SAP – Analytics Cloud

Platform: SAAS Cloud Services
Feature: Smart Discovery
Role: UX/UI (Team of 2)




Design Definition

Problem

More and more data is becoming available and it can be overwhelming to understand for the average decision maker.

Goal

Integrate SAP's predictive services into Analytics Cloud by offering a Key Influencer Analysis to help break down influencers in large datasets.

Details & Context

Analytics Cloud had built out two of the three strategic pillars with analytics and planning capabilities. This work was the first step towards offering predictive services.

Behavioural Assumptions

Decision makers are frustrated with the amount of data at their fingertips. Often, decisions are made by a gut feeling.

Strategic Approach

As a year-long project, this feature was costly for both design and development. It was crucial to understand the persona, technical requirements, & limitations.

Persona

Alex can perform light analytics to help make key decisions during his business day. He's looking for help in crunching large amounts of data, but can't do it on his own.

Technical Requirements

The Guided Machine Discovery's core is a Key Influencer Analysis, which required a deep dive to understand the processes.

Knowledge Experts

My team included myself, another Interaction Designer, Data Scientist, Development Manager & Development team. This project also required regular collaborated with teams in Dublin and Paris.

Limitations

The full design scope was not feasible for the first release, therefore, this design was developed and released in two phases.

Project Timeline

The Process of Design

Why was this project successful?

Project success was achieved by designing with stakeholders instead of to stakeholders.

Iterative Design

Working iteratively provided the best possible designs. Each design carried pros and cons, which were reviewed to decide how to move forward. This project benefited from an amazing team who enjoyed participating in the design process.

What did I learn?

• I love designing systems, both for my projects, but also for my own project management.
• Including key stakeholders throughout the entire process helps guarantee the product’s success.
• In most cases an agreement can be made on how designs can be de-scoped for release, but there are times when you can’t give in if you feel the UX is being compromised.

A Picture Says a Thousand Words

Project Design Stages

This was the first predictive feature designed for BusinessObjects Cloud. The process involved gaining an understanding of the predictive process, breaking down both applications, and deciding how the features would integrate.

What I Started With
Information Architecture
High Fidelity Specs
Prototypes
Working Wall
Wireframes
Visual Design

Here is an example of how iteration solved a design problem.

How can we remove confusion for the UX tab pattern which currently has data and workspace in the same tabs style?


Why was this change necessary?

• Predictive needed to be more prominent within the application.
• The Guided machine Discovery is an automated analysis that runs on an entire dataset. Because of this, the feature should be placed within the Data space.
• Customers found it difficult to find and access the data space.

How does this help the customer?

Our user works with data every day. Having a clear understanding of how to navigate to and from the data space is imperative for a successful product

Pen to Pixel: A Path For Growth

I am a life-long learner, and each project I take on drives my personal growth. Large and complex problems like this one continue to offer me a great deal of personal satisfaction and growth.

This project allowed me to think freely about the beginning-to-end workflow of our predictive users. I was able to create a design strategy which worked with the persona, technical requirements and scope limitation but also addressed the inherent risks. The design process generated a simple design to the customer, which was then validated during user testing. Through the process, I was able to defend my UX when needed while working with an amazing and collaborative team.

"A question is sometimes better than an answer." Charles Ives

Here is a sample from user testing.

Takeaways

How To Measure Success

Success to me means realizing that I am a member of a larger team. Including stakeholders in the iterative design process helped shape an efficient and successful product.

Large & Complex Projects Offer Growth

I enjoy designing systems for my projects as well as my process. I enjoy thinking in terms of macro and micro when designing projects.

Predictive As A Path For Growth

I enjoy working in the predictive space. I'm excited about where we are heading with machine learning, natural language processing, and artificial intelligence. I will take any opportunity to continue learning in this area.

Breaking Up Large Projects

It is necessary to break large projects into achievable phases. There are times where you must defend user experience, and there are other times when an alternative is easily found.

Hover & Blank States Can Delay Complexity

To minimize confusion for the user, the analysis is revealed in stages. Blank states and hover states were used to reinforce this information hierarchy and allowed the user to consume the data.

Left-to-right Consumption Eased The Learning Curve

Revealing the key influencers first in the top left of the analysis established this bar chart as primary data. Each new chart appearing to the right of this chart is driven by the previous.

Avatar
Bob Werner

As a systems thinker, I love breaking down complex problems into elegant design solutions

View Profile