Throughout the months-long discovery process, we organized user interviews to build a solid understanding of how clients interact with their data across our, their own, or with third-party data collector and management platforms.

The other pillar of the research aimed to deepen our understanding of the client-side operational and decision-making process related to data utilization. We delivered this research through remote discovery interviews channeling in insights from client-side business unit leaders and analysts, data engineers and developers related to the data-pipeline operations, and external consultants.

At this point, in-house decision-makers and stakeholders could brightly see that we have to come up with a new solution delivering a collaborative platform for data engineers, integration-side developers, and data team leaders to have a common understanding and livable control over the recommendation systems' output.

Through a series of weekly recurring workshops, we involved the developer team and the data team to break down the jobs that have to be done by internal and external data professionals if we want a reliable process for how we ship recommendations to large enterprise companies. 

With the needs of the key personas in mind, we have begun designing the initial processes and possible flows, while also attempting to structure the list of other functions related to data flow management to get a full picture beyond the project's scope.

With further co-creation sessions and design studios we started to identify the most viable and feasible interaction concepts and to define the most fundamental building blocks of what a data pipeline has to be designed by.

Due to the extremely abstract and highly complex tasks that users had to accomplish, the vision was to visualize the data flow in a way that could be both holistic and detailed at the same time.

At that time, whiteboard applications were not yet widespread, making it challenging to provide examples. Eventually, by analyzing 3d tools, architectural, game-development, and pre-production software cases we decided to develop our own whiteboard tool.

Highlighting the most important aspects, the design process consisted of the following steps:

In the first step, we had to map out and analyze feasibility limitations, and the critical components of a deeply visual interaction where, zooming and paning and connecting boxes will be the essentials of usability.  

Since we had benchmarks about whiteboard interactions and ergonomics (Illustrator, Shaderforge, Blender) we decided not to test separately the usability of the whiteboard solution, we rather made tests to understand if the users can effectively and confidently interact with the new data operations and representations - so we focused our resources to iterate these aspects.

After the first live demo, the in-house data team became the early-adopter group that started to use the unpolished product - their preliminary insights transformed into a co-creation, where we defined most of the roadmap items together.

As we understood it in the discovery phase the success of our solution won't be affected only by the well-designed data flows, filters, and best- performing recommendations, but on that how the business part of the other teams will prioritize the common tasks.

After the rollout, we invested a lot to highlight our data-editor's benefits to convince client-side professionals - in belief that they will be our inner ambassadors who possibly will help us to move along with projects, or help to convince decision makers to choose us.