Like the majority of businesses, transportation companies realize they are also in the arena of (digital) services surrounding their core business. And no longer is a website and some app enough, which is why we recently helped a Danish transportation client explore how a chatbot-based service could offer something valuable and different.
The fields of transportation, commuting and travel-planning are complex, but generally well-serviced arenas in Denmark, meaning that competition and demand for well designed solutions is fierce. Amongst others, those combating for users and customers are the likes of DSB, Google Maps, DOT (a Zealand co-op for public transport), Movia, Citymapper, Arriva, Metro, Rejseplanen (a Danish travel-assistant) and ride-sharing opportunities like GoMore.
And let’s not forget the Danes’ preference for biking, and the fact that approximately 50% of all Copenhageners use the bike as primary means of transport every day 🚲 💨.
So how can any one of these companies stand out and offer something unique, which enhances the customer experience? Our client wanted to explore, if AI and conversational interfaces could do contribute with something else.
This led us to the following starting point:
Which are the primary concerns of urban commuters, and how might a AI and a conversational interface help relieving those painpoints?
The scope was decided as to aim for a beta-product for younger digitally savvy customers (15–25 year olds), which could be tested by expert users. The goal with the beta-product was to get feedback on the concept of whatever we would come up with, as well as provide feedback for using chatbots as a channel and UI in broader terms for the target group.
Evidently, we then needed to explore two high-level aspects before we could design and build the beta-product:
We needed to examine what was technically possible, i.e. what kind of data is available to support an NLP-based service and which systems are open for integration with API.And we needed to figure out who to focus on and what the primary concerns was for that specific group.
In other words, we needed skills from the entire palette of our team, who took the lead on the project, but worked together with the company throughout the process.
While our AI- and development team were assessing potential data streams, integrations and API documentation to clear up the technical limitations and possibilities, I lead a design research effort to cover the hopes and dreams of the young digital-savvy target group.
“… design will tell us what AI product to create, how it should look or feel, who it’s for and which features are important”.
So, since we aren’t exactly 17-year old high-school students ourselves, we needed in-context insights from the users themselves, to understand what’s important to them. In this case, we needed to gather qualitative knowledge on imporant aspects of transportation in their lives, but also actually see and experience it in action, so we could dig into main points of interest.
Therefore, we decided to carry out a version of cultural probes — as described by e.g. Gaver, Dunne & Pacenti (1999) as well as Mattelmäki (2006) — where daily tasks were given to a small group of 10 users. Since the goal was to capture real-life situations easily, and on the go, we didn’t instruct them to use notebooks, diaries and cameras — but SnapChat. Every morning, the users were given tasks, which they carry out with their phones. This is a relatively new way of gathering research, but feedback from both the client and users involved was very positive. If you’re interested in hearing more about how we work with design methods at BotSupply, you’re more than welcome to reach out.
After a week of in-the-field research and interviews, we had a whole day of workshops, first mapping the insights, and then generating ideas on top of it. In its totality, it looked like this 👇
Making sense of it all
By using design, we mapped out a mix of pictures as well as explicit pain points and statements about transport.
Specifically, we used the methods of affinity diagramming, dot-voting and how-might-we questions, to open up potentials of the newly gathered insights. For the latter part of the workshops (the solution-focused) we also invited one of the research subjects to co-create with us and bring in another perspective.
For me and the project’s progression, the most important outcome of the research was a shared understanding and common focus between the client and ourselves.
Eventually, we had a shortlist of wishes and potential services, most of which we could actually build, and some of these were also very much reasonable to create as a chatbot service fuelled by NLP. Additionally, they are all potential iterations and additional features, meaning that the research is not only valuable short term in defining the concept of the product, but can also work as a backlog of additions going forward.
So far so good. We had a user-verified idea and true market differentiator, and we knew it would be technically feasible 🙌
As per the original goal, we aimed to develop a good-enough version of a chatbot concept, so that our client could beta test with expert users. The final concept was a travel assistant, who could get to know your favorite places, and which would, proactively, suggest trips, provide countdown-like updates and alert on last-minute changes while also providing alternatives — all in the shape of a conversational interface powered by our custom NLP.
Besides the overall concept, we also designed a persona to fit the concept and needs articulated from the users. The goal is to have a common idea of why certain design choices is fitting going forward, just like when designers use personas in developing new solutions. This bot has the personality of a mix of so-called helicopter mom (who’s a tiiiny bit too excited, eager and emoji-using) and a geniunely helpful person, who’s always there looking out for its “kids”. This is how we ended up defining her (in Danish). From left to right, we have elaborated on the bot’s background information, it’s quirks and features, hopes and dreams and finally language. what it contains here is the result of working through the user-insights and what fits the concept:
We always advocate at least being aware of how a chatbot will be perceived, and generally advice to create a personality for the bot — especially if it’s the face and voice of a new service for a brand. Add to that, that humans will anthropomorphize a service like a chatbot no matter whether it’s designed for or not (as e.g. mentioned here & here), and you have a pretty good reason to put in an effort.
If you’re keen on learning more about how we use design and character development from screenplay-writing and game development, don’t hestitate to reach out. We’re always happy to exchange ideas and insights (psst! And we love coffee ☕).
Meanwhile, through a little bit more bumpy development phase than hoped, we eventually ended up with a satisfying product though we had to make a few turns and twists along the way.
To Infinity… and Beyond 🚀
As I write this, the fate of the project lies in the hands of the people who’s gonna use it: The young commuters of Copenhagen. Our client is conducting the tests of the MVP to verify the idea of a chatbot-based solution, and we can’t wait to hear the feedback.
We’re thrilled to have been part of the project, pushing the boundaries of what AI, conversations and design can do together.
Next up will be a series of thoughts and impressions on new projects, my upcoming debate gig at Chatbot Summit Tel Aviv and maybe a post on the occasinoal crazy experiment.
See you later, 🐊
A rich text element can be used with static or dynamic content. For static content, just drop it into any page and begin editing. For dynamic content, add a rich text field to any collection and then connect a rich text element to that field in the settings panel. Voila!
Headings, paragraphs, blockquotes, figures, images, and figure captions can all be styled after a class is added to the rich text element using the "When inside of" nested selector system.