Working on Data and Storytelling

When you read about data science or data analytics, you read a lot of exhortations that analysis alone is not enough: you have to be able to tell a story with the data if you want to get action or receive respect from your coworkers.

That’s easier said than done.

It’s one thing to tell a story around your data if that data is a one-time study, but if you have to report data week after week, because it is a part of a standard metrics report or dashboard, that becomes much harder. Sometimes there is nothing to say. It is just a weekly statistics, just like the monthly unemployment rate is.

Current Project: Marketing Data Visualization on Power BI

Right now I’m working with Power BI and I am just beginning to learn how to produce an end to end product on a bi-weekly basis, so telling beautiful engaging stories for each market will not be sustainable – at least not initially. In this type of scenario, standardization and design aesthetics will be the focus rather than storytelling. So far I have figured out how to schedule periodic data refreshes from Quickbase to Power BI Service (rather than to Power BI Desktop on my desktop – takes me out of the process), let the data flow through the tables and charts in the report and hopefully be ready for automatic distribution to Sharepoint. Unfortunately, Sharepoint is not ready yet so I don’t know if automatic distribution is possible.

Instead, I’m going to have to figure out how to automatically produce the reports in PDFs and save them into Dropbox rather than push the reports to Sharepoint. First signs suggest that is not possible. It looks like the capability to download the report into PDF was just rolled out in February 2019. And I’ve read VBA cannot manipulate Power BI Service because Power BI environment is not set up as object modules. Maybe Python would be the way to go but I would have to first learn it.

There is something called Paginated Report Builder so I will have to investigate that and see whether paginated report builder can lead to automated report distribution into Dropbox.

Having played with Power BI, I find it to be a data visualization tool, and not a simulation tool. But the data visualization tools available is not unique to Power BI; Excel has a lot of these visual tools too. What Power BI can do much more easily is deal with large datasets and can pull in data from multiple sources, sometimes automatically. It also has a feature where once you set up the instructions on how to pull in the data and clean it up, Power BI will do the same data clean up every time you refresh the data. The other aspect of Power Bi is probably the distribution side although I haven’t worked on that yet because I was waiting for Sharepoint. But the tantalizing part I saw was sending or embedding the report to a website. I’m wondering if I can make the Sharepoint site similar to a website with all of the pretty visuals and deploy the reports there. I’m hoping to make the information more visually engaging and eventually interactive.

My Attempts at Storytelling Rarely Succeeded

There’s a lot to learn so storytelling is out for now but that is always in the back of my mind. I’ve always tried to show data in a different way or to change the aesthetics of the report to make it friendlier or interesting, but most of the time people are not ready for change: they want the same thing as has always been done. The most success I had was at SAIC when I rearranged a spreadsheet showing umpteen columns of data into a report format that told a story about contract details, financial health, and forecasted project profitability. I was initially apprehensive to roll this out in a review because it was so different from anyone else, but it actually generated a buzz. I once showed the makeup of the contracts we had by showing a histogram of the sizes of the contracts. In the early years, the vast majority – maybe 80% – of the contracts were in the $10,000 range and that histogram was probably the juice that instigated the project managers to go for larger value contracts. A couple of years later, that same histogram was showing more $100,000 contracts and a couple of million dollar ones, so we moved from small contracts to more million dollar ones.

That is the power of data storytelling.

But it is not often one gets to do that. Even Microsoft can’t do storytelling in Power BI.

A Beautiful Example

I quite often look for examples of interesting Power BI design or even just design in general – it could be Powerpoint design or annual report design or editorial design. In my mind, New York Times deploys some of the most interesting techniques in storytelling. The latest was about a young man falling into the alt-right echo sphere: The Making of a YouTube Radical. Not only is the story of how YouTube may be inadvertently creating white supremacists interesting, the way NYT uses data to tell its story is creative and instructive. NYT comes up with some the neatest way of using data to tell stories.

Other notables in using visuals or data for story telling are: Wired, McKinnsey, Fast Company, and HBR.