Data vs. digital: That’s a big tension within many organizations.
Chief Data Officer s and Chief Digital Officers don’t always agree about some important things, said Joe Caserta, president of consulting firm Caserta, during his DATAVERSITY® Enterprise Data World Conference presentation titled Building a Foundation for Disruption and Advanced Analytics. What’s the disconnect between the two roles that share the CDO acronym?
The Chief Digital Officer is really about the customer experience, about being the customer advocate. The Chief Digital Officer wants the customers to buy something as quickly and as easily as possible, no matter what device, and to capture and share the data about the transaction with any other part of the application and with any other part of the business that’s going to make that experience better, Caserta said. And to do it all as quickly as possible.
The only way a company can know anything about its customer is through data — every single transaction and touch point. A business won’t stay in business very long if a web application isn’t customized specifically for a particular customer, if it doesn’t know what that person bought last and what their preferences are — and not because the individual told the company what those are, but because the seller observed it and wrote some algorithms to learn it.
The data world is moving fast. But the Chief Data Officer has legitimate concerns that may slow down innovative experiments. Their role includes the very important duty of making sure that that data is governed and is consistent, quality-assured, and secure.
“So, we have this conundrum where we have to have really fast speed-to-value, instantaneous change management, and then we need to be able to be safe and secure,” Caserta said.
The Chief Data Officer has to ask questions like: Where do we need to enrich our internal data with external data? How do we monitor usage of our data? How do we come up with security methods and standards? That person has to accommodate schema drift with the plethora of disparate unstructured and semi-structured data filling the coffers and provide accountability for the data. The business can’t tolerate continual changes to metadata that affect ETL processes and jeopardize appropriate data flow.
“The type of data that’s available to help you run your business better, it’s broken into about 24 different categories,” Caserta said. “Things like weather data, satellite data, images of parking lots to try to count cars to figure out how your business is doing, demographic data, genomic data, and so on.”
That’s a lot of data to manage.
Meanwhile, the Chief Digital Officer is by virtue a disrupter in an environment where everything is disruptive. You might think of what this person does as cannibalizing the business — but for good reason.
Caserta gave as an example The New York Times. They knew when they started going into the digital business that newspaper sales would slow because now its customers have an app. But if they didn’t take that step, they risked being wiped out by all the other digital news media. “They would be out of business. It’s very, very obvious to see it for them,” he said.
How to bring the two perspectives into alignment? Caserta says it’s starting to be the case that sometimes the data officer is reporting to the digital officer, or vice-versa. “When that happens, it seems to be more cohesive and the business tends to run a little better,” he said. Setting standards, making the data platform stable, and the processes reusable come together with innovation — building things really fast and as cheaply as possible to get a quick handle on what does and doesn’t work, and generating revenue with data.
Swimming in the Data Lakes
What does this alignment look like? There’s a managed environment where people can get to their data quickly and easily with a very tunable Data Governance strategy, he said. Data lakes are the only option, the places that help a business be at the core of a data ecosystem, which itself is everything about data — ingestion, governance, pipelines, warehousing, analytics, machine learning, and data science.
“We need to have some kind of an environment where we can just take that data, put it to rest and explore that data,” he said — unstructured, semi-structured and even structured. It’s the safe space for data quarantine until somebody tests it to see if it’s okay to come into the general data population.
Data is becoming 80 percent exploratory and 20 percent production, he noted.
“We need to have a sandbox where we can experiment, we can discover data, explore data, analyze data and test our theories on data,” he said.
No amount of pre-development meetings will deliver as much knowledge as just having users have the ability to look at data. Once data is in a query-able format, he said, “we’re probably doing as much analytics off the lake as we are off the data warehouse.”
And once data is moved into the general data population, “it goes through a very rigorous Data Governance lifecycle.” That requires knowing the organization around the data and the people who deal with it; how the data is identified, quality-assured and secured; how the data is made accessible; how data entities are mastered; and how they are integrated into data processes.
“And then how are you actually systematically doing data retention and archival based on regulatory requirements or compliance,” he said. “Every component of the data ecosystem needs to have its own policy for these things. When we build the data lake, or build a data ecosystem, every one of these things is affected.”
The whole environment is built around a data platform, with the Chief Digital Officer supporting an engaged business and a chief data organization composed of Data Governance coordinators, data libraries and data stewards. And more — IT gives the enterprise architecture and the data engineering, and data scientists do advanced analytics.
“You need to bring them all together as a team, virtually or physically, and do this as a unit,” he said. “And if you don’t do this as a unit within the business, and tech, and data science, and Data Governance all together, it’s going to be really, really hard to succeed.”
Check out Enterprise Data World at www.enterprisedataworld.com
Here is the video of the Enterprise Data World Presentation:
Image used under license from Shutterstock.com