What Is Ticket Deflection Rate? (And How to Calculate It)

Every support team wants to answer more questions without adding headcount. The metric that proves you are doing it is ticket deflection - the share of support requests resolved by automation instead of a human agent. This is the number CFOs ask about, and it is the reason AI support platforms exist.

What ticket deflection means

Ticket deflection is the percentage of incoming support requests that a self-service channel - typically an AI chatbot or knowledge base - resolves completely, so no human ticket is ever created. A deflected ticket never costs you agent time.

The formula

Deflection rate = Deflected requests / Total support requests x 100

For example, if your chatbot resolves 340 of 1,000 requests end-to-end, your deflection rate is 34%. Count only fully resolved interactions; an interaction that escalates to a human is not deflected.

What a good deflection rate looks like

Most teams start between 15% and 25% with a basic FAQ bot. With a well-trained RAG chatbot and thoughtful handoff rules, 40% to 60% is realistic. The best-performing knowledge-base bots push past 60% for product-heavy companies, because the same questions repeat constantly.

How to lift your deflection rate

• Train on real tickets - the questions customers actually ask, not the docs you think they ask.
• Keep the knowledge base fresh - index new articles immediately and retire outdated ones.
• Add a confidence threshold - let the bot answer when confident, escalate when not, and never guess.
• Close the loop - route unresolved conversations back into your docs so the bot learns.

Why deflection is a profit metric

If each resolved ticket costs $4 in agent time, lifting deflection from 25% to 55% on 10,000 monthly requests saves $12,000 a month - against an AI bill that is a rounding error. That is the math behind Chatonn's analytics dashboard, which tracks deflection, resolution rate, and CSAT in real time so you can see the ROI on day one.