11
December
2020
|
09:00
Europe/Amsterdam

A user-centred approach to designing AI concepts

Summary

Part two of the Prosus AI series on how to achieve AI-by-design product and business innovation

Every company wants to innovate. But it takes a carefully designed approach and team to translate a high-level goal such as 'AI-by-design innovation' into real and valuable product ideas. In this blog, we will highlight some of the practical steps that our Squared Initiative team, established to work on AI innovation in parallel to incremental AI, goes through to get to a winning idea with a potential to disrupt the business.

Our squared approach follows five steps: discover, define, concept, test and inspire. A disadvantage of presenting a five-step approach to innovation is that this might make the innovation process seem like a linear one. In reality, the beginning and end of each phase is usually not clearly defined and requires a highly flexible and iterative approach. Yet having this phased approach avoids a random, potentially never-ending search for a winning idea.

In this blog, we cover the first three steps of the process, namely discover, define and concept. We highlight what it takes to find AI-driven ideas that can help future-proof your organisation: radical, yet practical applications of AI that solve real and big customer problems. We’ll discuss testing, concept de-risking, scaling ideas and inspiring others in upcoming blogs.

Step zero: what's the focus?

Any innovation team needs a mandate: which part of the business can it, or should it, disrupt? This is what you do in step zero or the scoping phase: investigate which part of the business or market to zoom in on. You talk to leaders in and outside of your organisation and quickly size up the (future) opportunity of different segments.

Criteria that you can apply for scoping include the obvious ones, such as revenue potential, the ability for AI to disrupt and the estimated mismatch between customer needs and current products. But on top of all of those criteria, there's another essential one: what excites the team? If you have a top-notch team for your initiative, you need to give them some freedom to work on topics that make their hearts beat faster.

The major challenge in scoping is figuring out how close you should stay to the priorities of your top leaders. Solving the top problems on the minds of your leaders will make sure your solutions will be picked up rapidly and at scale. Venturing into topics that are removed from daily business increases the risk of creating good concepts that never get scaled due to a lack of urgency — though this might be exactly the mission of a Squared Initiative: finding the "unknown unknowns". A way of mitigating this risk is by ensuring that the daily squared team has the capabilities to prove their own concepts at scale and be as bold as they want.

The major challenge in scoping is how close you should stay to the priorities of your top leaders

Discover: where can you find the best opportunities?

In the discover phase, you immerse yourself in context. You need to understand the field or industry you’re in, as well as the needs and values of your users. And you certainly want to research the latest AI trends beyond your context. This ensures that the initiative builds upon existing knowledge and will help spark ideas in the concept phase.

“Deep empathy for people makes our observations powerful sources of inspiration.”

- David Kelley

 

When we launched the Squared Initiative, our team members had never worked with each other before — and didn't know much about each other's fields of expertise. Our data scientists thought service design was a buzzword, and the service and UX designers only knew the basics of AI. We gave each other masterclasses to understand the value of AI and service design, and to plan out a joint approach. However, as a team, we only started to really understand and combine both fields much later in the process.

When combining and analysing insights, we were not sure how user insights balanced against AI opportunities. Our in-depth user interviews gave us many user insights, and our desk research into promising AI companies ended up in an extensive benchmark of AI opportunities. But what is more important in picking a focus area for the team —putting tech or the user first? We chose to focus on user insights and use AI opportunities for inspiration — to stress the importance of user pull, instead of tech push.

What helped us in this phase was the virtual equivalent of a great war room. Our wish to have infinite wall space was granted with our Miro account, where we could collaborate visually. In this digital playground, we could create models and drawings, capture insights, discuss, and recreate as much as we wanted. This helped create a shared understanding of the context — an invaluable start of any innovation project.

Define: what mission do you solve?

During the define phase, your team translates endless clusters of user insights into a small number of missions (or problem statements). You are synthesising information to formulate actionable missions to invest your team’s time and effort.

