Data now has a very different value post-COVID-19.
The value of data is not something new. Before the pandemic, big data analytics and the virtues of a data-driven culture were already hot button topics for the C-suite. The increase in chief data officers (CDOs) is a testament to the rise in awareness.
However, when COVID-19 turned into a pandemic, companies found that they needed different data.
Joe DosSantos, chief data officer at Qlik, knows this well. He is, after all, at the frontlines.
"So, the [top] questions were: Where the heck are my employees? Where do they live? What offices do they go to? How are they affected by this thing?" he suggested.
This data was available. But it was hidden in HR and other department data silos. "Very quickly, people wanted to know what the office closure policy is, the work-from-home support, the technologies that they need, etc. — so, a lot of concern about employee data."
C-suite started asking for more business transparency. They also saw the need to do what-if analyses more frequently.
For example, companies wanted to know whether they should lay off their employees, furlough them, or reduce their salary. Or which product lines to focus on, and which to stop, said DosSantos.
This changed the data conversation right at the top. In the past, predictive analytics and AI were "delegated off quickly because it sounds very techie and boring."
Now, decision-makers wanted to become more data-driven. "And this is the moment we CDOs have been waiting for," said DosSantos.
The time is here
One consequence of COVID-19 was that the C-suite began to appreciate data operations (data ops). The pandemic also caught out those who did not have one. "Because the time that you need the data is not the time to ask where it is," said DosSantos.
He observed that companies wanted to know how they could take available data, build pipelines, "and make the data work for them." They also wanted to speed up the analytics process to guide their operational decisions that now directly impacted their business survival."
"For example, a lot of retailers needed to figure out how to get the right [products] into the retail channels. And so they are really trying to get the data for action," said DosSantos
Another COVID-19 upshot was that the questions management asked changed. "If you ask how much money we made, that's descriptive analytics. How much money will we make is predictive analytics. But it's still using the same data."
As companies started to ask more what-if analysis questions, data workers began worrying about the richness of data. It meant adding more valuable data, including third-party data, for analysts to understand and offer more insights.
All these concerns made data ops necessary said DosSantos.
Data mind games
Data ops success, at the end of the day, depends on the right data governance framework. And this raises some hard questions on data ownership.
Most of them concern data ownership. When one department or division does not see the advantage of sharing their data, it creates both a data gap and analyst rift.
DosSantos noted that from a line of business perspective, it is easy to see why such turf wars occur. For example, in a bank, if you are measured by assets under management, why would you share the data of your division's best customers with the rest of the company? "Even though it makes sense from a corporate perspective."
It boils down to alignment. "I think you need to start out with highlighting what the alignment is," he said, including offering clear incentives for sharing data. An internal center of excellence or small data-driven projects can help, but ultimately it needs a strong corporate political will to shape old habits.
DosSantos felt this could be difficult but can be done successfully. He also suggested answering four questions to get data ownership right:
There is no single answer or blueprint for answering these questions. Every company will have its challenges.
For example, DosSantos argued B2C companies are doing better in entering data because it is usually captured by POS machines, lies within behavioral pattern reports and entered using standard forms. B2B companies are a different story. "You have data that is often entered by sales reps. And those are not the best people to enter them."
No matter how difficult, companies need to address all four questions to create a proper data governance framework. Such a framework will then allow everyone in an organization to have a single source of truth.
It also frees up time for data scientists and analysts to do the work they were employed to do, and not waste time cleaning up bad data or wrestling with erroneous entries.
Let's play along
While leadership plays a substantial role in successful data governance, data workers need to do their fair share. They need to become adept at persuading and working with their peers in other divisions and departments to create shared value for sharing data.
"I often have said that my job is 10% technology and 90% psychology. You have just got to keep reassuring people that you're there to support them," said DosSantos.
He noted that getting this alignment is also becoming critical as companies try to find their feet in the post-pandemic landscape. "One thing that the pandemic has actually shown us is the importance of the time value of data.”
Here, the value of a data point diminishes faster as the market dynamics change quickly, as it is doing right now.
This is already pushing a lot more companies to look into cloud-based data lakes. "They can make data available in a more real-time kind of construct. And if you combine that with the semantics and the ability to ask any kind of question that you want it's really where you're going to start to get some noticeable difference."
But without strong data governance, these decisions will be fraught with erroneous assumptions, bias, and security issues. And in today’s market, there is no room for such decisions.
Photo credit: iStockphoto/Tero Vesalainen