While the marketing industry has been talking about “big data” almost non-stop for the past five years, many publishers and marketers are still having trouble putting together capable analytics teams and establishing workflows that allow them to get the most out of their data.
As a follow-up to Brian O’Kelley and Catherine Williams’ conversation on the uses (and misuses) of big data, we asked three key members of the AppNexus data science team to give us their best tips for finding and activating data-driven insights that improve business results. Here’s what they had to say:
Build a data science team with a diverse skillset.
“I think an ideal data science team should really be a mix of people with all different kinds of skills. There should be someone who is really sophisticated in terms of coding and engineering knowledge. There should be someone with a statistical background. And there should be someone who understands client needs because, a lot of times, the client’s asks will be very different from what they actually need. Data science teams can get in the weeds, which makes it hard to pull back to see the larger perspective and effectiveness of whatever it is they’re doing. So, having that mix of personalities and backgrounds creates an opportunity to catch each other when that does happen.” — Liz Zoidis, Lead Client Insight Analyst (Publisher Technology Group)
“Two qualities that make a good data scientist is being ego-free and auto-didactic. They should be able to go to their peers and give and take knowledge without worry about how they’re going to be judged. If you’re able to learn and ask questions, you’ll do really well.” — Adam Petranovich, Data Scientist, Buy-Side R&D
When investigating your data, start small and know what you’re looking for.