The formula for ticket deflection rate is one line of arithmetic. The part that takes judgment is deciding what actually counts as a deflected customer, and getting that wrong produces a number that looks good on a dashboard while hiding a real problem.
Here's the calculation, a worked example, and the specific mistake that inflates the metric.
Direct answer
Ticket deflection rate is calculated as: (self-service sessions that resolved the issue without a ticket) ÷ (self-service sessions + tickets filed on that topic) × 100. Worked example: if your help center or widget handled 400 questions on a topic and 40 of those still became tickets, deflection rate is 360 ÷ 400 = 90%. The measurement mistake to avoid: counting every self-service session as "deflected" without confirming the customer's problem was actually solved, since a customer who reads an article, doesn't find their answer, and gives up without filing a ticket looks identical to a successfully deflected one in raw session counts, but represents a failure, not a win.
The formula, worked
- Count self-service sessions on the topic: widget conversations or help center visits that addressed a specific question.
- Count tickets filed on the same topic: support tickets that reached a human despite self-service being available.
- Apply the formula: deflected = self-service sessions − tickets on that topic; rate = deflected ÷ total sessions × 100.
- Example: 400 total sessions on a topic, 40 became tickets → 360 deflected ÷ 400 = 90% deflection rate.
The measurement mistake that inflates the number
Raw session counts can't distinguish a customer who found their answer from one who gave up and left. If you count every non-ticket session as a success, a confusing or outdated article that causes customers to silently abandon looks identical to one that actually helped them, both show up as "deflected."
Frequently asked questions
What's a good ticket deflection rate?
It varies heavily by product and topic complexity, but the number matters less than whether it's measured honestly. A high deflection rate built on silent-abandonment sessions is worse than a lower, honestly-measured one.
Does deflection rate account for customers who gave up without an answer?
Not by default. Raw session counts treat a silent abandonment the same as a successful self-serve resolution, which is why pairing the rate with a direct feedback signal matters.
Should deflection rate be measured per article or overall?
Per topic or per article is more actionable, an overall site-wide number hides which specific articles are actually solving problems versus which ones customers bounce from.
Conclusion
- Deflection rate = deflected sessions ÷ total sessions × 100, a simple formula.
- The real measurement risk is counting silent abandonment as a success, both look the same in raw session counts.
- Pair the rate with a direct feedback signal to catch what raw counts miss.
- See what is ticket deflection for the full concept this metric measures.