<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=293842391037690&amp;ev=PageView&amp;noscript=1">

Astronomer Blog

Culture / Data Science / Dev / Growth

"Way More __________": An Interview with Dustin Larimer

This post was originally published on the blog for USERcycle.

Part two of the Keen IO Interview Series! We really enjoyed hearing from Stephanie Stroud in our last interview (which if you haven't checked out, you can and should read it here), but now Keen IO's Dustin Larimer is taking center stage. We first met Dustin when we visited the Keen IO offices and loved the perspective he brought to data visualization and design. His work at Keen IO as a Design Engineer is as interesting as his path to Keen IO itself. It was truly great to chat with him. Enjoy!


 You’ve had a very diverse past including getting an MFA from SCAD and consulting for companies at Techstars. What attracted you to join Keen IO as a design engineer?

Dustin Larimer: The awesome humans behind it all! I met the founders during Techstars Cloud, in San Antonio, and we became fast friends. I was invited into their Burning Man camp the following summer and suddenly found myself in this big goofy tribe with some of the most profoundly creative and inspiring people I’ve ever known.

While they were busy building Keen, I had been consulting as a “devsigner" for a handful of startups and cranking away on side projects of my own, but was hitting a wall. Contract work has its perks, but it also means brief, shallow involvement with whatever problems you might be trying to tackle. I think great design happens somewhat slowly, as insights and connections fall into place over time. I wasn’t able to really sink my brain into something for more than a few quick sprints, and that was making me a crappy designer.

I also realized most of my side project efforts were stagnating because I was trying to do everything myself.. customer dev/research, design, end-to-end engineering, networking and fundraising.. the whole bit. In hindsight I think it was pure ego driving and defending that approach. I was standing squarely in my own way of creating anything of actual value to anyone else. Nothing was working, and I was way past burnt out.

So, I made two lists: “No more _______” and “Way more ______”. The “No more” list quickly filled up with activities and busy-work habits that sapped my time and energy. The “Way more” list went in a different direction: more time with friends, introspection, creative conversation, writing and teaching. There wasn’t a single thing I was currently doing that made the “Way more” list. I knew it was time to make a radical change, surround myself with amazing people, and just see what I could do to help them. So that’s what I did, and it’s turned out to be one hell of an adventure.


What prompted Keen IO to start looking into and start building customizable dashboard templates? Did you find that creating their visualizations was causing frustration from customers? Was it a commonly requested feature?

DL: It actually came about from a series of support pool rotations. Everyone at Keen does support, so we all get to see first-hand what our customers are building. We noticed that most of the support questions regarding data visualization were part of an internal dashboard project of some kind. In the process of troubleshooting a chart configuration bug or some other issue we would catch a quick peek of what they were really trying to create.

Most of the dashboards we saw were unstyled HTML files with a few empty chart placeholders. Visual style wasn't a top priority.

The idea for the dashboard starter kit was simply to give our customers a shortcut; to make it quick and easy to create something that works well and looks nice. They were all starting from scratch and repeating a lot of steps to get started, so we just decided to see if we could take some of that work away.


The Keen IO dashboard repo has been starred over 6,500 times. Why do you think it caught on so widely?

DL: I wish I knew! I remember the day we made it public and shot out a few tweets. A few of us high-fived and cheered when it hit 100 stars, but we had no idea what was coming. It just kept climbing. Then we posted the repo on Product Hunt and it went through the roof.

The best part is how small the codebase actually is. We built this on top of Bootstrap, the well-known and well-documented layout framework. We considered other layout frameworks, but Bootstrap is the most ubiquitous and well known, and we wanted to minimize the initial learning curve. The custom styling behind our dashboard kit is just under 1.5kb of uncompressed CSS… less than 80 lines of code. I think there’s a fascinating lesson about product design and code sophistication in there somewhere, but I’m not entirely sure what it is yet :)


Keen IO has a number of recipes online for creating custom visualizations. How did you decide which visualizations to include? Are you considering adding any in the future?

DL: Most of those recipes were inspired by customer questions and popular integrations. We’re definitely planning to add more recipes in the near future, and will be going deep on open source data analysis and visualization tools as well. Tools and methods go hand-in-hand, and we’re putting a lot of thought into both.

Mike Bostock wrote an excellent piece on the power of an example that really resonates with us. Quick examples and how-to guides help users ramp up and learn quickly, creating positive mental associations between inputs and outputs. Our platform is very flexible and opens up a lot of new possibilities for recording and learning from event data, but those possibilities aren’t always explicit or obvious. A few good examples can help demonstrate that potential and spark all kinds of new use cases and inventions. We’re investing heavily in our customers’ imaginations.


With the rise of d3.js and other interactive frameworks (such as Highcharts), where do you think the future of data visualization is headed?

DL: I’ve noticed some really cool projects popping up that combine D3 and React, and I’m excited to see what new dataviz tools take shape there. I think (hope) we'll also see more attention given to mobile performance too. The JS ecosystem is moving pretty quickly toward an ideal of small, modular packages that do a single thing well, and big, monolithic visualization libraries don’t really fit in that world. I’d love to see what replaces them.

Another exciting trend is the rise of dataviz-as-a-service businesses. It’s content creation, business intelligence and some pretty awesome tech, all rolled up into one.


How can data and design be balanced without one overshadowing the other?

DL: Designers will always find access to new tools, better technology, and more data to work with. Modernization doesn't stop; that's a given. We just happen to have a lot of data to work with these days, and it’s putting interesting pressure on designers.

I think it’s important to remember that data is just one input, and hopefully not the only one. It should inform the designer’s understanding, but letting data drive the design process entirely is lazy and short-sighted. In some cases it can even be dangerous. At the same time, ignoring the data you do have is just as irresponsible.

In the end, good design will always hinge on how deeply you can drill down into what truly matters, and how quickly you can call out your own bullshit. You need to get your own biases and assumptions out of the way, and data can help with that.


Dustin is a product designer and engineer who spends his time helping people make things at Keen IO. Follow him on Twitter, Github, or Medium.