Customers
The person behind the ticket: their record, their history and their health score.
A customer is the person a ticket belongs to. Every ticket can be linked to one, and once it is, the ticket screen shows who you are talking to and what has happened with them before.
Customers live at Customers in the sidebar.
The list
The list shows name, email, company, country, and where the record came from. Search matches name, email and company: type and press Enter. The search and the sort live in the address bar, so a filtered list is a link.
Add customer takes a display name and, optionally, an email, a company and a country code. The email is unique within the workspace.
Where customer records come from
Two ways.
By hand, through Add customer or through the API.
By arrival. When a message arrives on a channel, the pipeline resolves who sent it and links the ticket to the matching customer. Email matches on the address, which is why keeping the email field accurate matters more than the rest of the record.
Linking a ticket to a customer
On the ticket, the crumb row has Link customer, which opens a searchable picker, and Unlink customer once one is linked. Linking is a write-level action.
Once linked, the ticket rail's Customer tab shows the email, company, country, when they were first seen, the channel, and the customer's recent tickets.
The customer page
Open a customer and you get their profile, their lifetime totals, and three tabs: conversations, tickets and activity.
The page carries:
- how many tickets they have, and how many are open,
- their most recent tickets, capped at ten,
- their most recent conversations, capped at ten, each with its channel, outcome and sentiment,
- sentiment and intent distributions,
- when they were first and last seen.
Neither recent list is pageable. For a customer's full ticket history, filter the ticket queue by that customer.
Edit customer changes the profile fields.
The health score
Each customer carries a health score from 0 to 100 and one of four labels:
champion, healthy, monitor or at_risk.
It is a plain formula, not a model. It starts at a neutral 70 and moves on recency of contact, the balance of positive against negative sentiment, how many times an AI agent handed off to a human, explicit churn signals, how many tickets are open, and how many conversations ended resolved.
The page lists every factor that moved the score and by how much, so any number can be explained. Treat the exact figure as advisory. It is a prompt to go and look, not a measurement.
Deleting a customer
Deleting is admin only, and it is refused while anything still references the record. The refusal names what is holding it.
Nothing is cascaded and no ticket is ever removed by this call. A customer with tickets cannot be deleted until those tickets are gone.