The majority of applicants to The University of Adelaide interacted with a Facebook Messenger chatbot during the institution’s busiest period.
The school says the bot freed up staff for more valuable tasks and returned overwhelmingly positive feedback from the potential students because it was deliberately designed as a “complementary” feature that reflected its values.
According to the sandstone institution, the chatbot has allowed the university to engage more students in less time, while maintaining satisfaction on its busiest applicant day of the year. But it was never designed as a replacement for staff.
The bot, available through Facebook Messenger and the university’s website, helps Australian applicants determine their adjusted ATAR — effectively their selection rank — to calculate their eligibility for certain courses. It is a relatively straightforward task but one which had previously been a manual one, requiring staff guidance over the phone. Thousands of students also attempt the task on “Results Day” — when ATARs are released and students can begin calculating eligibility and applying for offers.
This year over 80 per cent of the 4,500 applicants used The University of Adelaide’s chatbot instead, with 82 per cent of the feedback being “awesome”, according to the university. 16 per cent of applicants said it was a “neutral” experience and only 3 per cent rated it “terrible”. Phone calls to the university on its traditionally chaotic and stressful day fell 55 per cent compared to 2016 levels, prior to the chatbot.
The University of Adelaide’s director, prospect management, Catherine Cherry, tells Which-50 this improved student engagement is where the chatbot has had the biggest impact.
“This is where the chatbot has provided significant value beyond the resources that we have available and has allowed the staff to engage in longer conversations with students who have nuanced questions about their result or need additional support with completing their application.”
Under the hood
The bot, built with Oracle cloud software including Oracle Digital Assistant, was first implemented in 2017. It had an immediate impact, Cherry says, but the underlying artificial intelligence and machine learning means it has improved over time.
The chatbot had to mirror the values of the university’s staff to ensure students were comfortable when engaging with the team, according to Cherry.
“We also wanted to ensure that users saw the chatbot as complementary to human resources, only answering certain queries. At the same time, it was important that the chatbot provided responses that were personalised, natural and conversational, especially given the stressful circumstances and need for the students to get an answer quickly.
“The artificial intelligence and machine learning built-in to the chatbot was instrumental in achieving this outcome, as the more conversations the bot had with students, the better equipped it became.”
This is evident in the student feedback which has improved each year, Cherry said. The University of Adelaide is now working on a similar solution for international students.