What is ticket deflection? Formula, worked example and honest measurement

What is ticket deflection?

Ticket deflection is a support strategy where customers resolve their own issues through customer self-service — FAQ pages, knowledge base articles, chatbots or AI answers — before a ticket is created. Zendesk defines it as reducing the number of support tickets by providing self-service resources such as AI-powered chatbots, community forums, FAQ pages and knowledge bases. The unit that matters is the support need: one question, one problem, one intent. If it gets answered without an agent ever touching it, it was deflected.

The same idea goes by other names: case deflection, call deflection when the avoided contact is a phone call, or simply support deflection. Teams track it because a deflected need never enters the queue: no handling time, no wait for the customer, no agent cost. The catch is that "no ticket" is not the same as "problem solved" — more on that below.

The ticket deflection rate formula, with a worked example

The most common form counts intents:

Ticket deflection rate = intents resolved without a ticket ÷ total support intents × 100%

Vendors count differently, so check the denominator before comparing numbers:

• Freshchat's bot KPI guide illustrates deflection with ticket volume: 1,000 agent-handled tickets per month drops to 500 after a bot launch, a 50% deflection rate.
• Intercom's Fin metrics are conversation-based: the resolution rate is the share of Fin-involved conversations that Fin resolved.
• Zendesk's ticket deflection rate (its "self-service score") is a ratio, not a percentage: total help center users ÷ total users in tickets. Four self-service users per ticket user scores 4.

Worked example — sample data. A website logs 1,000 support intents in one month. FAQ, knowledge base and AI answers resolve 420 of them without a conversation ever being created. The rest become tickets.

42% is the headline number. Before you report it, run the two checks in the next section.

Deflected is not resolved: keep the number honest

A visitor who reads an answer, gives up quietly and leaves still counts as "no ticket". That is false deflection, and it inflates the rate. Two checks keep the number honest.

Repeat-contact check. Fix a look-back window — seven days in this example — and find deflected visitors who come back and open a ticket about the same issue. In the sample data, 63 of the 420 do:

Confirmed deflection = (420 − 63) ÷ 1,000 = 357 ÷ 1,000 = 35.7%

The headline 42% was inflated by 6.3 percentage points. Match on the issue, not just the visitor: a return about a different problem is a new intent, not a failed deflection.

CSAT spot-check. Survey a sample of deflected visitors right after the self-service answer. Sample data: 120 respond and 102 say the issue was solved to their satisfaction — 102 ÷ 120 = 85%. A rising deflection rate with falling satisfaction means people are stuck, not helped.

Keep the window and the denominator fixed from month to month, and report the headline and confirmed rates together so the gap stays visible.

How to raise ticket deflection

Four levers, roughly in order of effort:

1. Build the knowledge base from real tickets. Mine the queue for recurring questions and write one clear answer per question. Deflection only works when the content matches what customers actually ask; stale or generic articles send people straight back to the queue.
2. Pin an FAQ for the top questions. A short, visible FAQ on the site and in the chat widget catches the handful of questions that make up most of the volume — pricing, shipping, password resets, business hours.
3. Add AI answers grounded in that content. An AI layer can match a visitor's own wording to the right answer and handle follow-up questions. Keep a human handoff for anything it cannot answer confidently; see AI support agents for what agents can and cannot do, types of chatbots for rule-based versus AI options, and set up AI chat replies for the setup steps.
4. Suggest articles before submission. When someone starts typing in a contact or ticket form, match the text against the knowledge base and show likely answers before the form is sent. This is the last chance to deflect a need that already reached the form.

Keep the path to a human visible at every step — deflection that traps people produces false deflection, not savings. And for the tickets that still arrive, canned responses cut the handling time per ticket.

Deflection vs self-service rate vs first-contact resolution

These three metrics answer different questions and use different denominators. The table computes all of them on the same sample dataset: the same month, 1,000 intents, 700 of which touched self-service, 420 resolved without a ticket, 580 tickets created, 406 of those tickets resolved in a single touch.

Usage is not outcome: 70% of intents touched self-service, but only 42% avoided a ticket. FCR is the agent-side mirror image — Zendesk defines it as one-touch tickets ÷ total tickets received × 100 — and its denominator is tickets, not intents. Never compare rates computed on different denominators or different datasets.

How to see deflection in TapCub

TapCub has no report named "deflection rate", but the boundary this page is about, resolved before a conversation exists, is built into how its help widget works.

• Pinned Q&A is the deflection surface. In Live chat › Knowledge base, pin your most-asked Q&A entries. They show in the visitor-side FAQ, and opening an answer creates no conversation and uses no quota. In the worked example, these are the self-service resolutions.
• "Not solved" is the boundary. Clicking Not solved or asking for a person creates a conversation and brings the question to an agent in Live chat › Inbox. That click is the moment an intent stops being deflected and becomes work.
• Watch the proxies together. Live chat › Reports shows the knowledge base hit rate (questions answered by an exact Q&A match or a retrieved answer), resolution rate and satisfaction. A rising hit rate with falling satisfaction is the "deflected but not resolved" warning from the section above.
• Let AI answer first, honestly. In Live chat › Chat settings › AI assistant, AI-first mode answers from the knowledge base and hands off when the material is not enough, instead of guessing. Setup: set up AI chat replies.

Sources

• Zendesk, Ticket deflection: Enhance your self-service with AI.
• Intercom Help, Update to Fin performance metrics.
• Freshchat Support, Measuring KPIs for your bots in Freshchat.
• Zendesk, What is first contact resolution (FCR)? Benefits + best practices.