data discovery mountain of data concept

When’s the last time you were talking data and analytics and someone stopped the conversation to ask what a term meant? If you’re wracking your brain now, you’re not alone. It’s a rare situation to run into.

Analytics professionals like being on top of things, which means sometimes when we hear something new in a conversation we just nod and make a mental note to look it up later. And then work gets busy and before we know it we’re resolving to just do our best to pick up said buzzword through context clues in other conversations.

This all-too-familiar scenario is why I’ve decided to just answer the question that everyone is thinking, “so what is data discovery anyway?” After all, if one of the versions of data discovery out there keeps being touted as “the end of traditional BI,” we should probably get some clarity on it, right? Of course that statement is a bit hyperbolic, but it still can’t hurt to have a solid definition of data discovery and five key facts focused on how it can work with, not against, our traditional BI environments.

Sound like a plan? Okay, here we go.

Data Discovery Defined (Sort of)

One reality we need to embrace when we set out to define any concept is that there’s going to always be some debate. This is especially true with data discovery since there are two different and prevalent definitions out there. One has to do with information management processes and the other has to do with front-end analytics, which is what we’ll discuss here. So don’t be scared of the “sort of.” All that means is I’m going to try to give you the simplest definition possible for the angle I’ve chosen to take first.

At its heart, data discovery is a way to let people get the facts they need to do their jobs confidently in a format that’s intuitive and available.

That’s it. There are more details around how data discovery works in different contexts, but that’s the core message. It makes it so when someone has a question relevant to their function at an organization they can get the answers without a big turnaround time or reliance on a technical resource.

In terms of the details, there are five key facts you need to understand to fully grasp data discovery:

  1. It’s fast.
  2. It’s usable.
  3. It’s targeted.
  4. It’s flexible.
  5. It’s collaborative.

Before we break these down, I’d like to introduce you to Ironside’s number one tip for BI data discovery: be careful. This process can get messy fast, and you need a solid core of governance principles in place to make it work right. It’s always best to put a balanced system in place that can give agile results but also maintains high data quality standards.

1. It’s Fast

data discovery speed concept

Data discovery is designed to answer immediate, spur-of-the-moment questions. As Ann All of Enterprise Apps Today puts it, it’s optimized for “easy one-off data analysis .” An ideal data discovery solution allows information from many sources to be accessed whenever needed. Features supporting this in many data discovery products include quick connection to many data source formats, search functions for narrowing down data that’s brought in, and recommendations of optimal display structures for different types of analyses. It all results in a spontaneous question turning into a dashboard with answers in the shortest time possible.

2. It’s Usable

data discovery riding bike usability concept

Presentation is also a core focus in data discovery. It stays as far away from actual query syntax as it can get. Instead, it relies on intuitive, drag-and-drop interfaces that make analysis steps clear and provide many prestructured templates, visualizations, and workflows. Users may stay in the dark about what’s going on to get them the data they see, but that’s not the point. Remember, as Antivia says in their blog, “none of these groups need or want to see the original workings of the data scientist .” For them, it’s about having informal access that caters to the needs of the moment and answers more questions than it raises.

3. It’s Targeted

tree with target in the forest

Data discovery isn’t meant to be a monolithic, enterprise-wide practice that’s everything to everyone. It’s optimized to meet a specific need. It gathers quick, trusted initial results that satisfy the different business departments when they need a quick fix. If the same question recurs over and over again, chances are it should be brought beyond data discovery and into a more formal business analytics structure through work with data scientists or BI analysts. But that’s what’s great about it. Patterns like that can be uncovered and added to the larger BI reporting ecosystem and still be valid at the moment they first come to light.

4. It’s Flexible

data discovery flexibility concept tying rope

Even though data discovery has a specific focus, it’s not narrow. It can apply to any department or function it can access data for. In fact, to support this, some companies have started using data lakes as a central repository for all their data assets so users from anywhere in the business can get at the whole picture. This flexibility supports the speed and usability aspects of the approach, since it makes the analytics process the same across the board.

5. It’s Collaborative

data discovery collaboration concept jugglers

As I hinted at in point 3, data discovery shouldn’t be separate from other analytics processes and techniques in place in our environments. On the contrary, it should strengthen them. The idea that “traditional BI” and data discovery are diametrically opposed is false. As Southard Jones states in Wired, “data discovery remains one small piece of the larger pie that is business intelligence .” Rather, we should see data discovery activities as a gateway to continuously improving more formal reporting, data science, and information management activities.

For subject matter experts in these areas, data discovery can be a valuable barometer that shows them their organization’s interests and lets everyone pitch in to help move goals forward whether they’re technical or not. It can also help with overall data quality and standards, since administrators and architects will include this kind of analysis in their planning when developing an infrastructure and it requires a dedication to governance to work right.

What Is Data Discovery to You?

Now that you’re armed and ready for the next time you hear “data discovery” in a business analytics conversation, take a minute to think about how it might apply in your situation.

  • Are there lots of questions that you or others have that go unanswered or take too long to resolve?
  • How would you meet the architecture requirements necessary to put data discovery in place?
  • How would you formalize data discovery findings that should be part of daily analytics processes?

Ironside can answer these questions and many more. Discover how we can help you excel in analytics by grabbing a copy of our Lookbook. We look forward to becoming a trusted partner in your journey to data discovery.





Download Ironside's Lookbook



References

  1. All, Ann. Data Discovery Is Changing Business Intelligence. Enterprise Apps Today. September 18, 2014.
  2. Antivia. Busting the buzzwords: what is Data Discovery? May 29, 2015.
  3. Jones, Southard. The Battle of Business Intelligence: Data Discovery vs. Traditional BI. Wired. August 2014.