Welcome to the Customer Service Management app
The future of customer service
The future of customer service is no longer just about resolving tickets quickly. It's about providing exceptional end-to-end customer experiences that becomes a net positive for the entire business. As a result, world class organizations are already adapting to leverage the latest technologies and deliver this net positive effect.
We built Customer Service Management for this exact reason. In order for customer support teams to do their best possible work, they need to have a clear picture of what’s happening in their organization and stay connected to development, operations, and product. Atlassian brings your customer support team on the same platform to the teams building and running their products and surfaces critical context with the Teamwork Graph. This is especially valuable in an AI world, as we can help you build human-AI collaborative teams, not take the human out of the loop.
In order to do this, there are three key roles that should matter to your organization.
Help-seeker
The help-seeker is your external customer who is reaching out for support. They interact with your organization through your support channels. The goal is to deliver them a seamless, fast, and personalized experience so that they get the help they need, when they need it.
Support agent
The support agent is the person who is working to resolve customer requests when they come in. They work with the relevant stakeholders to set the fastest and best path to resolution. In this new AI world, their goal is to focus on bringing their empathy, expertise, and judgment to the complex requests that AI can’t fully resolve on its own and provide a personalized support experience for customers.
AI support manager
We acknowledge that every organization has a different level of AI readiness and desire to adopt. We encourage teams to explore what is best for them.
Atlassian’s belief is that when AI is incorporated in a way that keeps your support team in the loop, it elevates, not hinders, the overall customer experience. It’s important to understand the AI support manager is not a net new position, but an evolution of how your support teams work in an AI world.
The AI support manager is meant to ensure you keep a human in the loop at all times. They monitor the AI agent's effectiveness, review past conversations to understand the AI's reasoning, provide coaching and feedback to guide future responses, manage agent versions, run tests and evaluations against golden datasets, and use optimization tools to surface improvement opportunities like missing help articles. They ensure the AI agent keeps getting better and consistently meets the support team's standards.
As you continue reading the guide, you’ll learn how to set up Customer Service Management to empower and deliver the best experience for all three of these roles.
Establish your queues and Service Level Agreements (SLAs)
Queues help your team organize and manage work items, giving a focused view of work to make it easier to triage, prioritize, and take action on customer requests. When a customer submits a request, it will come into your queue so you can get to resolving their request quickly.
Service Level Agreements (SLAs) help you track and manage how quickly your team responds to, and resolves, customer requests. SLAs make it easy to set clear expectations, prioritize work, and measure your team’s performance.
Add customer and organization context
When help-seekers reach out, they just want their request resolved quickly. Most often, once they’ve finally gotten to the front of the queue to a support agent, help seekers are met with more frustration because they have to answer a series of questions like what product their request is about, what region they are based in, etc. which only slows down the speed of support.
The more you know your customer, the more you can deliver personalized, faster, and more accurate customer service. That’s why it’s important to build context on your customers.
This can be done two ways, with the first being customer profiles. Customer profiles make it easier to understand who a customer is, what organization they belong to, their product entitlements, and details of any previous interactions. This way, you can get right to the core of the problem rather than needing to waste time gathering the basics each time a customer reaches out.
The second way you can build context is with organization profiles. Organization profiles help your team quickly understand the organization’s history, relationships, and needs, so you can deliver more effective support.
Once you have established customer and organization profiles, this context is available right at your team’s fingertips as it helps support agents understand the complete picture of the support request and who the request came from right away, rather than having to search for it. That means faster, more personalized resolution as you can see things like location, purchase history, past requests, account manager, support tier, number of licenses, and more.
Build your AI agent experience
The customer service AI agent is available for customers on Standard, Premium and Enterprise plans. If you are currently on a Free plan but want to explore the AI agent’s capabilities, we recommend you try a Standard plan free trial.
Customers don’t care how much money AI saves your company. They care about how quickly, accurately, and effortlessly you solve their problems. That’s why the customer service AI agent is built into Customer Service Management and focuses on helping your teams deliver the best possible customer experience. Whether it’s 24/7 self-service or understanding when to bring a human in the loop and providing all the necessary context for fast resolution, you merge human empathy with AI-native speed to set a new standard for customer service.
If you do not feel comfortable deploying the AI agent publicly at this point in time, you do not need to set it up now. It does not go live unless you set up, publish, and add customers that can access it.
The first step to getting your AI agent setup is to define identity, which shapes how your AI agent introduces itself, represents your company, and interacts with customers. From the agent’s name to its tone of communication, the first message the agent sends when starting a new conversation, and suggested questions customers can select to start the conversation, setting identity the way you want ensures the agent is an accurate representative of your customer support team.
Once you’ve established identity, the next step is to connect your knowledge to the agent. Learn how to build your help articles in the “How to deliver exceptional help-seeker experiences with Customer Service Management” section.
