How to Deal with Dirty Data
How to Deal with Dirty Data
If there's one thing marketers don't want their data to be, it's dirty.
Just the word “dirty” evokes an emotional response, says 20th Century Fox's VP of global enterprise operations Thomas Stilling, leading people to envision Pig-Pen's dust cloud following them around.
But with so much data being integrated and updated, marketers need to come to terms with the fact that their data probably isn't as clean as they want it to be.
“Data becomes dirty and evolves and isn't really that clean anyway,” Stilling said during his presentation at the recent Microsoft Media and Entertainment Day in New York, sponsored by the Media and Entertainment Services Alliance.
So, how do companies deal with this mess? According to Stilling, the first step is to enter a “no judgement” zone and face the dirty data head on.
“It's not going to bite you,” he said. “It won't crawl into your house at night and lie next to you as you sleep.”
Once they overcome this hurdle, Stilling said, companies need to focus on the following five factors to protect their data and its usability. And while he primarily referenced metadata during his presentation, these best practices are applicable to any data housekeeping.
1) Define scope
Before companies get too deep into the weeds, they need to identify what they're hoping to achieve: quality or quantity. This scope should be business-driven, Stilling noted. And while quantity is often the first thing people focus on, he said, quality is what the structure should be based on, including what the company can drive from this data.
2) Build a roadmap
Once these definitions have been established, Stilling said companies need to create a framework that will answer the following questions: Who will support this effort? What needs to be accomplished at each step? When does this need to be completed by? Where does the final set of data go?
Being clear and specific, he said, is important.
3) Conduct an honest assessment
While it's easy to sweep dirty data under the rug and act like it's not there, Stilling said it's important for companies to do an honest data assessment.
“You always have to tell the truth ,” he said, “and you have to tell the narrative as the data suggests.”
Stilling suggested taking a methodical, documented approach to this assessment, which includes understanding the background, testing the company's “truth”, and mapping out different data inputs and how they're used.
4) Abide by the rules
It's important for companies to stick to their purpose-driven plan and not be distracted by unnecessary “bells and whistles,” Stilling said. Just because a company can do something, he noted, doesn't mean that it should.
“Structure will set you free,” he said.
At the same time, Stilling said companies need to keep their plans flexible enough so that they can be adapted for the future.
5) Measure the right metrics
If companies really want to determine whether they're leveraging data effectively, then they need to measure the right metrics.
“Metrics should never surprise,” Stilling said.
That's why he encouraged conference attendees to track practical KPIs that are meaningful and impactful.