What is containment rate?
Containment rate measures how much of your contact volume an automated channel deals with on its own. If 1,000 people call and 400 of them get what they need from an AI voice assistant or IVR without speaking to an agent, the containment rate is 40%.
The term is used for any automated front end: IVR, chatbots, voicebots and other customer self-service. It’s one of the headline numbers vendors and project teams use to describe the value of automation, because every contained contact is one an agent doesn’t have to handle.
How containment rate is calculated
The basic formula is:
Containment rate = (contacts handled entirely by automation / total contacts entering the automated channel) x 100
The detail matters, because small definitional choices change the number a lot:
- What counts as “handled”? Some measure only that the contact ended without a transfer. Others require a positive outcome, such as a fault logged or a question answered.
- What about abandoned contacts? If a customer hangs up mid-conversation, some reports count it as contained. That flatters the figure.
- What about repeat contacts? A customer who is “contained” but calls back an hour later and asks for a person wasn’t really helped.
- Which contacts are in scope? Containment across all calls is very different from containment of the call types the automation was designed for.
When comparing figures, always ask how they were defined.
True containment vs false containment
It helps to separate two kinds of contained contact:
- True containment. The customer’s need was met. They got the update, logged the fault, or completed the change, and didn’t need to make contact again.
- False containment. The customer left without help. They gave up, hung up or switched to another channel.
Only true containment reduces workload and improves the experience. False containment just moves the problem, often into angrier calls later. Pairing containment with first contact resolution, repeat-contact rates and customer effort shows which kind you have.
Containment in a surge
Containment is most valuable when demand spikes. In a call avalanche after a storm, most callers want the same information, and an AI voice assistant can give it to them instantly. Every routine call contained is a queue position freed up for a hazard call or a medically dependent customer who needs a person. In that context, containment and good service work together.
Improving containment the right way
- Automate the right contacts. Focus on high-volume, low-complexity requests the AI can complete end to end.
- Connect to real data. Automation that can actually check an outage or log a job contains far more than one that only reads out general information.
- Make the handover easy. Counter-intuitively, a clear route to a person builds trust and makes customers more willing to try the automated option.
- Exclude what shouldn’t be contained. Safety, vulnerability and complaints should go to people.
- Review real conversations where customers dropped out or asked for an agent, and fix the cause.