When it comes to customer experience, speed and accuracy are critical elements. Customers and internal employees need access to the information they’re looking for quickly, or your company might have lost a sale — or frustrated a customer.
Delivering that fast, accurate customer experience can be even more difficult when you’re an international company with a large catalog of digital solutions and services. That’s why Atlassian, the global software company with more than 9,000 employees that makes software to help teams organize, discuss, and complete their work, turned to process orchestration to hyperautomate their tech stack and deliver improved customer and employee experiences to support Sales and Finance-related activities.
Meet Suzie: Atlassian’s savvy support bot
Suzie is Atlassian’s ticket-support bot, which connects different endpoints across their sales and finance systems and orchestrates bots from various robotic process automation (RPA) vendors. A few of the processes and activities supported by Suzie include:
- Policy configuration for business users
- Atlassian product quoting process
- End-to-End subscription process management
- Servicing customer subscriptions to Atlassian products
- Auditing and licensing reporting
Today’s Suzie is a powerful bot, connecting disparate systems and endpoints, including humans, into a tech stack that has helped reduce ticket turnaround times by 93% in some cases, from an average 45-60 minute response time down to two minutes. Suzie’s automations have also yielded an almost 30x return on automation spend, driving cost savings on manual development. However, Suzie wasn’t always so efficient.
The Challenges of Relying on RPA
“Initially, Suzie was completely RPA-based,” explained Brajesh Bharti, senior engineer, Atlassian.
Trying to automate complex processes with a diversity of endpoints solely with RPA introduced a number of challenges into the system that Bharti and his team needed to solve.
Firstly, the out-of-the-box orchestration layer for RPA isn’t mature enough for Atlassian’s needs. Since these processes were being implemented for business users or end-customers, this orchestration immaturity meant these non-technical personas didn’t have the ability to visualize the workflow end-to-end.
Another big hurdle to overcome is that the initial RPA-based solution was implemented using scheduled, time-based triggers instead of real-time processing. This hampered real-time processes within Suzie and built natural delays into the automated workflows.
“We were not able to provide a quick response to our customers, because we originally had scheduled pull-based triggers that ran every hour, as requested by our business users,” Bharti elaborated.
Additionally, RPA is not suitable for long-running processes, or complex processes that require human interaction.
“One of our use cases for Suzie is generating price quotes,” said Bharti. “This process requires some level of manual approval from a human. That approval needs to go to different business departments, and different users.”
Those requests will take time to be approved, delegated to another human, or require adjusting. The old process built as it was would hit delays at this stage.
The old system further limited the flexibility Atlassian desired, as they did not implement Decision Model and Notation (DMN). With DMN, you can configure the policy on the fly, without relying solely on the developer. Process orchestration that leverages DMN delivers a level of flexibility and agility that RPA does not.
All these challenges added up to a need for a process orchestrator that could help connect the diverse endpoints within Atlassian’s processes, streamline the workflow, and improve the experience for customers, prospects, and Atlassian employees alike.
Supercharging Suzie with Camunda
Atlassian used Camunda to solve these challenges and maximize Suzie’s efficiency and ability to provide superior customer experiences.
Workflow process orchestration layer
First, Camunda provides the orchestration layer for the workflow. Now it’s very easy to track the workflow from end-to-end, and both IT and business users at Atlassian can visualize the workflow, the events, and where the system may be impacted upstream or downstream. It’s also easy for the IT team to define the orchestration layer for the workflow.
Real-Time responses powered by a flexible architecture
Now that Atlassian has moved their bots under a common orchestration with Camunda, they can fully take advantage of real-time processing. Camunda provides API-based, subscription-based, and configurable real-time processing for end-to-end processes.
Camunda also allows for a flexible architecture that enables integration with other systems; in Atlassian’s case, that means they can integrate with platforms, including an enterprise service bus (ESB) (MuleSoft) and different Oracle and NetSuite pieces, Workato for API connectivity, and UiPath for RPA – with Camunda at the heart of their tech stacks’ automation fabric. It also accommodates developer-friendly elements, such as DMN and BPMN that can easily change on the fly, and empowers developers to use the tech stack they are most comfortable with, including Java, React, Node.js, or Python.
