Containment Rate
Definition
Containment rate is a metric specific to AI chatbot and IVR (Interactive Voice Response) systems that measures what percentage of initiated conversations are fully handled by the automated system without any human agent involvement. It differs from deflection rate in specificity: deflection rate measures all self-service resolutions (including knowledge base, FAQ, etc.), while containment rate specifically measures what proportion of conversations a particular automated system contains versus escalates. A chatbot with a 70% containment rate resolves 70% of initiated conversations autonomously and escalates 30% to human agents.
Why It Matters
Containment rate directly measures the efficiency and capability of an AI chatbot investment. A chatbot that cannot contain interactions is an expensive routing layer rather than a true support automation tool. For support leaders evaluating chatbot ROI, containment rate translates directly into cost savings — each contained conversation costs a fraction of a human-handled equivalent. However, containment rate must be balanced against resolution quality: artificial inflation (e.g., ending conversations rather than resolving them) damages customer experience without genuine cost savings.
How It Works
Containment rate is calculated by the chatbot platform from session data: (Sessions ended without human escalation) / (Total sessions initiated) x 100. Most chatbot platforms provide this metric natively in their analytics dashboard. Analysis should segment containment by intent category — chatbots typically contain 90%+ of simple FAQ-type intents but may contain only 40-50% of complex process intents. This segmentation guides content and capability development priorities. Low-containment intent clusters reveal where AI capability or knowledge base content needs improvement.
Containment Rate — 100 Chatbot Sessions
Contained
73
resolved within bot
Escalated
27
to human agent
Contained breakdown
Real-World Example
A 99helpers customer reviews chatbot containment rate monthly by intent category. They find their chatbot fully contains 91% of 'product information' and 'pricing' intents, but only 38% of 'billing dispute' intents. Billing disputes are escalating because the chatbot cannot access account-level billing data in real time. They integrate the chatbot with their billing API, enabling it to look up recent charges, explain line items, and issue small credits autonomously. Billing intent containment rate improves from 38% to 74%, removing hundreds of monthly escalations.
Common Mistakes
- ✕Maximizing containment rate at the expense of resolution quality — define containment as 'customer confirmed resolution' not just 'no escalation requested'
- ✕Treating low containment as solely a chatbot problem — low containment for certain intents may indicate that those issues genuinely require human judgment or authority
- ✕Not segmenting by intent — overall containment rate hides whether the chatbot is excellent at some things and poor at others
Related Terms
Deflection Rate
Deflection rate is the percentage of potential support contacts that are resolved through self-service, AI chatbots, or automated tools without requiring a human agent, measuring the effectiveness of automated and self-service support.
Chatbot Deflection
Chatbot deflection is the process by which an AI chatbot resolves a customer's inquiry completely, preventing the need for human agent involvement and reducing the volume of tickets that reach the support team.
Self-Service Rate
Self-service rate measures the percentage of customers who successfully resolve their issues independently through knowledge bases, FAQs, AI chatbots, or other self-service tools without contacting a human agent.
Ticket Escalation
Ticket escalation is the process of transferring a support issue to a higher-tier agent, specialist, or team when the current handler lacks the authority, expertise, or tools to resolve it.
Customer Satisfaction Score
Customer Satisfaction Score (CSAT) is a metric that measures how satisfied customers are with a specific interaction, product, or experience, typically collected through a simple post-interaction survey asking customers to rate their satisfaction on a numeric scale.
Ready to build your AI chatbot?
Put these concepts into practice with 99helpers — no code required.
Start free trial →