"We deflect 70% of tickets." It sounds impressive. But what does it actually mean?

In the support technology space, deflection and resolution are often used interchangeably — by vendors, by analysts, and sometimes by support leaders themselves. They are not the same thing. The distinction has significant implications for how you measure your support operation, what tools you buy, and what results you can realistically expect.

Defining the Terms

Ticket Deflection

Deflection occurs when a customer is prevented from submitting a ticket. The customer had a question or problem, started to submit a ticket, and was redirected to a self-service resource — a knowledge base article, an FAQ, a chatbot conversation — that resolved their issue before the ticket was created.

Key characteristic: No ticket was submitted. The deflection happened pre-submission.

How it is measured: Deflection rate = (customers who found self-service answer / customers who started ticket submission process) × 100.

The problem with this metric: It is very difficult to measure accurately. You can count how many customers clicked "this article helped" in your help centre. You cannot count how many customers found the answer and left without clicking anything. Deflection rates are often overstated because the denominator (customers who would have submitted a ticket) is not directly observable.

Ticket Resolution

Resolution occurs when a submitted ticket is answered correctly and the customer's issue is addressed. The ticket was created, it was processed, and it was resolved.

Key characteristic: A ticket was submitted and resolved. The resolution happened post-submission.

How it is measured: Resolution rate = (tickets resolved without escalation / total tickets submitted) × 100. For AI resolution specifically: automation rate = (tickets resolved by AI without human involvement / total tickets submitted) × 100.

Why this metric is more reliable: You know exactly how many tickets were submitted. You know exactly how many were resolved by the AI vs by a human agent. The denominator is not estimated — it is a count.

Why Vendors Conflate Them

Deflection numbers are easier to make look impressive. If your chatbot shows a knowledge base article to 1,000 customers and 300 of them click "this helped," you can claim 30% deflection. But you have no idea whether those 300 customers would have submitted a ticket if the chatbot had not intervened — some of them were just browsing.

Resolution numbers are harder to inflate. If the AI resolved 600 out of 1,000 submitted tickets, the automation rate is 60%. That is a real number with a real denominator.

Vendors who lead with deflection rates are often selling chatbots or self-service tools. Vendors who lead with resolution rates are selling AI ticket resolution engines. The two categories do different things.

The Business Value of Each

The Value of Deflection

Deflection reduces inbound ticket volume. Fewer tickets submitted means fewer tickets for agents to handle. This is genuinely valuable — but only if the deflected customers actually got their issue resolved.

A customer who was deflected to a knowledge base article that did not answer their question is not a success. They will either submit a ticket anyway (making the deflection rate misleading) or give up and churn (making the deflection rate actively harmful).

Deflection is valuable when: The self-service resource genuinely answers the customer's question, and the customer is satisfied with the self-service experience.

Deflection is harmful when: The customer is redirected to an unhelpful resource, cannot find the answer, and either submits a ticket anyway (frustrated) or gives up (churned).

The Value of Resolution

Resolution addresses tickets that have already been submitted. The customer has already decided to contact support — the question is how quickly and accurately their issue gets resolved.

AI ticket resolution delivers value by:

• Resolving L1 tickets in seconds rather than hours
• Reducing the volume of tickets reaching human agents
• Improving consistency — the AI always checks the knowledge base before responding
• Scaling without headcount — the AI handles 10x the volume without additional cost

Resolution is valuable when: The AI resolves tickets accurately and quickly, improving customer experience and reducing agent workload.

Resolution is limited when: The ticket requires human judgement, system access, or information the AI does not have.

How to Measure the Right Metric

If You Are Evaluating Deflection Tools

Ask vendors:

• How do you measure deflection rate? What is the denominator?
• What percentage of deflected customers submit a ticket anyway?
• What is the CSAT score for customers who were deflected vs customers who submitted a ticket?
• How do you distinguish genuine deflection from customers who would not have submitted a ticket regardless?

If You Are Evaluating Resolution Tools

Ask vendors:

• What is your average automation rate across customers? (Ask for median, not maximum.)
• How is automation rate calculated? (Tickets resolved by AI / total tickets submitted?)
• What is the re-open rate on AI-resolved tickets?
• What is the CSAT score on AI-resolved tickets vs human-resolved tickets?

The Metric That Matters Most

For most support operations, the metric that matters most is total cost per resolved ticket — the fully loaded cost of resolving a ticket, including agent time, tooling, and overhead.

AI ticket resolution reduces this metric by resolving a large proportion of tickets at near-zero marginal cost. Deflection tools reduce it by preventing some tickets from being submitted — but the savings are harder to measure and the risk of customer dissatisfaction is higher.

A Practical Framework

When evaluating support automation tools, use this framework:

Question	Deflection Tool	Resolution Tool
When does it act?	Before ticket submission	After ticket submission
What does it do?	Redirects to self-service	Resolves the submitted ticket
How is success measured?	Deflection rate (hard to measure accurately)	Automation rate (easy to measure accurately)
What is the failure mode?	Customer not helped, churns or submits anyway	Customer receives incorrect AI response
Best for	High chat volume, strong knowledge base	High email volume, repetitive L1 tickets

Why TR-AI Focuses on Resolution, Not Deflection

TR-AI is a ticket resolution engine. It does not intercept customers before they submit a ticket — it resolves tickets after they are submitted.

This is a deliberate choice. Deflection tools work well for customers who prefer self-service and are willing to engage with a chatbot. But a large proportion of support customers prefer email, submit tickets outside business hours, or have already tried self-service and failed.

For these customers, deflection is not an option. Resolution is the only path to a good experience.

TR-AI resolves L1 tickets within seconds of submission, regardless of channel or time of day. The automation rate is measurable, the quality is auditable, and the impact on agent workload is immediate.

If you want to see what a resolution rate looks like on your actual ticket data, start a free trial — no credit card required.