Knowledge is a combination of:
- articles you created and appear on your support site
- all other documentation you may have but not want to make customer facing
It is the foundation for how your AI agent responds to customers so you can have confidence that the AI agent is delivering consistent, high-quality support.
Knowledge isn’t the only foundation to help your AI agent respond to customers though. By connecting actions, you can let your AI agent perform specific tasks by interacting with APIs or services. This turns your AI agent from “just an answer bot” into an actual problem-solver because you are extending its capabilities and automating common support tasks. All this contributes to faster resolutions for customers and more time for your team to focus on the tickets that require their attention.
Sometimes, there will be questions that come in that the AI agent might not be equipped to answer based on knowledge or actions alone, or the issue needs to be escalated to your team for their expertise.
With guidance, you can prepare by shaping how your AI agent responds to customers. Tell it exactly what to say or do in specific situations such as asking follow up questions when the incoming request is unclear to deliver accurate, helpful, and consistent support. There are two types of guidance:
- Response guidance: Tells the agent exactly how to reply in a specific situation. You can provide a sample response, and use Markdown formatting to make replies more structured and engaging.
- Handoff guidance: Tells the agent when to escalate a conversation to a human, like when a question is too complex or sensitive.
When a request is handed off per the guidance you have established or when a customer expresses frustration, this is where true AI-human collaboration takes place.
With handoffs, the AI agent is able to seamlessly bring your support agent in when needed with all the necessary context for faster resolution. The AI agent auto-fills out a form based on their chat with the customer, which creates a ticket that is shared with the customer and your support team. As a result, your customers don’t have to go through the frustrating experience of repeating themselves, and your support team spends less time gathering the background info and more time getting the request resolved as quickly as possible.
Before you publish your agent to be available for your customers, you can test the agent anytime you wish. When testing, you are simulating a live conversation with the agent as if you are the customer. As a result, you can ask a range of questions, including both common and unusual, to make sure it’s accurate and helpful when deployed to your customers.
If you’re happy with all the configurations from above, you can now publish your agent and make it available to respond to your customers!
You can publish your agent from any configuration page in the customer service agent’s settings. In the floating bar at the bottom of the screen, select Publish.
Work with Rovo, an AI teammate
We just shared a lot of AI configuration to deliver exceptional customer experiences but don’t worry! We’ve got something for your support teams too.
With Rovo, an AI teammate built into Customer Service Management, you get a 24/7 AI teammate to help you resolve requests faster and more confidently. It:
- Brings the relevant knowledge you need and similar resolved cases right to your fingertips when working on a request, helping reduce the time to resolution
- Drafts updates you can review, edit, and then send to customers to keep them in the loop about progress
- Summarizes progress so you can get up to speed and get teammates in the loop fast when requests are handed off
Connect with the teams building and running your products
While some requests can be fully resolved by frontline support teams, others require escalations and forwarding to subject matter experts. Not to worry, the magic of Atlassian’s platform and the Teamwork Graph enables you to forward work from Customer Service Management directly to the teams building and running your products.
Forward requests your developers' inbox in Jira: Spend less time chasing after the right stakeholders or find that their escalations were slipping through the cracks. Requests land directly where developers work with all the context needed so they can get to resolving fast.
Get insights on your team’s performance
It’s important that you have a clear view of how your support team is performing. That way, you can find opportunities for improvement and ensure you are always showing your best when delivering customer service.
The Customer service overview dashboard gives support managers a high-level view of all customer service activities. It's designed to monitor trends and performance across multiple customer experiences and projects. You can see charts for information like AI agent conversations, resolved work items, median full resolution time, and more.
Empower customers to self-serve with help articles
Help articles are the foundation for your support site. By establishing a robust collection of articles, you can:
- Provide instant self-service: Help customers resolve common issues on their own, 24/7, without waiting for a human agent to answer their questions.
- Create a consistent, trustworthy customer experience: Every customer gets the same, accurate guidance instead of a variable answer depending on who works on the ticket.
- Keep your team focused on complex cases: Reduce the volume of repetitive, common requests coming to your team so they can focus on the complex, high-value cases that require their empathy, expertise, and judgment.
Before you can make help articles available for your customers, you will need to have an active Confluence subscription.
Then, you can connect an existing space if you already have written help articles, or create a new space if you’re starting from scratch. You can also connect multiple spaces if needed, and you can always edit articles as needed so you have the most up-to-date guidance available for your customers.
Meet customers where they are for instant help
When customers need help, they just want it fast. Below is an overview of the channels you can leverage with Customer Service Management to make getting help as easy and accessible as possible:
Customizable support site
Your support site is the front door to your support experience, where customers can search for and read help articles to self-serve answers to their questions or submit requests that require your team’s expertise. Customize the support site to meet your brand and deliver a consistent support experience.
Allow customers to reach you via email. You can create a new email address or use an email address you already own. Once established, requests sent to the designated email address automatically convert into work items for your team so you don’t have to worry about missing requests or managing multiple inboxes.
