Key elements of the role
Customer Experience Strategy | Research | Stakeholder Management | Stakeholder Workshops | Competitor Analysis | Customer Research | Copywriting | Guerrilla Testing | Wireframes | Interactive Prototypes
The problem
To understand the loyalty program's benefits, Qantas Frequent Flyer customers must phone the call centre and complete a questionnaire. The thorough questionnaire touched on lifestyle, products, purchase frequency, etc. A standard call would last 30 minutes. The problem to solve was how to reimagine this experience.
Background
For customers to fully understand ways to earn and boost points, the process would be to phone a Qantas call centre. This consisted of listening to a pre-recording and the inevitable having to hold, ‘your call is important to us’ messages. To understand and learn the full benefits of the rewards program, the questionnaire could take up to 30 minutes in total to complete.
Other considerations
A customer could request a customised email highlighting the Qantas business partner offers, the number of points that could be earned and a link to the offer.
I was brought in to lead the user experience strategy two months into the program. Up until that point, there hadn't been any UX input.
Development (dev) work had been carried out to understand whether a virtual assistant experience could be a viable option. The solution of a chatbot experience still needed to be validated with customers.
Apart from the call centre scripts, there wasn’t any existing assets for the virtual assistant experience. The scripts from the call centres would need to be reviewed to distinguish whether they could be used verbatim in a virtual assistant experience. The short answer to this would be no.
Qantas hadn’t previously built a virtual assistant. This would be a first.
A separate team conducting research into the characteristics, look and feel of the avatar to be used. At one point, it was between a human and kangaroo avatar. I kid you not.
Approach
The timings of the delivery framework had been agreed prior to me joining the team, therefore my immediate task was to understand the problem statement and being brought up to speed (and any UX decisions made) on the project thus far. This started with me facilitating a stakeholders workshop.
The session was used as a 'questions and assumptions' exercise on the product. This was to ensure all stakeholders were aligned and that there was a clear goal. This in itself was a valuable exercise as there were a few 'grey areas' and unanswered questions that needed to be fleshed out during the workshop.
Due to the fast-paced nature of the work, the approach consisted of daily team stand-ups & reviews of UX research and UI design, ensuring openness, consistency and collaboration.
Discovery
The initial task was to review the decisions and approach made of taking one system (call centre questionnaire) and analysing whether this would translate verbatim into a simplified tablet chatbot experience. As this was an interactive product, a natural flow of language, interaction and timing, was going to be essential for a chatbot to be useful and usable, and ultimately a success.
The first port of call was to review the language, phrasing, type, length, order and the number of questions the call centre questionnaire required to be completed. A considerable benefit to reviewing the call logs was that it quickly identified customers pain points and goals. This led to understanding where and how they wanted to learn how to boost their points.
Reviewing the scripts, they worked well for a two-way telephone conversation. For this product experience to be useful and implemented successfully, it was critical to review and address all areas to ensure the best customer experience could be developed. These issues would include from being short on time to customers privacy concerns (where to place the tablets).
A clear distinction was needed to identify what questions were required and what was superfluous for a chatbot experience. Being time sensitive and frictionless from a customers perspective was paramount.
While in the process of reviewing the order, wording and phrasing of the questions, I quickly identified the call centre questions were too long and not transferable to a chatbot experience. The call centre questionnaire was broken into 5 categories; Home Express, Lifestyle and Entertainment, Travel, Financial, Reward and had a whopping 31 questions!
Due to the tight launch time frame, senior stakeholders decided the responses to questions would be predefined. This meant I would need to reframe and redefine the problem. What would have been a chatbot experience with AI capabilities would now become a virtual assistant who would guide the customer more linearly. However, still keeping the experience conversational and as natural as possible. With this in mind, additional research was conducted on smart form industry leaders and their approach. Due to the amount of tech and dev effort already committed, it was agreed to continue with the original approach, rather than use a smart form platform.
Define
As the project already had considerable effort and budget invested and stakeholder buy in from earlier customer conversations and business decisions, the 'end product' had already been defined - a chatbot experience on a tablet device. It was now my responsibility to lead a multi-disciplinary team using human-centred design methodology to create and build an engaging, informative and joyful product. The product would be an MVP for tablet only. However, the other virtual assistant projects that were on the digital roadmap would be leveraging and reusing assets and data from this project. No pressure!
Ideation
As the glue for this product is the interaction of copy and tone of voice, the first task was to craft the questions in a manner that would resonate and encourage customers to engage with the virtual assistant. One of the first decisions was ‘how might we’ reduce the number of questions and amount of copy from the call centre script and craft those questions into ones that are shorter yet still informative. My copywriting skills had to be dusted off as initially there wasn't a copywriter on board to collaborate with. Addressing the copy and having the correct balance of friendliness and being informative was of paramount importance.
Organising the questions and logic into a system that is scalable and repeatable.
Many questions were identified as not adding value, they were culled as they did not fit in with an overall virtual assistant experience.
