There’s no shortage of value in analytics, especially as it relates to the kind that data science, artificial intelligence, and machine learning can provide.
They are technologies that can not only make sense out of mind-boggling volumes of data, they can take on the responsibility of mundane tasks, thereby freeing up humans to focus on problems that computers just can’t solve for.
But even in a world that is more reliant on speed and efficiency than ever before, automation has its limitations. And in Episode 5 of Domo’s CURIOSITY video series (see snippet below), three data experts—Donald Farmer, Mike Bugembe, and Ryan Schrupp—discussed why.
For one, a machine can’t place values on the relationships you have with your customers or clients, Ryan said.
“If we went purely by numbers right now,” the senior vice president of enterprise research at U.S. Bank said, “we would downgrade clients, and we would take credit away from them. But we know those types of decisions are going to dramatically impact our relationship with these clients.”
Another problem we must consider is that we live in a very different world from the one of even a year ago, which makes it almost impossible to rely on historical data to the extent we used to, Donald said.
“When you look at historical data sets, they shift over time,” the principal of TreeHive Strategy said, “so concerns we had a few years ago may not even be captured in that data set in a way we can identify.”
To help Donald drive this point home, Mike called out the recent exams debacle in the United Kingdom. In short, because students couldn’t sit for year-end tests this year due to the pandemic, an algorithm was developed and deployed to predict how they may have fared.
“It was disastrous,” said Mike, who lives in London and authored the best-selling book Cracking the Data Code. “So, whilst we have increased our sophistication in the algorithms that can help us with decisions, I still think at this stage humans are the better big data machine.”
For organizations to arrive at a place where they are getting the most out of what data and automation promises, they have to be able to fuse the best of what man and machine bring to the table, Mike said.
“That’s where we’ll really get a massive benefit,” he added, “and a step-change in our ability to make the right decisions.”
Ryan agreed.
“At U.S. Bank, we’re trying hard to inject more data into everything we do because we can optimize some of the decisions that are being made and we can do so in a way that doesn’t produce errors,” he said. “But at the same time, we’ll always need our relationship managers to take what the data is telling them and do the right thing with it.”
How do you go about finding the right balance? By first getting some people together to talk about how you’re going to treat the results that come from your algorithms, Ryan said.
“You have to have deep conversations on it,” he added, “and then strategize about how you’re going to handle it.”
In Mike’s opinion, it all starts with identifying what your key decisions are.
“So many organizations try to apply data and AI and really complex analytics, but for what?” he said. “They have decisions that could really do with the support of analytics, but they’re spending all their time on a fancy use case that really adds very little value.”
To learn more about the role automation can and should play in helping users and systems take better actions, watch or listen to Analytics Is Not Enough.