To understand which missions could have the biggest potential user impact, we created a common understanding of our users. This consisted of needs-based personas (i.e. personas based on needs rather than on demographics); quantification of these needs and personas (i.e. how often does this need occur in our target audience and how large is this persona segment?); and user journeys with prioritised pains and gains.

Note: if the above seems like a bunch of gibberish, you can read more about needs-based personas here and more about customer journeys here.

Again, AI and service design were closely connected. The mission that the Squared Initiative works on shouldn't only address the biggest needs of your customers, but should also uniquely be solved through AI. In this phase, our team started to reap the benefits of a shared discover phase. We all had an understanding of what our users value — the designers didn't need to convince their team members what direction should be taken. Building on each other's expertise, we could take AI opportunities into account when prioritising mission areas.

The process of selecting the 'right' mission can create uncertainties in a team like our Squared Initiative. Our ambitions of the initiative are big: we aim for missions that allow for non-incremental innovation while containing a significant AI component and solved real user problems. To make an informed decision, we ranked potential missions on these criteria. But even if we had found data to back up the score, our ranking was still based on assumptions and the intuitions we developed about the user needs. It can be hard to see a clear, single winning mission early on. Instead, we decided to opt for multiple mission directions (problems to solve) simultaneously and see which ones sparked the most innovative ideas. Why put all your eggs in one basket anyway?

Selecting the ‘right’ mission can create uncertainties, because ambitions are big

A practical example to illustrate this phase is by thinking about e-commerce. Imagine that one of your team's findings is that your online shoppers experience a tension between getting endless inspiration and actually acquiring an item. Such tensions can be a starting point for ideation: how can AI play a role in resolving this tension? For example, by trying to converge the endless browsers to buyers through relevant and timely recommendations.

Concept: how do you solve the mission in the best way?

Time to solve the mission — or missions in our case. During the concept phase, you break that larger mission up into smaller actionable pieces in the form of How Might We (HMW) questions. And you come up with loads of ideas to solve the mission. This means walls/Miro boards filled to the brim with post-its with ideas. You cluster ideas, turn the most promising ones into concepts, and score your concepts on predefined metrics (e.g. noveltydesirabilitydata availability).

One challenge in this phase was how to integrate AI into the ideation. We first tried incorporating AI in our HMW questions (e.g. How might AI help solve user problem X). Although a good HMW should open up endless creative possibilities, this AI-integration rather withheld us to come up with ideas. Also, we realised this AI-focus creates a strong bias: you might end up solving problems in a too complex, AI-driven way. It is better to keep ideation open to any type of idea, and introduce AI later as a core metric for prioritisation: is this a problem that can uniquely be solved through AI techniques? And how strong is the AI component?

Another learning is to brainstorm on user-centered HMW questions and to introduce some AI-inspired What If Questions. We also used the Machine learning interaction cards of Futurice’s IA design toolkit — and used these like What If questions (e.g. What if you would solve this by means of Prediction). Both methods helped us broaden the solution space and come up with ideas through the lenses of AI and ML.

We ended with a bunch of diamonds in the rough. After creating a few more iterations, our team had winning concepts.

AI-focus in ideation can create a strong bias: you might end up solving problems in a too complex, AI-driven way

At the end of the concept phase, you should end at a high: your team found a potentially very impactful concept idea. You might have uncovered a radical new business opportunity that builds on some of the core needs of your users and couldn't be possible without the latest AI techniques. On top of that, you might end up with a long list of other ideas that may not be a match with the ambitions of the Squared Initiative, but can still create impact. The team can pitch these ideas to other teams throughout your organisation.

We will discuss our next phases (test and inspire) in more detail in our next blog on how to quickly de-risk the concepts of the Squared Initiative. This is where it's time to move from ideation to being critical. Be prepared for your darling to get killed — or at least to get some bruises.

Stay tuned for the next edition and feel free to get in touch with us at [email protected] to continue the conversation on what AI innovation really means, and how to achieve it.

We’d like to thank our colleagues from the Prosus AI team, OLX Group and Koos Service Design for their suggestions and help in editing.