Steep learning curves seem to be a necessary evil for BI software. Looking through nearly any online course catalogue, you’ll find a wealth of material on how to use Tableau, Power BI, or Google Data Studio. 

These analytics behemoths opt to deliver the most flexible and feature-rich solution possible. In doing so, they've geared their products towards ‘technical users’, assuming that those users are already knowledgeable in analytics, or are able to spend the necessary time and effort getting up to speed.

Our own platform, DeepSea, is in a unique position. It functions as a means of sharing ML models & insights between data scientists and ‘non-technical’ domain experts (such as business executives, mechanical engineers, etc.).

Although we’re unique in terms of DeepSea’s machine learning integration, providing an end-to-end analytics solution (extending to dashboarding & data collection) requires us to make those features useful to both technical and non-technical users, aiming to:

1)  Provide the flexibility a data-scientist needs to integrate ML models and visualize model outputs.

2)  Empower non-technical users to explore their data, through visualizations, filters, and model configurations.

This duality often causes power to clash with ease-of-use during the design process. When it does, rather than ‘dumbing down’ our platform, we can uplift our users. With that in mind, here's my working formula for making DeepSea both easily learnable and surprisingly powerful:

‍

‍Educate our users.

DeepSea has a non-linear workflow, which includes data collection, data ingestion, and dashboarding with integrated machine learning models. For a system with this many pieces to be understood, we need to explicitly communicate how each one fits together.

We can use a few strategies to communicate DeepSea’s underlying workflow, including: 

• Contextual hints, such as styling, text, and diagrams—that remove ambiguity around what something is, what action is expected, or what the result will be.‍

• Documentation, including a brief ‘getting-started’ guide, searchable bite-sized information, and example use-cases demonstrating how to effectively use DeepSea’s toolset.‍

• Immediate visual feedback, allowing users to preview the results of potentially confusing or consequential actions.‍

• Notifications & links, clarifying the hidden results actions have across different parts of DeepSea.

‍

Account for human error.

In order to make a product that feels safe to explore, we need to account for the possibility of human error. By creating a clear and forgiving system, we put up ‘guard rails’ that give people assurance they can experiment safely, enabling them to 'learn by doing'.

We can create a forgiving experience by:

• Reducing irreversible decisions, allowing people to experiment freely without the fear of repercussions.‍

• Warning users about destructive actions, prompting them to give these actions additional consideration.

• Explaining the implications of decisions ahead of time, allowing users to understand the effects a consequential decision will have within the overall system.‍

• Providing easy exits, such as cancel and ‘x’ buttons, that allow people to effortlessly leave unwanted situations and areas.

‍

Reduce interaction cost.

‘Interaction Cost’, a common(ish) term among UX designers, can be summarized as the overall mental and physical effort expended to accomplish a goal. It’s the sum of confusion, comprehension, memory-load, reading, clicks, scrolling, and waiting required to get something done.

When dealing with complex data, some extra cognitive load comes with the domain (in comparison to consumer apps like Netflix, etc.). This makes it especially important to take advantage of the quick wins that make our product easier to use. We can reduce interaction cost with:

• Visual hierarchy & grouping, meaning that similar elements share styles and locations, while elements on the screen are given visual weight in accordance with their importance.

• Repeating UI & interaction patterns, meaning that the UI will look and work similarly across different parts of the DeepSea, so people can refer to previous experiences to contextualize future ones.

• ‍Clear and concise vocabulary, with visual aids and descriptions for 'technical' or uncommon terms. By being diligent about the quality and brevity of our writing, we enable our users to spend less time reading—and make it painless to do so when necessary.