Ticket Triage: How to Automate Triage With AI

In this article, we'll discuss the benefits intelligent ticket triage, how to set it up, and how other companies have set up successful automated ticket triage processes.

What is ticket triage? And should you automate it?

Triage (or “routing”) is the process of evaluating and assigning the inbound customer support requests to an individual, team or queue location.

The evaluation stage involves looking at the topic of the request and giving the request a topic tag.

The tag triggers that ticket to be routed or triaged to the right help desk queue, where it’s hopefully handled with the appropriate speed and urgency.

How to set up your ticket triage process

Here are 5 steps to consider when setting up your ticket triage system:

1/ Identify your goals

First decide what your priorities are. The goals you set now will impact your entire triage workflow, here are some examples to consider:

• I want to replace the time my agents spend reviewing our support inbox with an automated system so agents can focus on giving great service.
• I want to reduce first reply time to customers in “X” situation, because our CSAT surveys show they are most heavily impacted (this what James Villas did).
• I want to prioritize customers who are already subscribed to our newsletter, because we know they’re more likely to purchase (this is what Organic Basics do).

Ask yourself:

• What issues are most urgent to solve?
• How can customer needs be better met with routing?

Whatever the issue you’re facing, or goal you want to achieve, list your ideas and how ticket triage might help.

2/ Create a tagging taxonomy

Based on your goals, you'll need to build a tagging taxonomy.

For example, if your plan is to triage "really angry" customers to a high priority queue, you may want to implement a sentiment-based tagging taxonomy.

If your plan is to prioritized customers based on the topic of their request, you'll want to list those topics and common identifiers, then build a clear tagging taxonomy for your agents/automation to follow.

Read more in our guide to tagging taxonomy best practices.

3/ Decide what type of triage to setup

There are three types of ticket triage:

• Manual - An agent reads every ticket, assigns it a tag and routes it.
• Rule-based - Simple tags are applied and the ticket is triaged based on simple automation rules you set up.
• AI-powered - Intelligent tags are applied and tickets are routed with a high degree of accuracy based on sentiment or topic at a granular level.

AI-powered triage leads to the most accurate outcomes, but you'll want to balance that with your budgetary requirements.

Here’s how to setup all three:

4/ How to setup manual triage

If you have a handful of tickets, consider dedicating this task to one agent per session.

Your agent must:

• Read every ticket
• Assign them based

They’ll read and assign new issues as they come in, so the rest of your team knows where to focus.

Top tip: Don’t ask customers to assign their own severity level. 

“One of the things most companies get wrong in their customer service analytics system is letting customer's self-report issues on forms. It causes inherent distrust in any subsequent analysis—support managers hesitate to share insights, other teams question the validity of them, and the self-tagging is too broad or inaccurate to be used to automate other processes like triage.”—Kirsty Pinner, Head of Product

5/ How to setup rule-based triage

• First decide what factors will cause triage to trigger.
• Will it be the email address? Channel it came in on? SLAs (e.g. it has gone unanswered for too long)? Topic? Sentiment?
• Setup rule-based tagging
• Implement routing based on those tags

Using Zendesk? Here’s a short guide from them on setting up rule-based ticket routing.

6/ How to setup AI-based triage

The most accurate triage system leverages AI tags. AI is the best tool to categorize support tickets in real-time without the support agent workload.

• Decide what tags you want to apply (e.g. “severe sentiment” or “X topic”)
• Implement routing rules
• Monitor results

Sentisum integrates with most help desks to automate triage, and the platform is fully flexible to your needs.

Let's look at how other companies have implemented triage into their workflows.

Ticket Triage Example 1: How James Villas Used Ticket Triage to Reduce Response Time

In this interview, Johannes from James Villas. James Villas is a holiday rental company with thousands of hotels all over the world.

James Villas used AI-based ticket triage automation to handle a specific issue their customers were facing during the COVID-19 pandemic.

Hear from Johannes directly:

“Due to Covid-19 the travel industry had an influx in support tickets that made the management of our support operations extremely challenging. The team struggled with customers reaching out due to uncertainty, frustration, as well as necessary cancellations and travel date changes.

