Intercom defines containment as "the conversation closed without human intervention". That's an operational definition, useful for Intercom. It's not aligned with what you actually want to steer.
What does a high containment rate actually hide?
- Real resolution: the customer got their answer and is satisfied.
- Abandonment: the customer closed the tab for lack of a useful answer. Counted as containment.
- Channel switch: the customer left to email, DM via LinkedIn, or reopen an account. Counted as containment.
Without separating those three, your 75% containment covers both a 4.2 quality score and a 2.1 quality score equally well.
How do you calculate net containment?
The honest formula worth reviewing:
Net containment = gross containment − (7-day reopen rate × gross containment) − (rate of tickets opened by the same contact within 48h on another channel).
Why shouldn't you set a containment target for the team?
Committing to "containment +10 points" as a team goal mechanically pushes toward the worst case: aggressive auto-close, FIN mis-calibrated on complex topics, agents who resist escalating. Quality drops, containment climbs. The dashboard is green, retention moves the other way.
What should you steer by instead of containment?
- Net containment (formula above).
- Average CSAT on contained vs escalated conversations.
- 7-day reopen rate on contained conversations.
The surface metric is a communication tool. The net metric is a steering tool. Both can live, but you don't hand team decisions to the first one.