Based on the title, you might think I’m about to describe ways in which companies get data science wrong, misunderstanding what data science is supposed to mean and how to do it. And while that’s a perfectly valid blog post, and I’ll touch on a few of those themes in my tips for 2015 later on, for now let’s think of the title a different way. Think metaphysics: data science itself is a new, rapidly evolving type of perception of business reality for 21st century companies.
Consider an old-fashioned perception: hearing. Sound waves are constantly barraging us, and our highly evolved and complex ears are responsible for capturing that data. But without our brains, that sound data would be nothing—just noise, static, snow. By filtering the sound waves, identifying both patterns and anomalies, and interpreting, our brains give us the ability to perceive the world around us—they allow us to hear.
Modern companies collect vast amounts of data, like ears do, but interpreting that data correctly is what provides the accurate perception of business reality, turning data from static noise into valuable insights. When a company’s decisioning is data-driven, those decisions will be stronger and more reliable, and will lead to better results. Using the hearing analogy: I make better decisions about crossing streets because of my ability to hear 10-ton trucks coming, for instance. Similarly, data products and algorithms based on rich, accurate interpretation of data are more successful; my singing is much more likely to be on pitch if I can hear the rest of the karaoke music.
At growing companies, data science isn’t usually an initial focus; existential issues like finding market footing take precedence. But eventually, many companies realize that they need to reinvest and focus on their data—and that’s exactly what’s happened at AppNexus. It’s been an exciting process to be part of. We’ve evolved from using byproduct data almost as an afterthought, to bringing data front and center, making a deliberate, strategic investment in developing our data assets to their fullest potential.
Today Data Science is its own function at AppNexus, with a dedicated hardware cluster and ever-expanding expertise in data science tools and techniques. We’re more ready than ever to perceive whatever our data has to tell us about the reality of our ecosystem, our clients, and ourselves. And just in time! You may have heard that 2015 is the year of the Ad Tech Power Game—a.k.a. #ATPG—and what could be more useful than a new found set of ears (so to speak) to help us help our clients navigate?
As you prepare for the year of the #ATPG, here are a few tips:
- Data science isn’t just a buzzword, it’s a means of perceiving business reality. Be aware of what your data is telling you about the world. Don’t just look at the graphs that have the trend you’re looking for—be honest with the data, be rigorous in understanding it correctly, and then listen to what it’s telling you.
- For algorithmic uses of data, remember the 80/20 rule: while there are many sophisticated, modern tools you could use to do predictions and modeling, in most cases that complexity isn’t necessary in order to get 80% of the return. Avoid the hype of “deep learning.” Avoid the hype of buzz words. Know that it’s not always necessary to have the latest and greatest tools. Some of the best, most tried and true algorithms and techniques for extracting value from data have been around for decades.
- Finally, know that data science is fundamentally creative. When you’re hiring, look for creative problem-solvers. There’s no boiler plate, no one exact set of skills that makes a good data scientist—it’s about whether and how she can use data to solve problems creatively, to glean insight and translate that into business value.