‍The challenge was to prioritize tickets in a way, so that urgent cases were handled as quick as possible so that our customers could travel at ease.

‍Unfortunately that was not always possible due to the sheer ticket volume, as well as complexity of the cases.”

In this example, Johannes took a topic that was particularly important—customers in high-frustration situations—and leveraged AI-tagging to automatically identify them in the queue, pull them out, and send them to the priority queue.

The result? James Villas reduced reply time to urgent requests by 46% points and increased CSAT +11%. 

Another benefit mentioned by Johannes was that this freed up his customer service agents time to focus on what they do best: being human and empathising with customers.

Ticket Triage Example 2: How GoCardless use ticket triage to scale their customer support function

In a recent interview with Dimple Pattani, Head of EMEA Customer Support at GoCardless, she detailed how they managed to scale the support function alongside the rapid growth of GoCardless.

One of her key strategies was setting up a ticket triage function. As a company with 30,000+ customers, Dimple’s team deals with thousands of requests a day from customers.

They leverage ticket classification and automated triage to prioritize customer experience above all else.

At GoCardless, all support tickets come to one central location. A duo of dedicated agents then assesses and classifies each ticket to determine severity levels, applying the tags:

• B1: If the severity (or difficulty level) is such that the customer must be guided by a human agent.
• B2: If a knowledgebase article will likely solve the problem, but some hand holding might be needed.
• B3: If no human intervention is needed, simply sending the right article will solve the problem.

Based on these classifications, support tickets are triaged to different severity queues. The benefits are felt on both sides.

For the company, automation can handle the repetitive, easy to resolve requests through self-serve. Agents are unlocked to focus on more interesting, complex requests.

For customers, a fast response is delivered by the appropriate mechanism. No more irrelevant automated links to an article.

Dimple’s team could take this entire process one step further by automating the severity classification process—saving the time of two full-time employees.

Ticket Triage - FAQs

1. Why is ticket triage important?

Having a ticket triage process in place can lead to seriously valuable impact. It depends on the severity of the issue you’re currently facing and/ or how you choose to leverage ticket triage to drive results.

There are several high impact ways to prioritize your support requests with triage. They include:

• Prioritizing customers currently in the process of purchasing your product. Grab ‘em before they abandon the basket.
• Prioritizing customer requests that contain ‘danger topics’ like a customer with food poisoning (for Uber Eats). Get to these before they become a PR story.
• Prioritizing particularly angry customer requests. Negative sentiment will be dampened by a fast response, or heightened by a slow one. Get to these before they become a bad review.

On the other hand, you could want to build a customer service team that specializes in certain product areas or issues. Triaging tickets to the right team gives every customer a faster and more skilled resolution.

2. Why am I triaging tickets in the first place? What criteria routes tickets to which team?

Most customer service teams triage in order to prioritize them.

Gone are the days of first come; first serve in customer service. With smart triage, you can prioritize requests in a way that makes sense for both the customer and company.

Perhaps some requests are full of angry sentiment, or perhaps they’re about a topic you want to handle really quickly—e.g. A customer is at the airport right now and needs something urgently. These criteria, paired with a dedicated ‘urgent’ inbox that is handled first, could have a huge impact on customer satisfaction.

3. Who triages the tickets?

Especially if you’re a large B2C company, at this point you might be thinking, “we have thousands of requests a day, who’s gonna evaluate all those tickets?”

Well, there are a few ways customer service teams tackle this problem.

1. They pay two agents full-time to review each and every request. It’s time-consuming and boring, but the benefits are seen as worth it.
2. They have the customer self-tag their request with a topic or urgency, based on this information the ticket is routed automatically. This is less time-consuming but is largely inaccurate and only allows triage based on topic. Zendesk has a guide on setting up automated triage in their platform here—but the feedback we hear again and again is how awful it is at the job (“inaccurate, inconsistent, painful”).
3. They’ve automated ticket triage and prioritization with AI. AI works in real-time to categorize sentiment, intent and granular topics in incoming requests, which routes the ticket to any individual or team you’d like.

As an extremely repetitive, yet important task, we highly recommend automating the process with an AI triage tool for the most accurate, consistent and speedy results.