Blog  Perspectives  

Data Lakes: Is It Time to Jump In?

by Amobee, May 09, 2017

Data has some complex concepts. Often we use metaphors to explain how vast quantities of information behave and are stored. We’re used to the term database, the suffix of which is borrowed from building construction. We’re also familiar with data warehouse and data marts which adapted concepts from retail.

Data lake, which dates back to 2011, is the latest buzzword. You may have come across the idea recently (perhaps in this Digiday article). The term describes a heap of raw data that doesn’t follow a common structure thrown together with no central organization. It also might be likened to a basement, where you keep household goods that you’re not sure what to do with – yet.

Save it, then figure out how to use it

In the hierarchy of utility, a database would have the most immediate use, the data warehouse second and the data lake third. (A data mart, meanwhile, is a subset of a data warehouse designed for access from a single business unit, like sales.) That’s because a database is designed to store information that’s actively being used. A data warehouse holds more data and isn’t as easily accessed. The difference between a data warehouse and a data lake is that while the former is structured, the latter isn’t.

A database may be more useful on a day-to-day basis, but a data lake could have a potentially game-changing effect, if a company can make meaningful connections with the information. For instance, a CPG company might keep ad server data and shipment data in its data lake. While the two sets of data are separate, you could use the ad data to see if there’s a correlation between ad exposure and demand for goods. Years ago, those two data sets would exist in different systems and analyst would have to pull information from both and then try to synthesize them. A data lake allows more universal access.

Fishing in a data lake

Data lakes can be expensive to maintain and harnessing them requires some expertise. Merely having one isn’t going to move the needle on a business.

Most of the time, businesses aren’t sure what the data they’re throwing into a data lake will be used for. Unlike data warehouses, data lakes can accommodate totally new kinds of information. If a brand-new type of social media springs up, for instance, a data warehouse may not have a structure compatible with storing information about it. With a data lake, you can store the data for now and then act on it later. A data lake then is kind of an investment premised on the idea that data-collection methods will keep improving, and that novel applications for the data will be found.

At the end of the day, a data lake has to help you accomplish a business goal. Those could be relatively simple or incredibly complex, depending on your organization.

For more on using data and analytics to create business impact, see our blog series on building an analytical framework.

Read Next

All Blog Posts
Perspectives

Why You Need a Data Strategy

While brands have now heard loud and clear that they need to start using data in their marketing, many organizations often struggle to get off the ground because they don’t have a real data strategy.

November 9, 2016

Perspectives

The Keys to Bridging the Programmatic Data Divide

It should be an unbeatable combination: provide creatives with the data they need, and they’ll come up with mind-blowingly effective ads targeted with pinpoint precision. Unfortunately for brands, Turn’s "Mind the Gap! 2016 UK Agency Survey Report" released today suggests it’s not that simple.

July 25, 2016

Perspectives

Three Keys to Success in Data-Driven Marketing

Globally, 3.2 billion people use the Internet. How can marketers cut through the noise in a fragmented media landscape? A smart data strategy matters.

August 17, 2016