A Lifetime of Breaking Things, Maybe Fixing Them, Brought Me to Data Science
When I was a kid, I was the kind of kid that took things apart, but not in that “kid genius” way. I liked tinkering with things. I never got to the point where I would disassemble something, but I’d just take it apart in small bits and see what happened without them.
98% of the time I tinkered with something until I realized it was about to break irreversibly. The other 2% I didn’t realize it was about to break. I tinkered with my toys, I tinkered with my clock radio, I wouldn’t touch my Nintendo because that’s where video games come from, but I certainly tinkered with the family computer. I got familiar with how all those things functioned and what I could do with them. I didn’t put things together because I knew they would work again, I had to figure out what I did after making something work.
“Breaking” things in the house became so synonymous with me to my parents that I got phone calls in my 30s from my father asking me what I did to his computer because it stopped working as expected.
In school, I wasn’t what you’d call a “great student” when it came to my academics. I lost focus and got bored pretty easily in a traditional classroom setting and momentum pretty much took over from there. Music made sense to me. You could practice in order to change or perfect a performance. I got a four track cassette recording and began recording everything I could. I played with every setting, every dial, and every microphone I could get my hands on and I wasn’t happy with what I was coming up with, but I wanted more. So I took my “great student” attitude to college.
I went off to college majoring in Sound Recording Technology. It sounds fancier than it was, but I got my major in music performance and audio engineering. I took my guitars apart, every recording studio session was an exercise in trying to find the best sound for the situation for an endless amount of hours. I joined countless bands, recorded hours of music, and built weird little websites from the html I could teach myself. I graduated as computers were becoming staples inside recording studios and the explosion of home recording meant I could tinker to my hearts content in my apartment, undisturbed.
Wordpress’s first release was in 2003 and by 2006 I was building my own website over and over again. I assumed everything was html, even when I was using CSS and php to control how the site looked and MySQL to maintain the content. While I worked in studios, live venues, and at home, I also kept learning and hammering on Wordpress. Breaking it and fixing it. Breaking it and fixing it. Breaking it to the point of no return and starting from a backup.
I moved to New York City, hired to start a podcast network for the now defunct New York Sun. I set up a recording studio, began recruiting voice talent from the news room and recording, but while there I also wound up managing the website’s redesign, installation of Drupal, information architecture overhaul, and editorial system installation. I leaned heavily on my tinkering and breaking of the family computer, my own computer, and what I gleaned from working with Wordpress. I was the person generating early Omniture and Google Analytics reports. I accidentally turned all the writers against each other when I started posting pageviews of the previous day’s articles. At some point, I wasn’t referred to as the podcast guy anymore.The Managing Editor of the paper introduced me in meetings as being “from the internet.”
It was the analytics side of things that I latched on to. It immediately made sense to me while working with Steve Krug on A/B testing and working on speeding up the site with the development teams and encouraging user behavior through design changes.
The New York Sun’s business was still a newspaper first and a website second. As was the popular thing with newspapers in 2008, the business changed and the company didn’t. At the same time, it was the home recording boom that enabled me to work virtually anywhere as an audio engineer that also started putting recording studios out of business. I took my small experience with Wordpress, analytics, and the emerging social media and rolled it up into a new career in marketing.
I built on my experience at Symphony Space, managing the ticket information SQL database of thousands of live performances and an early music streaming service with tens of thousands of recordings and the website built in php, using site analytics to drive design decisions. I livestreamed performances, I designed (thanks to a decade of pirating Photoshop) and managed the development and release of iOS and Android apps.
My career was growing and my strongest asset was my use of analytics to help all other aspects of my job. I developed my own metrics for social media and email newsletters to better make content decisions based on audience reaction.
Engagement / (Reach (not impressions) / 1,000) = Engagement Ratio
I was never “trained” to do any of the work I was doing. I just did it. My strengths were the ability to make data-driven decisions, my weaknesses were my inability to speak the jargon my peers did.
In my last job as a Director of Audience Development, I was asked to put together a model three months to three years of email newsletter and site traffic from the newsletters. I spent a month working on a spreadsheet that never worked. I quadruple checked my work, trying to draw conclusions from the unsubscribe rate, signup placement, timing of the lightbox form on the site, links per newsletters, newsletters per week, and likely a dozen more features.
It was the most frustrating work I have ever done in my life. I felt angry that I was close but still completely wrong. I tried a few different ways to make it work, but everything I was doing was clearly wrong based on the results I was seeing.
When I was let go from that job, I job searched for a while but my inability to create something useful stuck with me. Why couldn’t I do that? Turns out the answer is more complicated than I had imagined.
I made the decision to learn how to actually build out that model and continue tinkering, but this time with some education.
