At a glance
Deflected volume by
Volume per week
Avg. FRT for live chat
CSAT for live chat
When you operate one of the world’s biggest global classrooms, with more than 1,800 courses from 140 partners across 28 countries, deep data dives aren’t an elective but an absolute requirement.
Yet when Michael Robichaud, Manager of Support Products, joined Coursera in 2014, he inherited a support infrastructure that was unable to meet the demands of the fast-growing educational business, let alone parse everyday data in a way that would help improve the operations team’s workflow.
At the time, Coursera’s support team was managing all support tickets from a single inbox inbound queue without action-based triggers and time-based automations that resulted in “dramatically high resolution times.”
“We were running into a lot of limits, particularly with the numbers of triggers and automations we could implement,” Robichaud said. “One of my biggest motivations at the time was to implement some granular logic in terms of routing and automation.”
In order to do this, one of his first initiatives was to switch to Zendesk Support from the company’s previous solution, Desk.com. Robichaud leads the support content team, as well as a team of developers responsible for internal tools and systems. With the bigger picture in mind, Robichaud wanted customer service software that could easily pull data from tickets into the company’s data warehouse.
“I had a desire to build internal tools to augment the functionality that exists in the core ticketing system and create custom apps in the ticketing interface that could connect with our internal systems,” he recalled.
“It was just a very clear choice that moving to Zendesk Support would allow us to do that and reduce a lot of the headaches we were having,” he said. “Support was a system we could grow with—which was really important.”
Robichaud ended up completely rebuilding Coursera’s support infrastructure, adopting Support both for external customer support as well as for the company’s internal IT help desk. His team also created two help centers, giving both learners and the company’s partners dedicated self-service destinations packed with quality content.
“Our help center is widely visited by our end users,” Robichaud said. “We’ve purposely put a lot of helpful information in there so it’s not only a place to go when you have an issue or need to fix a problem. It’s also a place to go to learn more about Coursera.”
Coursera’s help center deflects as much as 20 percent of projected volume.
“The forums have provided a transparent, open place where people can discuss issues or ask questions about the product and use the community as a resource rather than rely on 1-to-1 support, which can sometimes take longer,” Robichaud said.
As one might expect, Coursera’s total ticketing volume has steadily increased alongside the number of students enrolled in its courses. Without the help center and community forums, Robichaud estimated that their ticket volume would be 20 percent higher than it is today. “Guide has had a significant impact on us,” he said.
Zendesk’s “very rich APIs” have also made a difference, Robichaud said. For example, the team created a lightweight integration to alert the support team whenever there was a volume increase over a certain threshold. The team has been able to view historical data stored in Coursera’s data warehouse to better understand trends about days and times when volume spikes above average.
“Support has allowed us to be much more proactive and to dive in and understand the root cause of a volume increase—whether it’s an outage, a bug, or something else. “It’s an added value to the operations team, but also for our engineering teams, since we’re now able to surface bugs and issues much faster.”
Using the APIs, Coursera easily extracts data and then uploads it to the data warehouse, where it can be parsed and displayed in forms that users can digest. Robichaud’s team created a tool that generates weekly reports for agents to view their own support activity.
“We have a diverse set of issue types and tickets that we support,” he said, “but we’ve been able to create very granular reports and dashboards because the API allows us to augment the already great functionality with even more custom functionality that we can build into an app.”
One of the insights the team has been able to glean since implementing Support is a preference among a large number of its users for live chat over other channels of support.
“We’ve seen just incredible resolution times with chat,” Robichaud said. “Issues get solved in minutes whereas previously, it could have taken potentially a couple days with all the back and forth emails. We’ve also seen a dramatic increase in satisfaction with live chat, and the fact that Zendesk Chat integrates so well with Support is amazing.”
Metrics and analytics are the lifeblood of any operations organization. Indeed, the changes that Robichaud and his teams have ushered in provide management with deeper and more granular statistics. What’s more, as he puts it, Coursera now has an infrastructure that will “scale for years to come.”
“Everything from Zendesk Chat to Guide to the core Support ticketing APIs has been tremendous for us.”