How can data analysis become useful?
Data - specifically 'big' data - has been trendy for a while now. Data dashboards, data scientists, data sets - but no sign of any common sense use cases. So where to from here?
“Big data is bullshit,” as Harper Reed put it back in 2013. Reed, the former chief technology officer in the Obama administration, was putting voice to an enduring frustration with ‘data’.
Big Data is the fad that set sail to a thousand marketing campaigns: Data this, data that - we were promised data utopia but all we got was this stupid graph. “The ‘big’ there is purely marketing,” Reed said. “This is all fear … This is about you buying big expensive servers and whatnot.”
The term ‘big data’ isn’t a misnomer. There is a lot of data out there now. It is indeed big. Even at the smaller end of the business scale. But big isn’t the key part of effective data analysis. As the writer James Gleick quipped: “When information is cheap, attention becomes expensive.”
“There’s an assumption that more data - or big data - is better,” said Nina Monckton, the chief insight officer of NHS’s Business Services Administration (BSA). “We’re using big data because we need it to get the answers we want, but it depends on what you’re trying to do? What’s your hypothesis, what are you trying to achieve?”
“People assume they’re going to get some data and chuck it into a box and miraculously press a button and it’ll tell them something.”
Monckton isn’t a finance professional per se, but she recently spearheaded an effort within the BSA to eliminate waste through data analysis. The efforts culminated in £100m of savings within three months.
But far from ascribing the savings to just amorphous ‘data’, Monckton cautions echo Harper Reed’s misgivings about the data hype. “I do feel as if some of the robustness that’s gone into producing statistics has gone by the wayside in all the excitement about how many terabytes of data you can analyse in one go.”
Proper data analysis begins with hypotheses. It’s not data for data’s sake. “The first step was to come up with some hypotheses about where there might be waste in the system,” according to Monckton.
“Which we did through talking to people in the organisation, people outside the organisation and thinking through where are the areas where there could be either gaming or waste.”
Through this process, Monckton’s team created a list of twenty different hypotheses that had been sponsored by a specific person within the NHS or someone external. “They would basically say ‘Could you investigate this for me?’,” Monckton explained.
“And then we picked off the ones based on the amount of money spent on them. The ones where there would be the biggest opportunity for savings. Where it would be relatively easy, just looking at data and comparing to other peers. Where anomalies would appear that couldn’t be explained by clinical necessities.”
Monckton’s team looked for anywhere where there was a declaration of numbers. Anything that wasn’t a fixed price, where there was freedom for individuals to make claims, for example. “We thought about human nature and how people might interpret the rules in their favour.”
The real work comes after the analysis when recommendations are made. At BSA, either Monckton or another analyst will sit down with the person sponsoring that hypothesis and make their recommendations.
“We track the recommendations we’ve made,” said Monckton. “Once the recommendation is made, we check on it and go back and ask whether its been implemented. If people are just asking questions for the hell of it, and they’ve got no intention of doing something, it’s a waste of resource.”