- Australian Open
In an effort to activate their brand presence at the Australian Open, we partnered with Blackmores to create the Well Bot, an interactive AI chatbot designed to kick-start conversations about health. Conversing with users over Facebook Messenger or SMS, the Blackmores Well Bot used artificial intelligence and machine learning to provide users with tailored health and wellness advice, ultimately fostering a collaborative digital marketing experience that added value for both Blackmores and its customers.
Official partner of the Australian Open, Blackmores was looking for a way to translate their company pledge to help people “Be a Well Being” into an exciting digital marketing experience for sports fans. Together with Bastion EBA, we created the Well Bot, an interactive AI chatbot that used machine learning to start conversations about health and wellbeing.
The resulting chatbot (and accompanying moderation application) represented an innovative new approach to connecting brands to customers through the technology of their everyday lives. Using artificial intelligence to assess each user’s greatest health weakness and sign them up to a specially developed Action Plan, the Well Bot harnessed consumer interest in chatbot technology to facilitate fun yet meaningful conversations about healthy living.
One of the key goals of the Blackmores Well Bot was to facilitate realistic and meaningful conversations about health and lifestyle with users. We placed great focus on understanding potential user experiences across each communication interface – Facebook Messenger, SMS and event kiosk – so that we could model artificial intelligence conversation flows on patterns already familiar to most users.
Mapping out these user journeys was a core part of our process, and one that helped us to foster comfortable and intuitive conversations between user and chatbot. By focusing heavily on natural interactions we were able to create a digital marketing experience that not only increased brand awareness for Blackmores, but also demonstrated the potential of machine learning and chatbot technologies.
Implementation of natural language processing was crucial to our aim of facilitating meaningful conversations between the digital marketing platform and users. By allowing users to interact directly with the Well Bot in their own words, we were able to enhance the feeling of connection and customisation throughout the interaction.
However, in order to mitigate the limitations of this relatively new technology, we also applied a system of pre-defined responses as a way of giving structure to each interaction. This allowed the chatbot to steer users towards more productive discussions, and the end goal of registering them for a personalised Blackmores Action Plan.
To deal with situations when these approaches fail – such as when a customer’s response was flagged as sensitive or profane – we also developed a moderation application to accompany the AI chatbot. By integrating this specially designed web portal into the chatbot system we were able to automatically transfer flagged users to a human moderator for more specialised assistance.
Moderators were then provided access to the user’s entire chat history, along with the ability to communicate with users directly. Importantly, we made sure to structure the application such that messages continued to appear normally through the user’s preferred interface, helping to maintain a consistent user experience throughout the entirety of these interactions.
In order to develop an AI chatbot that would integrate seamlessly into the lives of users, we designed the Blackmores Well Bot as a system composed entirely of interconnected web services. While this meant the project lacked a single authoritative interface, it allowed for extensive integration into other pre-existing platforms, giving Blackmores the opportunity to reach out to customers more organically.
To implement this wide-ranging integration, we created a unique set of APIs to allow the digital marketing chatbot to integrate with the external interfaces of Facebook Messenger, SMS, and the Australian Open event kiosk. To protect user data throughout this process, we housed the whole system in the cloud using a AWS multi-tiered system architecture. This also provided a greater level of stability across each of the digital marketing campaign’s various communication interfaces.
With the unpredictability of real-life interactions in mind, we developed the Blackmores Well Bot such that it was able to recognise and communicate holes in its own logic. Using adaptive machine learning, the chatbot would monitor for situations when it was unable to understand a question or answer adequately and communicate associated analytics directly to our development team.
Combining this kind of artificial intelligence with an iterative web development process, we were then able to add or adjust streams of logic and conversation to better suit the needs of actual users. This process also allowed us to extend the digital marketing chatbot’s functionality after the initial launch, giving it the ability to answer questions about a user’s Action Plan, and even recommend specific Blackmores products based on their health queries.
As consumers look for more integrated ways to engage online and on mobile, chatbots like the Blackmores Well Bot have the potential to be increasingly effective tools for brands and businesses. This combination of mobile technology and artificial intelligence provides an opportunity to not only create engaging brand experiences, but also to improve customer service, drive sales and streamline purchase funnels.
For us, the Well Bot was a chance to experiment with how digital marketing chatbots could iteratively learn from customers across multiple interfaces and interactions. Indeed, this project and others like it represent a step towards the more intuitive interactions and personalised recommendations that are shaping the future of brand and retail experiences.