My name is Tommy O'Connor. I'm a user interface, motion, and branding designer based in Flagstaff, AZ.. I've worked with notable clients like GE Healthcare, Adobe, Disney, Universal Pictures, and even the UFC. I co-founded a local technology group, Flagstaff Tech Track. I love to take on big, complex challenges. I also have three small kids, hence, I fear nothing.

I'm always interested in new opportunities. Let's talk.
Curriculum vitae

  1. 1. Compellon
  2. 2. GE Healthcare

* Compellon *

My friend Marc Bir joined Compellon as their CTO in 2013, and contacted me shortly after. At the time Compellon had a rough prototype, and the beginnings of user interaction mockups from another design agency. Q: The problem they were solving? A: Data analytics for the masses. In a nutshell, what was proposed to me at our first meeting was, "We want to create a data analytics platform that is as easy to use as MailChimp. We want casual users to be able to pop in sports data and gleen new information from it - conversely, we want Western Digital to be able to dump 13 gigs of hard drive manufacturing data into our system and figure out why certain drives fail."

compellon v2 sketch

We knew right away that the mockups they had originally commissioned weren't flexible enough to give their users enough depth of power. They were friendly looking, but that was about all. We decided it was worth it start at the beginning and solve the problem by understanding a potential user's goals in exploriong complex data.

Honestly, the first thing we ended up tackling wasn't design, it was vocabulary. A lot of the vocabulary was understood by the developers and data scientists - but would frustrate and confuse anyone who wasn't either of those. We came up with a pretty good solution: if I didn't understand it then we needed to re-approach it. Essentially, I was the village idiot. Solving the vocabulary issues was a very difficult task because we're working with complex concepts and for the sake of the casual user we needed to provide clarity without dumbing down the application.

compellon v2 sketch

We then built the v1 version of the application very iteratively, with a design theme I like to call utility-chic. We knew that with any startup came exploration, wrong turns, ah-ha moments, so we wanted to implement and iterate quickly without a lot of fuss. When we started v1 we really only had loose concepts of how users will want to work with their data. We knew that Point A was: getting their data into the service, we knew Point B was: a user understanding their data in new ways - but, we didn't know how they would let the system know what they were interested in, and how we could package those interests into sets that could be tweaked, shared, and then how we could make it prescriptive in order to give them suggestions on how reducing some variable might increase another.

What we were building was something really unique in terms of design concepts. For some of the interface solutions there really were no established patterns we could turn to and say, "Oh yeah, so we need to use X component to let users work with Y visual concept." Also, color becomes an issue when you are expressing positive and negative values. When it comes to positive values, we (and the rest of the world) knew that green was the right color to express the concept of positive valuation. Conversely, if you thought red was the obvious color choice for negative values then you'd be wrong, as I was. The thing about negative (the number) is that it's not always negative (the feeling), sometimes in data a negative is actually a positive, and red is a color that comes with way too much emotional baggage - so we chose purple and let red be reserved for application states: notifications, errors, and the obligatory "Are you sure you want to delete this?" dialogs.

We recently built the v2 version and made some major UX improvements. We took the knowledge we've gained about how users are exploring their data and we made the system a lot more geared towards their goals. We were also able to make v2 a lot prettier and streamlined.

I've been told that I'm supposed to talk about the quantifiable successes of the projects I work on. Compellon recently closed a Series A funding round for $7MM. They're up to 25 full-time employees and signing contracts with some brands your grandmother would even recognize. They just took second place at the prestigious O'Reilly Strata Startup Showcase.

* GE Healthcare - Amersham WB System *

Once I processed the details for this project I became instantly intrested. It had all the facets of the type of project I always fall for: impossible deadline, complex subject matter, and a big name client. Just like a baseball player who can't help himself from swinging at the high fastball, this is the type of project I can't say no to.

Here's the summary of the project, and what we did with it in respect to research: GE Health created a product that will run a Western Blot test in a fraction of the time (a day) than it takes for scientists to do it manually (up to a week). The problem they anticipated is that scientists would be reluctant to switch to GE Amersham machine since they each view their Western Blot process as unique, akin to an artform<. During the course of my research I came to the conclusion that although each scientists sees their process as art, the desired outcome is cosistency. Much in the way that photographers used to see their time in the dark room as part of the process, the desired outcome was a properly developed shot - digital has since killed the dark room.

That impossible timeline I mentioned? I had six weeks to launch (this deadline was a hard date with basically a kill term which meant if we didn't meet the date they could take my children - I like my children). So I went to work quickly to assemble a small team, establish tech requirements, understand GE's server capabilities, and trying to outline the deliverables as best as possible. The first really important task was to figure out how to convince scientists that a machine process was better than their manual process - or, at the very least convince them that they should entertain such a thought.

We ultimately came up with an idea to help the scientists see that while their process may be unique, the real goal was to achieve consistent results. We designed a questionaire that allowed the scientists to tell us about their Amersham process, after they told us their process then they could see the results from other scientists who had participated. We wanted scientists to see that while their process was unique, it wasn't THAT unique. We also believed that if a scientist saw others interacting with the new technology through this microsite, even though they weren't commenting ON the technology, then it helps verify that the new system was in the process of being accepted as normal.

Applying design to this solution was a little tricky from a user experience standpoint. For this plan to work we needed a good amount of participation - we had to make the percieved barrier in completing the questionaire feel low. We also wanted it to be graphicly pleasant, and to tease the user farther down the page. To establish GE as a partner we made the questions flow in the standard steps of a Western Blot, showing the scientist that we're operating from a base of knowledge.

Bottomline: We met our deadline, and I got to keep my kids. I won't sugar coat it though, that was a rough six weeks. In the end the project turned out great, we had tons of participation, and GE has sold quite a few Amersham WB Systems.