June Dershewitz has spent year 20-year career driving data and analytics strategies for industry-leading companies, including Fortune 500 corporations and tech startups.
She is the Data Strategist at Amazon Music, as well as President of the Board of the Digital Analytics Association. She previously held the role of Director of Analytics at Twitch, Director of Digital Analytics at Apollo Education Group, and Vice President of Analytics at Semphonic.
In 1999 I left academics and landed my first job as a web analyst at a startup in San Francisco. There was no template for me to follow, so I made everything up as I went along. Looking back, it was a really exciting time to start a career as a data person. In my role as Director of Analytics at Twitch, I get to solve data problems that have a high degree of ambiguity, and I still have that pioneering spirit.
In the work I do, I’ve often found myself acting as a bridge between technical and non-technical teams. Helping very different people develop a shared understanding of a problem space is a fun communication challenge. Have you ever explained multi-touch attribution to a marketer? If so, you’re a bridge.
I get a lot of energy from participating in industry groups that extend beyond my day job. I’m currently on the board of the Digital Analytics Association, an organization that focuses on professional development and community for data analysts. I’m proud to say that we’re hosting our first large conference this fall. I’m looking forward to meeting people in real life that I’ve gotten to know online.
J.D.: At my company Twitch, we collect and record 60 billion event records every day that describe customer engagement with our product and platform. That’s a lot of data! One of the challenges we face is that traditional ways of shaping and managing data won’t always work on data sets as large as what we’ve got today. We’re still able to find ways to meet customer needs when it comes to reporting, analysis, and data exploration, but the size of our data definitely influences the choices we make. Even some of our aggregated data sets have billions of rows. Based on what I’ve heard from industry peers, I know I’m not alone.
J.D.: Make sure that your analytics practice has its priorities straight. Periodically examine the data-related activities that occur in your company and reflect on what you’ve achieved as a result. There are only so many hours in the day. Are you doing the things that will have the biggest impact on your organization? It’s easy to get caught up in tactical work that sacrifices long-term impact for more immediate gain. Watch out for that.
Here’s an example. A new analyst, left to their own devices, might engage in an ad hoc request cycle resulting in a very literal set of deliverables for each question they receive. You want a data table? Here’s a data table. Not good! However, what if that analyst limited their reactive work in favor of a proactive effort to enable self-service for a whole class of related questions. That’s leverage! It takes discipline to solve the general problem, but it pays off in the end.
J.D.: Jim Barksdale, the former Netscape CEO, once said, “If we have data, let’s look at data. If all we have are opinions, let’s go with mine.” Without data we are left only with opinions, and that is not a place of advantage. We could apply that sentiment anywhere in a business, but it’s especially true in marketing.
The most powerful application of data within marketing is when it gets used to drive a company’s overarching media investment strategy. How much should you spend on paid search? Or social? Or video? In total, or at the campaign level, or at the creative level? How do you justify all of those choices and tune them over time to reach the optimal outcome? With data, that’s how.
J.D.: I dislike the term ‘hacks.’ Transparent solutions are always better than hacky ones. If you feel you need a hack accomplish something, what are you working around? If there’s a more straightforward way to achieve the same result, do that instead.
J.D.: Company-wide KPIs are - and will always be - at the top of my list. However, I also place a lot of value on the metrics that we use to mark progress toward our team-wide objective, which is self-service access to data. Here’s how we measure it:
We periodically run a company-wide survey where we ask staff if they’re able to get data in a timeframe that meets their business needs. Our stakeholders are very data-driven and quite demanding. We set targets on satisfaction as our primary outcome.
We track monthly active users of each tool in our self-service analytics portfolio - think Tableau and a few other data interfaces. Data should be accessible to everyone who needs it in a way that is pleasant to work with. People vote with their feet.
This is basically the number of defective rows in a data set divided by total rows. For example, if there’s a 5% defect rate it means that 5 rows out of 100 contain some garbage. This metric is important because self-service tools are only as good as the underlying data.
J.D.: Look beyond your own company and connect with others. The DAA has a mentoring program where analytics professionals can pair up to learn new things and advance in their careers, and I’m completely sold on the value of mentoring.
A few years ago I mentored a data scientist who was thinking about joining a new team, and I helped her find the courage to say yes to that opportunity even though she felt it was at the brink of her abilities. Sometimes it takes a second set of eyes, or an extra boost of support, to help someone find a new way to drive impact.
J.D.: In the past couple of months, Tableau got acquired by Salesforce ($15.7B) and Looker got acquired by Google Cloud ($2.6B), and I don’t think this trend is over yet. My prediction is that we’ll continue to see the consolidation of vendors in the analytics space through the end of this year and into 2020.