MAIN -Datascience@2x

Designing for Data

Designing for Data

Year: 2016

Role: Design Lead

Plus: Ashley Romo

Methods: User Flows, Site Flows, Information Architecture, Low Fidelity Wireframes, Interface designs, UI Annotations, UI audits, Style Guide Creation, Design QA, Iconography, User Interviews, User Testing

Year: 2016

Role: Design Lead

Plus: Ashley Romo

Methods: User Flows, Site Flows, Information Architecture, Low Fidelity Wireframes, Interface designs, UI Annotations, UI audits, Style Guide Creation, Design QA, Iconography, User Interviews, User Testing

Year: 2016

Role: Design Lead

Plus: Ashley Romo

Methods: User Flows, Site Flows, Information Architecture, Low Fidelity Wireframes, Interface designs, UI Annotations, UI audits, Style Guide Creation, Design QA, Iconography, User Interviews, User Testing

Overview

In mid-2016, Datascience Inc. was about to make a pretty big pivot. What started off as data science as a service turned into data science as a tool. We were tasked with building the complex tools data analysts and scientists needed...and we only had 12 months to design, build and ship. This is what we ended up with.

Overview

In mid-2016, Datascience Inc. was about to make a pretty big pivot. What started off as data science as a service turned into data science as a tool. We were tasked with building the complex tools data analysts and scientists needed...and we only had 12 months to design, build and ship. This is what we ended up with.

Overview

In mid-2016, Datascience Inc. was in search of a product space. What started off as datascience as a service turned into data science as a tool. After a pretty big pivot, we were tasked with building the tools data analysts and scienists needed in this modern age. Oh...and we only had 12 months to design, build and ship. This is what we ended up with.

DataScience-Project-Image-1@2x
DataScience-Project-Image-2@2x
DataScience-Project-Image-3@2x
DataScience-Project-Image-4@2x

Look & Feel: Where Do We Start? 

Look & Feel: Where Do We Start? 

Being able to define how a product looks and feels from scratch is a rare luxury. At DataScience Inc., I was able to do just that. Sans any marketing team, sans over zealous ( but passionate ) engineers, sans misguided ( but passionate ) stakeholders...what would us designers do if left to our own devices?

Design systems. Yep. That's pretty much where we started. Who would have thought us, designers, as rational animals? I have refined my process in building design systems since then but I still stand by a lot of my decisions. Some things I'd repeat 🔄 forever: a standard system of measuring ( units of 8 ), tight typographic systems ( no more than 4 styles per typeface ), making icons in a 24pt box. One thing I'd try to forget: using a baseline grid.

Being able to define how a product looks and feels from scratch is a rare luxury. At DataScience Inc., I was able to do just that. Sans any marketing team, sans over zealous ( but passionate ) engineers, sans misguided ( but passionate ) stakeholders...what would us designers do if left to our own devices?

Design systems. Yep. That's pretty much where we all start. Who would have thought us designers as rationale animals? I have refined my process in building design systems since then but I still stand by a lot of my decisions. Some things I'd repeat 🔄 forever: a standard system of measuring ( units of 8 ), tight typographic systems ( no more than 4 styles per typeface ), making icons in a 24pt box. Things I'd try to forget: using a baseline grid, using Lato :)

DS-TYPE
DS-LAYOUTS
DS-ICONS
DS-COLORS
DS-LOGOS2
DS-COMPONENTS-revised@2x
TOP

There are 3 pillars to large scale product design: scalability, manageability, and common practices.

There are 3 pillars to large scale product design: scalability, managability, and common practices.

BOTTOM

The Art of Combining 

Initially, we had several core products all housed in their own respective apps. One of the more difficult tasks was collating them all under one umbrella. We all know combining things is hard unless it's a Taco Bell and a Pizza Hut. In most cases, the main struggle is structure. A wrong decision or two later and your app is riddled with information architecture problems.

At DataScience we solved some of these IA problems by doing half a dozen customer interviews and leveraging guerilla-style UX research activities like card sorting. We even managed to corner some potential users at the IBM World of Watson conference. Never underestimate the value of swag when you're trapped inside a monolithic conference hall. 

