Main Tasks
- Design the native & web app experiences.
- Design the on-board kiosk ordering & queue app experiences.
- Usability tests.
- Motion & interaction.
Once Alike Coffee was a subsidiary of the automation start up Aabak, where our aim was to completely redefine and automate the cafe experience. The strategy around this goal was by reducing the footprint by a multiple of 10, cutting staffing costs and brewing objectively the highest quality coffee, Once Alike would embed itself into the daily habits of the CBD worker. During the R&D phase we produced three iterative prototypes, one of which was put into public beta testing (Mk2). We produced two more prototypes post Mk2 before the company deemed the Mk4 was ready for production, where a further eight units were manufactured in Melbourne.
I joined Once Alike in 2018 at the end of the Mk2’s life cycle, my role was to lead the UX & UI design of all subsequent prototypes/production models.
- Design the native & web app experiences.
- Design the on-board kiosk ordering & queue app experiences.
- Usability tests.
- Motion & interaction.
Redefine the cafe experience for users with a service that produces the fastest & highest quality coffee perfectly tuned for the individual by the individual.
- Product inconsistency between cafes and baristas.
- Wait times during rush periods.
- Inconvenience of having to go to a cafe before placing an order.
- Overhead costs in setting up and operating a cafe.
- The city commuter who has not got a lot of time in the mornings to get their coffee in the morning.
- The specialty coffee lover who is extremely particular about their order.
- The landlord looking for a low maintenance amenity for their office building.
During the Mk2 beta testing we found that users would visit the site for around two weeks before returning to normal routines unless they wanted to show a friend. Although the experience was innately better (consistency of shots, grind weights & prep speed) we were still limiting users to place an order (on an iPad) and line up single file, thus making the experience no more convenient than a standard cafe. Until we had a mobile experience we weren’t able to truly tap into the users' habits and rituals.
.png)
In the early stages of the mobile experience design we came to a key junction, where we needed to decide between a cascading list style of menu or something else entirely. The cascading list had become very popular in food deliver applications and thus seemed like a safe choice in repeating that pattern. However during this discovery phase I had come across a hole in the wall cafe that had implemented an iPad to supplement its ability to take orders that used a cascading list menu style. During my morning commutes I was factoring time to observe these users, what I found was that one of two things would usually occur for new users:
- Users would select a coffee type and milk but would get confused as to why they couldn’t add to cart, this was due the fact that they had not made all required selections which needed to be scrolled into view by them.
- Users would know to scroll the view but would get struck with decision paralysis due to some coffee types requiring more selections than others. The ordering sequence would not be obvious for them, thus making this process slower than ordering verbally.
With more of our own testing we ultimately had opted for a ordering process that incorporated progressive reveal into its flow. This option allowed us mimic the process of ordering verbally where the process is usually broken into four parts, coffee type, coffee strength, milk type & an additional preference (sugar, extra hot etc). While a cascading list menu is usable across different hospitality types, in our specific case it was not the optimal solution.
.png)
Ordering kiosk
The Mk4 features a front mounted ordering kiosk for users who don’t have access to the Once Alike native and web apps. The Ordering kiosk has a significantly reduced feature set that only allows for ordering coffee, however detail in how users order their coffee is not diminished. Users can tailor milk type, milk volume, milk heat as well as water volume and water heat using using any of the coffee types as a base(ie long black, cappuccino, latte etc).
Queuing kiosk
In order to update & alert users on the status of their coffee the Mk4 featured a kiosk solely dedicated to the current queue of orders. The queuing kiosk displayed the users coffee in three states:
- Queued
- Preparing
- Ready
Native & web applications
The Once Alike native and web apps enable the best experience as they presented a map view of all active machines and their current wait times, it also allowed for users to create accounts. With accounts users could then engage with a loyalty system which greatly increased Once Alike’s user retention. The native experience had a particular advantage over the other platforms because of its more accurate location services. the web app was restricted orders within a 2km radius (this was to prevent unnecessary queue build up). Native app users could order from an extended range and always collect their coffee hot, the service was able to achieve this for native by only placing orders into the queue based on the users ETA.
For our Mk4 usability tests we wanted to conduct the test around the users entry points into the service. For this we constructed ten different scenarios where we could observe and measure the user’s initial, secondary and tertiary exposures to the experience. The users were broken up into two main groups and a further two subgroups for the scenarios.
- Group 1 will be a series of isolated tests on mobile and the Order Kiosk. Group 1 will be split into 2 subgroups to further test how order of events vary user experience.
- Group 2 will be focusing on the Order Kiosk and testing the experienced in a busier environment. After Group 2 have completed 1 order each they will also be split into 2 subgroups to test the more advanced options of the order.