Ideation of the Qantas script continued with crafting the questions and possible responses. There were two rounds of ideation and guerrilla testing to ensure we were on the right track. One approach considered a question and multiple answer scenario that would contain a level of logic depending on the answer given. However, this approach still comprised of 15 questions, still far too many.
“One opportunity that might be available to you is to look at a Qantas Frequent Flyer Points Earning Credit Card. There are a range of different cards, from different providers that are available which can earn points not only on spend but also have sign up offers which can provide you with access to bonus points. If you're interested I can send you the link to website with a comparison tool so you can look and see if this is something you might be interested in?”
An example of the script of how to earn points via Qantas credit card, I think you would agree this would be too wordy for a virtual assistant experience. Just imagine getting all that on one screen.
Early iteration mapping the conversation flow. This version included the multiple answer approach
An example of a multiple answer scenario. If the customer selected Woolworths (a Qantas partner) as their favoured grocery store they received a different ‘encouraging’ message to a non-Qantas partner. Note the copy was still work in progress at this stage.
A scaled-down version (10 questions) was guerrilla tested. This experience, however, was considered too laborious and time-consuming by participants. The goal was to be informed in a delightful and timely manner. This then led to further crafting of the script and what would ultimately be used in the interactive prototypes.
Throughout this phase, the virtual assistant UI design was in development. At this point, the Qantas new brand style guide was also in early development and hadn't touched on chatbot interfaces. The components from this project would inform the brand style guide.
"The virtual assistant had quite a journey with its names, going from "Peta Pilot" to "Matilda" before finally settling on the name "Skye" – which customers preferred. Many components like colour and styling were still being explored during this process. The avatar design was a placeholder until the final version received approval."
Prototypes
Low fidelity wireframes were built using Sketch with interactive prototypes created using Invision. Based on previous customer feedback and the development of the UI design components, high fidelity versions were quickly produced. Both sets of prototypes were guerrilla tested with customers. Feedback and data from the low fidelity wires were iterated and informed decisions on the hi-fidelity versions. It was these hi-fidelity prototypes that would ultimately test our hypotheses 'in the real world' - i.e. in the Qantas shop at the airport.
Interactive prototype, this version shows the progress bar moved to the top of the screen. Previous iterations had this placed at the bottom where there was an input field and action button.
This would have caused a great deal of dev effort for the two to work seamlessly and intelligently - this issue was uncovered, during guerrilla testing.
Test and learn
To obtain a level of qualitative data, due to the unavailability of a customer interview panel and time constraints, guerrilla testing was conducted. Two high fidelity prototypes on tablets were set up in the Qantas shop in Sydney domestic airport. This was an exciting experiment as this was the first time this had been attempted. Many customers said they would approach the tablet without encouragement. This was mostly in part to the conversational welcome message. Tick. Many said they found the virtual assistant incredibly useful and usable, as it showed them ways to boost points in ways they were not aware of. Tick. And very simply too. Tick.
The data from the guerrilla testing was summarised and played back with recommendations to the stakeholders and broader teams.
Hi-fidelity prototype
Summary
I moved into a new role while the first release of the project was still in development. This move was due in part to while working on the UX strategy for this product I had identified further business opportunities within the rewards program space. In particular an improved onboarding process for new and dormant customers that highlighted pain points and moments that matter.
My role during the development stage was to consult the team and stakeholders. This included guiding the newly appointed UX designer to ensure the designs and interactions were accurate to the agreed research and user experience. Consultancy also included collaboration and facilitation between the UI designers and the Qantas Design Language System team.
Some exciting challenges came up and were navigated during this product research. Namely validating a hypothesis that had a solution before any deep customer research had been conducted. A further challenge was two weeks into discovery, the concept of an AI virtual assistant was revisited and due to the complexities, as a group it was decided to explore and build a chatbot with predefined responses. This meant I would need to reframe and redefine the problem. What would have been a chatbot experience with AI capabilities would now become a virtual assistant who could guide the customer in a more linear fashion, however still keeping the experience engaging, conversational and natural as possible.
This task meant some additional research in this field and talking to what we ultimately needed, customers. I believe even though this project did not start off with ‘everyone in the room, who should be in the room’, as a team we managed to achieve a great outcome and a great product.
Meet your virtual assistant - Skye
Outcomes
Virtual assistant components becoming part of Qantas Design Language System
A cultural shift to human-centred design approach being instilled across Qantas digital teams
Increased collaboration of previously siloed virtual assistant teams and projects
UX learnings being adopted by other tribes working on virtual assistant experiences
Role led to a more strategic position within Qantas rewards space
Gaining a more comprehensive understanding and learning more about AI experiences
Major achievements
• Successful launch of the virtual assistant
• Excellent customer feedback
• Copywriting skills were utilised
Tools & tech stack
• usabilityhub.com
• surveymonkey.com
• Google Docs
• Keynote
• JIRA
• Confluence
• Pen & paper
• Apple Numbers (spreadsheet)
• Sketch
• Invision
Client
Qantas Loyalty