For example, Atlassian established an AWS SQL database to keep track of their subscriptions. Suzie was able to consume all of the tickets created in the Atlassian support system which relied on that database by using an event trigger.
Atlassian then configured the DMN table on the basis of the business criteria, and Suzie routes to the next activity based on the valid business rule for the event which has happened.
Suzie no longer has to wait for scheduled pulls, because the workflow now uses asynchronous events which can be triggered if something changes in the context of outside systems.
This solves the challenges of pull-based schedules RPAs are traditionally restricted to, and allows for better visualization of the end-to-end processes.
Better process visualization offers optimization opportunities
Camunda’s Optimize component helped deliver end-to-end business process visualization for Suzie’s ticketing support system. Optimize delivers powerful insights about process performance and usage to help business and IT users continuously improve their systems.
Atlassian is now able to see how many tickets are processed by Suzie on a monthly, quarterly, or annual basis, as well as the business and technical exceptions where data can be missed and the ticket can’t be fully processed. This improves Suzie’s management, allows developers to spot areas that need to be addressed to keep the end-to-end process running smoothly, and gives business users insight into the process and its effects for business strategy.
Overall, Camunda has helped Atlassian’s team to more efficiently design, deploy, and manage Suzie’s features to support a number of complex business processes and make Suzie more effective than ever.
The journey to today’s Suzie
Atlassian kicked off the project in November 2019. After evaluating multiple BPM platforms, Atlassian’s platform team decided that Camunda was the best fit for their microservices and tech stack.
Atlassian operationalized their Camunda instance in April 2020, and began to evolve Suzie. They started with automating the subscription management and subscription activation processes, and then moved on to adding multiple capabilities, including price quoting, auditing, adding humans into the loop, and the export control policy configuration.
Today’s Suzie contains almost 20 automated processes, and orchestrates nine different endpoints, including complementary products such as RPA (UiPath), iPaaS (Workato), and ESB (Mulesoft).
“Suzie is handling lots of processes we automated,” confirms Bharti. “And that one automated process orchestrates multiple processes internally. There is lots of capability inside Suzie.”
How Suzie has improved customer experience and operational efficiency for Atlassian
With Camunda, the Suzie bot acts as an end-to-end ordering system, real-time processing for quotes, orders, and refunds. With their process orchestration, Atlassian now enjoys these benefits of a super-charged Suzie.
The original, RPA-delivered 45-60 minute wait time between a quote request and an answer has now been cut to two minutes, even when manual approval is required. This 93% reduction in time makes the process more efficient — and effective.
“It’s tougher to quantify, but the efficient quoting process definitely has a positive impact on sales,” said Bharti. “The quoting process impacts customer satisfaction and then impacts the business as well. We are able to close immediately and add to the dollar value into the revenue bucket.”
Aside from helping close more deals and bring in more revenue, Camunda-supported Suzie also delivers cost savings to Atlassian. The team has seen a nearly 30x return on their automation spend, which translates into money saved on manual development effort.
The most notable benefit of course has been the improved customer experience and resulting customer satisfaction. Speedy, accurate responses make customers’ lives easier, and helps boost Atlassian’s reputation.
It also makes Suzie’s reputation almost legendary.
“Some customers thought they were interacting with a human due to the speed of the response,” confided Bharti. “In fact, even some Atlassian employees think Suzie is a real person handling all of these tickets!”
What’s next for Suzie?
Now that Atlassian has seen success with Suzie’s current capabilities thanks to Camunda’s support, Bharti hopes to increase the number of tickets handled in the next year or two from 8,000 monthly tickets to 20,000 monthly tickets.
Additionally, Suzie’s planned roadmap over the two years also includes more processes to support additional business units and activities, including end-to-end processes for sales, marketing, finance, and additional invoicing.
With Atlassian’s hyperautomated tech stack, including Camunda Platform, the sky’s the limit for Suzie’s potential reach.