Embedded AI chat widget
Customers can interact with an AI agent through a chat interface embedded in channels like on your website, support site, or in-product to get immediate answers to common questions any time of day. The agent uses your connected support content and when that isn’t enough, it gathers all the necessary context to escalate and bring your support team in the loop.
Voice
Give customers a direct line to your team with the Amazon Connect voice integration. Your team can receive and manage customer calls, and automatically create work items with call transcripts for follow-up.
While having a breadth of channels is important, it’s also important to deliver a unified, intelligent, and smooth experience where customers don’t have to repeat themselves. With Customer Service Management, these channels and the context gathered stay connected. A customer could start on the phone and move to the embedded AI chat widget on your support site but continue the conversation from where they left off, all without having to start over and repeat themselves.
Make escalation seamless
When your team is needed on a request, it’s important that they have all the context they need to jump in, get to resolving the issue fast, and most importantly not make the customer repeat themselves.
That’s where forms come in, which allow you to receive requests from customers and collect details in a structured format. With forms, you can tailor them to a specific type of inquiry, such as product support, billing, or feedback. They live on the contact page of your support website where customers can select and submit forms to request help, provide feedback, or make other inquiries.
Depending on your handoff settings outlined in the previous section, the AI agent can also direct a customer to a contact form or fill out and submit a form on the customer’s behalf.
How to empower an AI support manager with Customer Service Management
Think of the AI manager like a farmer growing crops. They can’t just plant the seeds in soil and expect that they will grow perfectly in time for harvest. The farmer will have to provide continuous, consistent care and attention, from making sure there is enough sunlight to consistently watering the seeds, in order to harbor proper growth.
Delivering an AI-first customer support experience with an AI agent is exactly the same. It isn’t “set and forget” once you have deployed it to your customers. You have to continuously provide guidance and manage the experience so the AI agent never stops improving and you are always delivering a great customer experience.
Get a clear picture of the AI agent’s performance
Just like how it is important to keep a pulse check on how your support team is performing, it’s equally as important to understand how your AI agent is performing and if it is actually effective in supporting your customers.
As the AI support manager, you should look to own metrics like the number of conversations, AI containment (the percentage of conversations resolved without human intervention), and resolution rate. In Customer Service Management, you’ll see key metrics reflected across a series of charts to help you understand performance and identify areas of improvement.
Help your AI agent get better with every conversation
Similar to how you train your support engineers by providing coaching and guidance, you can do the same with the AI agent. That way, it keeps you in the driver’s seat because you have oversight on the AI agent and can dictate improvement in a way that ensures the AI agent meets your support team’s standards.
With conversation review, you can:
- See into all conversations: Filter and search for specific conversations and see transcripts to understand each interaction.
- Understand the AI agent’s behavior: You’ll get insights into the logic of why the AI agent responded the way it did, from what it understood from the help seeker, the actions that the AI agent performed, to the content that the AI agent pulled back to generate its response.
- Provide coaching: Rate the conversation and provide feedback on the responses so you can assess conversation quality, guide your AI agent’s future responses, and improve future performance.
Have confidence your AI agent improvement efforts pay off
While you are able to review past conversations and provide guidance, it raises an interesting question. How do you know that the AI agent is actually getting better? How do you keep an eye on the progress over time, and that the changes you’re making are actually working?
First, it’s important to keep a record of every time you make a change to your agent. Whether that’s Identity, Knowledge, Guidance, Actions, or Handoff, versioning lets you safely update the agent configurations the way you want. And you can have confidence you can review and refine changes until you choose to publish it as only then will the latest version go live for your customers.
Once you have made any changes, it’s important to test the latest version of your agent. Testing isn’t just for when you are first configuring your agent. You can test at any time to make sure you are deploying an agent experience that you feel confident in.
When you want a bit more than just simulating a live conversation experience though, you can also run evaluations. Evaluations test how your agent responds to a variety of questions and improve its effectiveness. A large language model (LLM) judge will review the responses, determine whether each question was answered satisfactorily, and provide an overall resolution rate. This is critical to AI agent improvement because you can:
- Measure against a golden dataset: Evaluations require uploading a golden dataset, a set of questions created for testing your customer service agent’s responses. This question list can be made up of common questions you receive, key areas the agent will need to help customers with, etc. You can upload and see how the agent would respond to up to 50 questions at once. Note that this is different from the testing function, which simulates a live conversation with the agent.
- Confirm changes are working: You may have made changes to your agent, such as adding knowledge sources or guidance. With evaluations, you are able to see how different versions of agents based on your changes respond to questions and you can compare results accordingly.
And all of this is why we call it continuous improvement! By looking over the AI agent’s past conversations, creating versions, running tests, and running evaluations based on your agent’s previous behavior, you compound and build on previous enhancements.