Making Sense of Data for E-Commerce Growth

By Lillian Pierson

Data science in e-commerce serves the same purpose that it does in any other discipline — to derive valuable insights from raw data. In e-commerce, you’re looking for data insights that you can use to optimize a brand’s marketing return on investment (ROI) and to drive growth in every layer of the sales funnel.

How you end up doing that is up to you, but the work of most data scientists in e-commerce involves the following:

  • Data analysis: Simple statistical and mathematical inference. Segmentation analysis gets rather complicated when trying to make sense of e-commerce data. You also use a lot of trend analysis, outlier analysis, and regression analysis.
  • Data wrangling: Data wrangling involves using processes and procedures to clean and convert data from one format and structure to another so that the data is accurate and in the format that analytics tools and scripts require for consumption. In growth work, source data is usually captured and generated by analytics applications. Most of the time, you can derive insight within the application, but sometimes you need to export the data so that you can create data mashups, perform custom analyses, and create custom visualizations that aren’t available in your out-of-the-box solutions. These situations could demand that you use a fair bit of data wrangling to get what you need from the source datasets.
  • Data visualization design: Data graphics in e-commerce are usually quite simple. Expect to use a lot of line charts, bar charts, scatter charts, and map-based data visualizations. Data visualizations should be simple and to the point, but the analyses required to derive meaningful insights may take some time.
  • Communication: After you make sense of the data, you have to communicate its meaning in clear, direct, and concise ways that decision makers can easily understand. E-commerce data scientists need to be excellent at communicating data insights via data visualizations, a written narrative, and conversation.
  • Custom development work: In some cases, you may need to design custom scripts for automated custom data analysis and visualization. In other cases, you may have to go so far as to design a personalization and recommendation system, but because you can find a ton of prebuilt applications available for these purposes, the typical e-commerce data scientist position description doesn’t include this requirement.