As I’ve come to learn through my experience as a content marketer at Dataiku, being a non-technical person at an AI software company can be both a blessing and a curse. On one hand, the opportunities for learning are enormous, and with self-service analytics and dashboards at hand as well as more than enough data wizards around to help you, you rarely feel like your lack of technical skills is a blocker to your day-to-day work and decision making.
Yet, when you spend a significant amount of your time talking to (and, in my case, writing about) coworkers, customers, partners, etc. who all use Dataiku DSS to do really advanced and fascinating things with their data, sometimes you can’t help but feel like you’re missing out on the fun. This is exactly how I started feeling a few months ago, which is why I decided to finally start getting my hands dirty with Dataiku DSS.
1. Always Start With a Business Question
Since the start of my journey with Dataiku, I have had plenty of opportunities to get into learning about and using our product -- my onboarding session, watching our tutorials and product materials, talking to our brilliant engineers and data scientists, or now even following the learning paths of the Dataiku Academy. However, I didn’t feel like I was really learning enough (and, to be honest, wasn’t putting in that much effort) until I identified an actual need for it in my day-to-day work.
The problem I was faced with is one that I’m sure many marketers in rapidly growing teams can relate to: siloed data coming from and stored in various marketing tools. As part of the content team at Dataiku, I’m in charge of our video content, which is published across a variety of platforms: from our website and landing pages, to our YouTube and social media channels, to third-party platforms such as BrightTalk which we use to stream our webinars.
On one hand, this is great because we get to experiment with different video formats and reach a wide and diverse audience, but when it comes to tracking and analyzing the performance of videos, this also means...well, a lot of random CSV exports. The video naming, references and metrics were documented differently in every platform, and having them stored in separate places prevented me from having a global view of their performance (and thus from being able to convince my team just how awesome videos are and that we should be investing more in them).
How I felt trying to analyze video data from 7 different Google spreadsheets
It was only once I acknowledged this specific pain point that I really gained the motivation and direction to start learning how to use Dataiku DSS for data analysis.
2. To Get Past the Initial Learning Curve, You Need to Get Over Excel
The initial excitement of downloading and opening Dataiku DSS (and the embarrassment of having to ask IT for a new license because yours expired 6 months ago) is followed by a slightly less exciting phase, which I like to call “getting over your ex(cel)”. Yes, I stand by this pun and I will take it to the grave with me.
What I mean by that is the frustrating and uncomfortable, but necessary period of getting used to a new tool and discovering all the basic functionalities and formulas that most of us already know how to navigate in Excel. Do not let the initial learning curve and the frustration that comes with it discourage you, though. I can assure you that Dataiku DSS can do all the things that Excel can, and much more than that, it’s just a question of getting used to a new user interface and finding where all the functions and formulas are.
3. Use All the Resources You Have
My next advice is kind of a no-brainer, but still important to note: try to use all the resources you have at hand to guide you through your first steps with Dataiku DSS. At the time when I first started using it, the Dataiku Academy still hadn’t been released, but I personally found the Dataiku DSS reference docs very helpful, and I also looked at tutorials and sample project walkthroughs. The latter are great for understanding the basic logic and frameworks behind creating a data project.
4. Don’t Just Let Your Data Project Sit On the Shelf. Try to Get Insights, and Use Them!
My final piece of advice for non-technical folks starting out with Dataiku DSS (and technical ones, too, for that matter) is to not just stop at performing a data analysis that more or less works. In order to really go from noob to a fully functional Dataiku DSS user, you need to “operationalize” your data project. By this, I don’t mean necessarily operationalization in the sense of having a full-on machine learning model running in production, but simply just using your data to answer your initial business question(s), draw actionable insights, and communicate them to your team.
Even though my video analytics project may not have been particularly advanced or impressive, it still allows us in the marketing team to get a global view of the way people consume our video content, and which types of videos work best on which platform -- something we weren’t able to easily and accurately analyze before. This is why I’ve made a point of communicating these newfound insights to my team, which ultimately made me feel like my first Dataiku DSS project was successful and brought actual value.
In the words of Kim Kardashian, not too bad for a girl with no talent :)
My next challenge is going to be collaborating with our marketing analyst to build a Tableau dashboard to visualize my data and make it easier for everyone at Dataiku who works with video content to draw their own insights. In the meantime, I encourage everyone who's frustrated with Excel spreadsheets and want to make their day-to-day work with data a little bit more exciting to download the free version of Dataiku DSS, like, yesterday!