What’s a Data Product?
By, Kevin Le
Credit to “Designing Data Products” by Simon O’Regan for inspiring much of this article
Here at LionBase, we focus on connecting students and companies through data products. But, what is a data product? At first, the term “data product” seems quite broad. After all, data is integral in almost every piece of software — be it a website, a mobile app, or a video game.
To me, what really defines a data product is a product whose primary purpose and value are defined and driven by the use of data.
Think about it like this — YouTube in itself is not a data product. It is a platform for content creators around the world to share videos of all kinds. While data plays a pivotal role at YouTube, using data is not YouTube’s primary purpose. YouTube’s primary objective is to be a public space for anyone to have a voice. Simply being a common space for uploading videos is what drives the value of YouTube’s platform.
However, YouTube’s recommendation system is a data product. The goal of YouTube’s recommendation system is to use data to make a prediction on what content a viewer might like. Data and data science are what drive the innate value of YouTube’s recommendation system.
Defining and identifying data products is important because it lets us understand where the value of our work really lies so we can develop and iterate accordingly.
Types of Data Products
Ok, we’ve narrowed down the definition somewhat, but it’s still pretty broad. Well, that’s because data products still describe a huge set of products.
At LionBase, we focus on three main types of data products: algorithms, decision support, and automated decision-making. I have ordered these products in ascending complexity.
1.) Algorithms
When defining algorithms as a data product, it is best to view it as algorithms-as-a-service. An algorithm takes input data, processes it, analyzes it, and returns new information that serves an end user. For example, the mobile app Shazam “listens” to music, analyzes the audio, and returns a guess as to what the name of the song is. Here, the end user is an individual who wants to figure out what a song they haven’t heard before is called.
2.) Decision Support
For decision support products, our goal is to help users make data-informed decisions without making the decision ourselves. For example, an analytics dashboard such as Google Analytics can provide insight on how to improve a user’s website (e.x. understanding conversion rates), but Google Analytics will not make any changes to the website. There are two keys to any effective decision support product. First, determine what is actually useful to communicate. There are so many ways to analyze a single dataset, but only a few insights are truly meaningful. Second, communicate everything in a straight-forward manner, no matter how complex the behind-the-scenes work is. If a user cannot understand the information being provided, then the tool is useless.
3.) Automated Decision-Making
This is one step above decision support products. Here, we are making all decisions without the intervention of a human. YouTube’s recommendation system is a good example of this. A user is not recommending videos to themselves, YouTube is automatically performing such actions based on user history. The user only sees the final output.
Types of Interactions
Now, how can people actually interact and interface with data products?
We’re going to focus on three main interactions: APIs, dashboards and visualizations, and web elements.
1.) APIs (Application Programming Interfaces)
APIs allow [technical] users to access data from a platform. For example, the Twitter API lets developers grab Tweets from various users on Twitter.
2.) Dashboards and Visualizations
Dashboards and visualizations allow users to view important information in a condensed and graphical manner. It is important to design dashboards based on the users’ comfort with statistics and numbers. A major aspect of designing dashboards and visualization also involves determining what information to include. Any information displayed will influence a user’s decision.
3.) Web Elements
Web elements describe places on a webpage to hold or display some kind of data. For example, a search bar is a type of web element. However, web elements can extend to voice inputs and more. This is a way of seamlessly integrating data products into websites and applications.
Conclusion
In summation, a full data product functions in some intersection of the above data products and interactions. For example, a search function with a search bar would describe the intersection of an algorithm and a web element. Google Analytics would describe the intersection of decision support and dashboards. The list can go on!
I hope this blog post provides some insight on what data products are! If you want to learn more, I highly recommend reading this article on designing data products by Simon O’Regan which I credited at the top of this post.
~ Kev
Thank you for reading! Check out our website and like us on our Facebook page to learn more about LionBase. Interested in working with us? Send us an inquiry through our website here!
