Support Metrics
Deflection Rate
Definition
Deflection rate is the percentage of support queries resolved by self-service, such as help articles, FAQs, or an AI chatbot, without reaching a human agent.
In depth
Deflection rate measures how many of the people who come looking for support get their answer from self-service, rather than from a human agent. It is written as a percentage because it compares one number against another: the interactions resolved without an agent, divided by everyone who sought help.
The maths is straightforward. Suppose customers start 500 support conversations in a week. If self-service, your help centre and AI chatbot together, resolves 410 of them before anyone reaches an agent, divide 410 by 500 and multiply by 100. The deflection rate is 82%. Read the other way, that means roughly four in five questions were answered without taking up an agent's time.
Only count genuine help-seeking interactions, so the denominator reflects real demand rather than accidental clicks. A rising deflection rate is closely tied to ticket deflection and the quality of your knowledge base: the more complete and current your content, the more questions self-service can resolve on its own.
What's a good deflection rate?
There is no single benchmark that fits every team. A good deflection rate varies by industry, the mix of questions you receive, and how well your knowledge base is maintained. A team with thin or outdated documentation will deflect far less than one with broad, well-kept content and a capable AI chatbot answering from it.
Because of that, the most useful comparison is your own trend over time, not someone else's headline figure. Watch the rate climb as you close content gaps and keep answers current.
It is also worth remembering that a high deflection rate is only good if customers are genuinely helped. If people give up rather than get an answer, the number can look healthy while satisfaction falls. Track deflection rate alongside customer satisfaction, so you are measuring resolved questions, not abandoned ones.
Deflection Rate = (Self-Service Resolutions / Total Help-Seeking Interactions) x 100
Why it matters
Example
A SaaS company adds an AI chatbot trained on its help centre. Of 5,000 questions a month, around 4,100 are answered instantly, from password resets to billing dates, without reaching an agent. That works out to a deflection rate of 82%.
How Resolve247 helps
Raise your deflection rate with AIChatbot
Resolve247's AIChatbot handles 82% of customer questions on its own, trained only on your knowledge base, with an anti-hallucination guarantee so it never makes answers up. When a question needs a person, it hands over to your team with full context.
30 day free trial, no cc required!
Related terms
Frequently asked questions
How is deflection rate calculated?
Divide the number of interactions resolved by self-service by the total number of people who sought help, then multiply by 100. Most teams measure it monthly and break it down by channel to see where self-service performs best.
What is a good deflection rate?
It varies by industry, question mix, and how well your knowledge base is maintained. Rather than chase a universal target, track your own rate over time and pair it with customer satisfaction, so you know questions are being answered well, not just deflected.
How do you improve your deflection rate?
Improve the content customers actually search for, keep your knowledge base current, and add an AI chatbot that answers from that content instantly. Reviewing the questions that still reach an agent shows you the gaps to fill next.
How does deflection rate differ from containment rate?
Deflection rate counts any query answered by self-service before it reaches an agent. Containment rate is narrower: it measures whether an interaction is resolved within a single channel, such as a chatbot session, without transferring out.