The key insight we gained from these tests were around sign posting and that we were not doing enough of it on our apps. The first point of failure was around users being unable to fine tune their coffees (add more milk, decrease water, increase temperature etc) in the advanced adjustments, this was a critical problem to fix if we wanted user personalisation. The second point of failure was our sign posting for our tray dispenser, when prompted to order multiple coffees it was not obvious for users where to find the trays.
G1S1 (Test 1)
This group will be testing the onboard order app separately and isolated.
Testing first exposure without other users to take cues from.
G1S1 (Test 2)
This group will next test the mobile app under the same conditions as last. However, they will only be allowed to go up to the MK4 once their coffee has entered the airlock.
Testing UX for more confident users who have gone from Onboard to Mobile.
G1S2 (Test 1)
This group will be testing the mobile app under the same conditions.
However, this group will only be allowed to go up to the MK4 once their order has been placed.Testing UX of mobile app first adopters.
G1S2 (Test 2)
Repeat test above.
Further testing of the ‘app unlock’ feature.
G2 (Test 1)
Group 2 will form a queue and begin to order from the Onboard app.
Testing the onboard order app in a busy environment.
G2S1 (Test 2)
This group will be given either a flat white or latte with the condition to make a milk adjustment.
Testing the UX of the ‘advanced adjustments’.
G2S2 (Test 2)
This group will be given either a flat white or latte with the condition to make a milk adjustment after it has been added to the cart.
Testing the UX of updating the cart.
Group 1 + 2 (Final test)
Both groups will be asked to order 4 coffees per person in isolation, Group 1 using the Mobile app Group 2 using the Onboard app.
Testing the communication for use of trays.
During the manufacturing phase of the Mk4 we installed the first unit produced into a CBD location. We used this location to test our onboarding as well as our loyalty and review systems. Through beta testing were able to improve our sign up conversion as well as our post order reviews by incentivising engagement, this was achieved by allocating in app credit to users to spend on coffee.
We put several systems in place to minimise a users wait time without compromising their experience, however the most significant were:
Machine Queue Times - Users could check the current wait time of machines nearest to them.
Ordering Ahead - Users could place an order ahead of reaching the Mk4’s location, saving them from waiting in line to order on the kiosk.
Optimised ETA - For users ordering ahead this system anticipated the pace of a user against the current queue time. By doing this Mk4 would only start preparing their order once they were a certain distance away, allowing for users to always get their coffee hot.
User habits became a core tenant for the team to solve for, it required us to find the opportunities for improvement in a users coffee run during the morning and lunch rushes. By building a favouriting system that allowed users to customise and tune their order in great detail we gained an edge over cafes. Ordering through the web/native app users get their tailored coffee in one tap, versus a cafe where a user has to line up and state their order every time, with the best case scenario being that they will be recognised as a ‘regular’ and have their order memorised by the barista.
.png)
An early concern of the mobile apps was around how would we solve for keeping a users cup safe if they weren’t there. Our solution was to only open the airlock if the user had pressed unlock in the app, we had initially taken this one step further but only allowing the user to unlock when they were in a 100m radius of collection to avoid any accidental taps. However we found that users were trying to unlock remotely and getting there friends to collect coffees on their behalf, in response we turned this measure off.
The corner stone of our retention strategy came in the form of our loyalty and feedback feature. Through this feature we completely retooled the ‘coffee stamp-card’ by rewarding users for reviewing their coffees. We created a feedback loop where our product was continuously under QC with users alerting us to potential drops in standards, they would then be rewarded for this with Credits in which they would then be inclined to spend or bank for their next purchase. Loyalty and feedback performed extremely well as we saw a 92% engagement rate with a lift in customer satisfaction from 78% to 85%.
The mounted kiosk applications provide users without access to a personal device place and retrieve orders from the Mk4.
Once Alike was an experience that was unique for users and formative for the team behind it. This experience taught us that trying to facilitate relationship between people and automation is a challenge that is overcome with quality and convenience, not novelty. Once Alike had manufactured 8 machines and installed 3 into sites across Melbourne. The roll out strategy was to install machines into high foot traffic office lobbies within the CDB. Due to the COVID-19 pandemic, use of and further installation of Mk4 units were no longer viable in central Melbourne. With CBD workers now at home we had pivoted on our roll out strategy to suburbs, however this was not enough as the company was unable to generate enough revenue to sustain itself. Despite the company being unable to find success, this undoubtedly was still a net positive experience, being a small and scrappy start up pushed the team to develop and hone a breadth of technical skills that we would not have otherwise.