←→ { email: "user@example.com" }
2

Email → Phone

User adds phone number to account
identify({
  email: "user@example.com",
  phone: "+1234567890"
})
// Graph expands: anon ←→ email ←→ phone
3

Personal → Work Email

User updates account email
identify({
  previous_email: "personal@gmail.com",
  email: "work@company.com"
})
// Both emails now connected

​
Retroactive Enrichment

To maintain historical consistency, after a conversion is processed, newly found identifiers are not retroactively computed to change the attribution of that conversion.You may choose to re-attribute that specific conversion if you need to re-evaluate its attribution based on information available after its conversion_time. Please note that re-attributing a conversion breaks historical consistency, so proceed with caution.You should instead model your funnel using different conversion events, one for each key step of the funnel.