Define and organize customer fields
Use consistent customer properties in profiles, views, saved replies, and AI context.
Open Settings → Inboxes → choose an inbox → Goals & customer data to define the properties accepted and used for that inbox. Examples include plan, account type, or onboarding stage.
Use stable keys and clear labels
Give each field a readable label and a stable key, such as subscription_plan. Choose the matching type: Text, Number, Boolean, Date, or Preset.
Your integration should send values using the key and the correct type. A boolean true is different from the text "true". Use lowercase snake_case for new integration keys.
Changing a label or definition does not rename values already stored on customer profiles. Use field merging for that.
Put fields to work
Customer fields appear in profiles and conversation details and can provide AI context. You can also use them in saved-reply variables, inbox filters, list columns, and playbook conditions.
Use the customer-data ordering controls to put the most useful fields first. These shared field-order choices are separate from your inbox list’s column selection.
Signed widget identification can update a customer’s property snapshot without opening chat. Those properties do not automatically become trusted evidence for shared outcome tracking.
Rename or combine stored fields
Open Merge customer fields in inbox settings. This is a workspace-wide operation and requires owner or administrator access to every inbox.
- Find the exact source key.
- Choose a new or existing destination key in lowercase snake_case.
- Select Preview field merge and review affected customers and conflicts.
- Confirm the merge and follow its progress in Recent field merges.
Existing destination values win, including empty, null, false, and zero. The source key is removed from customer records, and the operation cannot be undone automatically.
Update saved replies, filters, playbooks, outcome rules, and your integration to use the new key; those references aren’t rewritten for you. A failed job can use Retry from checkpoint after the underlying issue is resolved.