Comment by mettamage

A few fun examples that I do as "data analyst" for the marketing department.

1. Writing my own graph/tree algorithms, so I can visualize traffic flows better than whatever Google offers, detecting key drop off points, bringing that to the relevant marketing people so they can come up with ways to combat it.

2. An email campaign needs to be sent out, but the key person that knew about it left. I dive in and realize the whole data source has been phased out. I go into the huge org I'm at, connect to the right people (my managers have no idea), get the access rights to the correct data source. The data engineer is on vacation, I dive into our Airflow project, look at how our dags are made, create my own dag + needed custom operators, and now the email campaign can be sent out, on time.

3. Someone asks me to research something with minimal information (regarding media spend and outcomes on that). So I go research it, turns out the conclusions are very sensitive. My colleague did the research before me and he looked at my conclusions and realized I came to the same realization. Together we come up with a strategy how to effectively communicate this such that these insights strenghten the marketing team. My analysis was a bit complex, so instead of creating a compelling visualization, I created a small plotly application (in a notebook) with the right visualization metaphores and right filters.

4. Programming small free apps that drive traffic, but are also useful app in their own right (think stuff like an advanced tax calculator, etc.).

5. Using Airflow, certain APIs for solid data and LLMs to optimize our SEO pages. The last step in the pipeline is an actual human reviewing it critically and editing the text. It saves them 80% of work.

Currently collaborating on a project where we look at how LLMs can be a mini data analyst and retrieve data. My colleague got this project, but we realized it's better if we collaborate on it.