Building the New Redivis Demo Video

During the summer of 2020, I had the opportunity to work at Redivis, a company that provides organizations with a platform to upload, organize, and query datasets.

My role as a data wrangler and storyteller intern was to create a revised product demo that would highlight Redivis’ unique features and capabilities of working with data. This new demo, targeted at organization administrators and new researchers, needed to have a clear, story-like narrative that would be easy for users of all background levels to understand.

After researching other data platforms’ marketing strategies and discussing with the Redivis team, it became clear that one of the most effective ways to communicate a product was through a video.

With the help of the team, I spent the last three weeks of the internship creating a targeted, concise video walkthrough of the Redivis platform. Throughout this process, I gained new insights into data discovery and analysis, communication, marketing strategy, and video making.

Step 1: Defining the story

Before filming the video, I needed to develop an engaging story that would highlight Redivis and its applicability to researchers’ needs. After conducting stakeholder and user interviews, I developed several criteria for evaluating the story.

The story should:

• Be concise
• Feature publicly available data that was large enough to show performance
• Appeal to researchers and administrators
• Follow an easy-to-understand, relatable narrative
• Most importantly, contain an aha moment where the narrative came together with impactful visualizations

With this in mind, I began searching for the perfect dataset. I compiled a handful of potential datasets and stories and received feedback from the Redivis team on each one. Ultimately, we landed on the Center for Medicare & Medicaid Services (CMS) Synthetic Patient Dataset, which I found through Google’s BigQuery public data. This dataset contained (simulated) medical information for over two million patients and is similar to the far more restricted and identifiable CMS data. The dataset was ideal not only because of its large size but also because many Redivis users work in the medical field. I decided to craft a narrative around this dataset to tackle the following research question:

What is the county-level relationship between the opioid prescription rate and unemployment rate from 2008 to 2010?

Step 2: Building the tables and visualizations

After uploading this dataset to Redivis with the proper metadata and documentation, I began manipulating the data to see how I could communicate a clear story whilst showcasing Redivis’ data querying tools.

The goal for this project was to transform the large, complex dataset into a small, clean table that would allow me to answer my research question. In a Redivis project, I manipulated the dataset to only include records for the number of opioid prescriptions by county and year. I brought in a labor statistics dataset and compared the number of opioid prescriptions with the county-level, annualized unemployment rate.

Below is a section of the final output table:

Using this table, I created several visualizations of county-level opioid and unemployment data on Jupyter Notebooks, Google Data Studio, and Datawrapper.

This series of maps depict the opioid prescription and unemployment rate by county in 2008 and 2010.

This Data Studio report contrasts the state-level unemployment rate with the opioid prescribing rate from 2008 to 2010.

Step 3: Writing the script

Once the data story was complete, it was time to start working on how to communicate the process. We wanted to begin by crafting a polished video script that could be used as a voiceover. To my surprise, this didn’t involve writing a script from scratch. Rather, I recorded Redivis CEO Ian Mathews walking through the demo step-by-step and then transcribed that recording into a first draft.

Step 4: Filming and editing

Once the initial script and audio recording were complete, I began filming scenes for the video by recording my laptop screen and walking through the project on Redivis.

I then manipulated the video and audio clips in an iMovie project to create the first draft of the video, which was nearly 12 minutes long.

Step 5: Feedback and iteration

I presented the first draft of the video to others on the Redivis team to solicit input. Ian, my mentor and Head of Design Erin DeLaney, and I worked to revise the script and process, taking note of the overall flow as well as little things like the tone of voice, tense, etc. I would then remake the video in its entirety, re-recording all of the audio and re-filming each clip.

I went through this iteration process several times and produced four distinct videos, each one more polished and concise than its predecessor.

Each round of feedback I gathered was incredibly useful. If time permitted, I would have replicated this process a few more times.

Impact

After two weeks of video editing and remaking, I published the final demo video. It is currently featured on the Redivis homepage to help users and prospective organizations better understand Redivis and get started working with the tools.

I am extremely grateful for the opportunity to have created this demo video. In constructing this video I not only practiced data manipulation and how to communicate complex and compelling stories, but also how to research, plan, and execute a strategic and impactful project. I look forward to seeing how Redivis continues to grow and change in the coming years.