I have developed chatbots for several federal departments and universities. The complexity of these projects depended on:
- chatbot software
- user needs
- client constraints.
Below are 3 separate case studies about chatbots used for different users, using different software.
I have also worked on a global voice assistant, but as I signed an NDA I can’t use it as a case study.
Department of Employment and Workplace Relations
I was contracted by DEWR to create their new digital assistant for the Workforce Australia website. This was a new platform where jobseekers need to sign in to report their mutual obligation requirements (obligations). The entire website was developed with a new IA and no search function. The objective of the digital assistant is to:
- help jobseekers learn how to report their obligations
- locate the information jobseekers need to find a job
- reduce contact centre calls.
Software used
We used Microsoft Copilot (previously PVA) as the chatbot interface and Content Workflow as the mechanism to gain and record content approvals. The digital assistant sits in the authenticate space and can only be accessed when a jobseeker signs into their account.
It returns responses based on the trigger phrases, so the trigger phrases need to be carefully developed to match user needs. The content in the response has no bearing on how Copilot matches the intent.
Copilot allows you to create conversation trees in the backend, rather than having to design in third party software and upload. The conversation trees I created allow jobseekers to flow through topics to help them meet their obligations and find work.
Project successes
The Workforce Australia digital assistant has a 90% answered success rate. It was also nominated for an award at the 2023 Contact Centre Awards.
I introduced popular topics, which sit after the welcome message, based on the top 3 intents from the previous week. I also developed content that wasn’t related to Workforce Australia but that our users were asking for, including:
- where to get food help
- homeless services
- services for domestic violence
- mental health help
- self harm help.
I feel the importance of a chatbot, especially for government services, is providing direction to the correct support services even if they aren’t part of your organisation’s remit.
The first 3 images show the user view of the digital assistant. The last 3 images are examples of conversation design within Copilot.
Department of Veterans’ Affairs
I worked with the Department of Veterans’ Affairs (DVA) to develop their digital assistant for the DVA website. This was done at the same time as the complete redevelopment of their website. The objective of the digital assistant is to:
- help veterans locate the support services they need
- answer queries about their entitlements and how to get them
- reduce contact centre calls.
Software used
We used Microsoft QnA Maker as the chatbot interface and GovCMS as the mechanism to gain and record content approvals. The digital assistant was named Pat (short for repatriation) and sits across the entire website.
This digital assistant is a basic QnA bot that returns the correct response based on the confidence level we set it at. This means that the content created in the response needs to use keywords to match potential user intents.
Project successes
I trained APS staff to write for the digital assistant based on the software we used. Making sure they were aware of the importance of keywords in the response. This was to ensure continuation of content standards and patterns after I finished my contract.
We were able to quickly adapt and create new responses that helped veterans find providers and services to help them with their trauma.
Intercountry Adoption Australia
Intercountry Adoption Australia is a program that is part of the Department of Social Services. I was contracted to completely overhaul the Intercountry Adoption Australia website, including the IA and content, then develop a chatbot. The objective of the chatbot was to:
- help users learn which countries they are eligible to adopt from
- explain the adoption process
- reduce the need for staff to respond to call backs and emails.
Software used
We used Microsoft QnA Maker as the chatbot interface alongside Bottal, a specialised conversation design software used by DSS and Services Australia.
Project successes
On the back of the easier readability of content, the chatbot was able to easily answer questions about eligibility and what countries you can choose to adopt from. It also allowed users to quickly access the web page they needed for more detailed information. There was also a significant drop in the number of call backs requested, as the information was now easy to locate and read.
Contact
Let’s talk about how I can work with you to develop an effective chatbot:
Other case studies
You can view case studies for: