Being a Data Analyst at Jellysmack
In honor of my one-year anniversary at Jellysmack, I thought it was the perfect time to look back at the past year and reflect: what does it mean to be a Data Analyst at Jellysmack?
✨The Data Department
At Jellysmack, there are currently 30 people in the Data Department, and it is expected to double this year.
Three different professions are represented: Data Scientists (DS), Business Analysts (BA), and Data Analysts (DA). BAs and DAs are both part of what we call the Analysis Team. It’s a diverse crowd: from engineers in Big Data to graduates of business schools, to graduates in Math and Economics.
🤝 The role of the Analysis Team: an interface between Business and Data
Data plays an important role in the business decisions at Jellysmack. It allows to provide indicators, determine trends, and measure the consequences of the decisions we’ve made.
However, some problems remain: which pieces of data, amongst the 50 billion+ collected by Jellysmack, cater to our Business requirements? And how to use it to answer efficiently?
To find out, the Analysis Team regularly follows a sequential process allowing them to leverage the talents and skills of both the Business and Data Analysts.
- Business understanding: Analysts have to understand exactly what the business question is, what kind of data it requires and what problems it allows to solve. After all, the role of the team is both to provide the requested data and to use their expertise to suggest new indicators and new approaches aligned with the business question.
- Exploratory Data Analysis: Jellysmack has several terabytes of data and it is essential that the Analysis Team can explore the data to master the datasets and monitor the quality so they can provide the Business with a coherent and reliable answer. In certain cases, the Data Analysts have to aggregate and organize existing data into a new easy-to-understand graphic to better understand the dataset and therefore address the problem.
- Transformation and restitution: whether it’s in a spreadsheet or in a database, raw data may not mean much to us mere mortals (or at least to those of us outside of the Data Department) so the team’s role is to transform this raw data into information that is valuable and actionable for the business. To that end, the way the information is restituted is also essential. A presentation that uses the proper graphs depending on the information to present, and in a user-friendly way, allows emphasizing the key points that respond to the Business requirement.
The role of the Data Analyst
- 🔎A tour guide in a labyrinth of data
As stated above, Jellysmack has several terabytes of data and gets more and more every second (about 6% more on average). It is consequently essential for our DAs to know the Data Model inside and out: where to look for which data and how to analyze it.
That’s why it’s important for Data Analysts to be able to write high-performance SQL queries, to locate and retrieve the data they need.
- 🧪An alchemist on a perpetual quest for new data
Although Jellysmack has very recent data, its rapid Business growth constantly calls for new indicators and new data to tap into all the time. We can pursue two different avenues:
- We can use existing data to create new data (for example, by comparing and aggregating data coming from Youtube and Facebook).
- We can look for new sources of data and expand Jellysmack’s Data Catalog. We can then transform the data in Python. Transformations are based on a variety of concepts such as Asynchronous Networks or the application of mathematical and statistical methods.
- 🎨A Data Visualization Artist
Making quality data available and explaining why it’s relevant is the first step in a DA’s work. However, their job doesn’t end there. It’s about more than just setting up an Excel spreadsheet.
Working in tandem with the BAs, a DA has to fully understand the initial Business request so that they can present a coherent data visualization. The visualization presented must allow the Business to immediately see the answer to their question and dig deeper and explore new approaches to analysis.
A Data Analysis Project 📊
Data Analysis projects are often initiated by Business requests. In this case, the original request was: What theme keeps showing up again and again in the videos that we publish on Snapchat?
The sequential process detailed above was used for this project.
In the Business request, we’d been asked to identify the topics that showed up again and again in Snapchat videos published by Jellysmack, but after further examination, we realized that another important piece of the puzzle was missing because the goal of the analysis was actually twofold: not only to identity the topics that kept coming up, but also to identify which topics had the biggest impact. This brought a new comparison and performance aspect into the mix.
The Exploratory Data Analysis was easy enough: we chose to focus on the tags that Jellysmack used on Snapchat. So, it was easy to find out which words came up the most often. But, frequency didn’t tell us anything about performance. So, to analyze performance, we had to take an extra step and transform the data with Python. We calculated a Z-score for every word that appears in the tags, based on the performance of the video that it’s linked to. By aggregating the data of all of our videos (on all of our channels), we’ve been able to figure out which keywords are associated with the best performance.
At that point, we had calculated data from our raw data. The next step was to determine how we would present the results to the Business requester in a clear and meaningful way. We chose to present two graphics and a PowerBI report:
- First, a chart presenting all of the words in the existing tags, categorized by Z-score, starting with the highest (100 being the word with the best performance). The user could then see all of the tags and prepare their performance.
- But, in order to make the information easier to read and more aesthetically pleasing, we also presented a word cloud showing the 30 tags with the best score.
We now had two graphics allowing us to respond to the initial demand:
Join the team! 🤸♂️
I hope that this article has given you an idea of just what’s involved in Data Analysis. The Analysis team is constantly growing and is always looking for new talents. If you’d like to learn more about Jellysmack or join the Team, please don’t hesitate to get in touch.
You can find all of our job offers here: https://jobs.jellysmack.com/
