Finding the Most Efficient Path: Using Ticket Deflection and Escalation in Practice

In a customer service ecosystem with multiple channels — phone support, email, webchat, in-app messaging, and more — a customer question can come in from many different touch points and take various paths from there. In order to achieve maximum efficiency from a time and cost perspective for the support team, however, there should be an overarching strategy to both improve ticket deflection and ensure that tickets are escalated to agents when necessary. 

Read on (you can click to expand the infographic on the right) and gain an understanding of how to first route your tickets through AI and automation to offer the most efficient support for your customers.

1. Improve Ticket Deflection with AI 

Typically, an incoming customer query is first routed to some feature that is designed to deflect tickets. This can be an QuickSearch Bot, for example, that suggests knowledge articles with the hope that the customer will then be able to self serve. 

Right off the bat, this can improve ticket deflection — so, even before a ticket is created, many customer issues can be resolved.

If these knowledge articles are not adequate, though, an AI-powered classification engine can take over to determine the issue type, label the conversation, prompt the customer for additional information and then route the ticket accordingly. 

Improve the path of a question through ticket deflection

2. Open the Ticket and Route Accordingly

Initially, the AI-based classification engine is trained through a process based on supervised learning aided with NLP algorithms. Sample issues from the existing ticketing system are fed to the engine during this training. The more practice the classification engine gets, the more accurate it becomes at classifying customer issues. Over time, the classification engine becomes fully versed in the specific ticket types of the organization it is being used by.

Once categorized, issues may be separated based on urgency or VIP status (some of these may get immediately routed to an available agent). Importantly, many non-VIP issues can be resolved fully by bot-based automation. For example, a bot can gather the customer’s order information, access the shipment database, track a package and let the customer know when the package is due to arrive. 

Even customer issues that cannot be resolved by automation alone can still benefit from first interacting with a bot, as bots can collect routine information that eventually aid agents in resolving issues faster. Using automation in these ways massively reduces ticket backlog for agents. 

However, some customer issues will always require human attention. 

3. Create a Seamless ‘Automation to Agent’ Handoff

Once a bot can no longer be helpful, the issue should be handed off to an agent to avoid any customer frustration. Urgent issues should be routed to a live queue, while non-urgent issues may take a little longer for the agent to respond. And that’s okay — it’s important to point out that the customer will not be waiting on hold, because if they are directed towards messaging based-interactions, the customer can leave the conversation and return when an agent is available without losing any context. 

If the messaging-based agent still isn’t able to resolve the issue, the ticket may need to be routed to a phone representative who is trained in empathy and can assist quickly and with nuance. 

Often, customer issues are resolved using a combination of agents and chatbots, sometimes across multiple channels. It’s important to develop workflows that both improve ticket deflection, increase rates of automation — while also escalating those issues to a phone call as soon as it becomes necessary.

In fact, this is the next frontier of customer service: the ability to have conversations that transcend a single channel and happen continuously across mediums. This is what Helpshift refers to as Connected Customer Conversations because it illustrates how effective conversations begin, unfold and conclude: sometimes quickly and autonomously, but other times, asynchronously with human agents and over time.

For an in-depth, complete step-by-step breakdown to modernizing your support organization —  go ahead and download The Ultimate Guide to Digital-First Customer Service.

Similar Posts