We opted for side navigation because it carried more visual weight and was easier to scan. One benefit we had was that most of the core products were in some form of production and it was easier combining things we know worked. Once these structure issues were solved we put some color and aesthetics on that structure by way of our design system. This was hard sometimes...but a good design system has some give to it. 

Another challenge of combining different tools/products was working with multiple teams. Combining products is one thing...combining teams is another. We designers were few. In order to keep our sanity, we had to set up a cadence. Every other sprint we delivered to a different team...since it takes a bit longer to execute designs on the engineering side...somehow this worked out. One useful thing we did was set up a whiteboard with our sprint goals. This kept us on track and reassured stakeholders that we were running on time.

Initially we had several core products all housed in their own respective apps. One of the more difficult tasks was collating them all under one umbrella. We all know combining things is hard unless it's a Taco Bell and a Pizza Hut. In most cases, the main struggle is structure. A wrong decision or two later and your app is riddled with information architecure problems.

At DataScience we solved some of these IA problems by doing half a dozen customer interviews and leveraging guerilla style UX research activities like card sorting. We even managed to corner some potential users at the IBM World of Watson conference. Never underestimate the value of swag when you're trapped inside a monolithic conference hall. 

Once these structure issues were solved we put some color and aesthetics on that structure by way of our design system. This was hard sometimes...but a good design system has some give to it. 

Another challenge of combining different tools/products was working with multiple teams. Combining products is one thing...combining teams is another. We designers were few. In order to keep our sanity, we had to set up a cadence. Each other sprint we delivered to a different team...since it takes a bit longer to execute designs on the engineering side...somehow this worked out. One useful thing we did was set up a whiteboard with our sprint goals. This kept us on track and reassured stakeholders that we were running on time.

Remaking a Classic

One tool that is ubiquitous in data science work is the Jupyter Notebook. A Jupyter Notebooks is a tool "that blends computations, outputs, explanatory text, mathematics, images, and rich media representations of objects." At DataScience Inc., we determined that a bespoke Python notebook integrated into our platform of apps would be handy for our users. 

Remaking a Jupyter Notebook was no small feat. The one issue was updates. Whenever Jupyter updated their notebook, we had to QA ours to ensure nothing broke. Ultimately, this maintenance time was too big of a time suck and we ended up sunsetting the DataScience Notebook once our platform launched.

Sold to Oracle

During my tenure at DataScience it was evident to me that we were courting some bigger clients. While large companies like Facebook and Google used their own proprietary data science tools there were still some companies looking to build or buy a toolset to suit their needs.

We had successfully transitioned our current clients to the new platform and, when I left for a different opportunity, DataScience Inc. was on track to launch the platform as expected. In the following year, it was announced that Oracle had purchased DataScience for an undisclosed sum. 

It was a fun, challenging and sometimes harrowing experience building a product on such a short timeline. We had our moments in the sun and our cloudy days but design played a crucial role in the sellability of the product. In the end, it all paid off and I'm happy for my former colleagues who found success 👏.

During my tenure at DataScience it was evident to me that we were courting some bigger clients. While large companies like Facebook and Google used their own proprietary data science tools there were still some companies looking to build or buy a toolset to suit their needs.

We had successfully transitioned our current clients to the new platform and, when I left for a different opportunity, DataScience Inc. was on track to launch the platform as expected. In the following year it was announced that Oracle had purchased DataScience for an undisclosed sum. 

It was a fun, challenging and sometimes harrowing experience building a product on such a short timeline. We had our moments in the sun and our cloudy days but design played a crucial role in the sellability of the product. In the end it all paid off and I'm happy for my former colleagues who found success 👏.

Extras

Here's some videos the marketing team made to help with some of the actions you can take on the platform:

Here's some videos the marketing team made to help with some of the actions you can take on the platform:

More projects

Here Comes the SunProduct Design

Walls of InteractivityProduct Design

Funnel DynamicsProduct Design

Simplifying the ComplexProduct Design

A House for EonsProduct Design

Doober!Product Design

Jobsite InspectorProduct Design

Designing for the HungryProduct Design

Find Your FitProduct Design

Everything ElseProduct Design

logo_footer

© 2022 Jonathan Brazeau

© 2022 Jonathan Brazeau

© 2022 Jonathan Brazeau

© 2022 Jonathan Brazeau