It’s the business buzz phrase of the moment, but is ‘data democratisation’ such a good idea when so many of us find it hard to understand statistics?
It was Clive Humby, the mathematician and data scientist behind Tesco’s Clubcard, who coined the phrase “data is the new oil”. It has since become something of a corporate cliché, uttered by every consultant and CEO who wants to be seen to understand the digital economy.
For Ved Sen, head of business innovation in the UK and Ireland for Tata Consultancy Services, this maxim doesn’t quite ring true. He would instead describe data as “the new plastic”, because “we create a lot, we struggle to know what to do with it and it tends to turn up in the wrong places. And, for all the talk about the democratisation of data, business is not yet culturally geared up to handle this. There’s a lot to do.”
Certainly, there are benefits to be had from data democratisation: the relatively new school of thought that data should be made readily available throughout an organisation instead of being kept in silos. The idea is that this provides a basis for more informed decisions throughout the organisation, as well as promoting innovation. Some even claim that empowering employees by “digitising” a company, as Western Union calls it, will be a game-changer.
Yet only 27% of executives surveyed by MicroStrategy last year said they felt that they’d built the right organisational culture to support a data democracy. Indeed, the many hurdles to be surmounted in deposing a data dictatorship – aside from obvious ones, such as concerns about security and, for SMEs, simply finding the time – are only starting to become clear.
Trust and transparency
Matthew O’Kane is global head of AI solutions at tech services company Cognizant. He believes that companies need to handle the democratisation process with great care. In part, this is because of the potential interdepartmental sensitivities about data ownership, especially in larger organisations. A recent multi-industry study by McKinsey found that, for teams requesting access to internal data beyond their departmental remit, the response time could be measured in months in 53% of cases.
O’Kane cites one retail bank’s board-level diktat that ordered the centralisation of all data overnight but also gave reassurances that no material would be used without the consent of the team that originated it.
Once a fully accessible centralised data store has been established, promoting it as such can be helpful, he says. For instance, AT&T rebranded its internal online marketplace, Amp, as a company-wide data hub last year. This sort of “personification can create trust in the data”, according to O’Kane.
And trust is vital. When employees are offered greater access to data, they don’t necessarily believe what they’re presented with. Respondents to a 2018 survey by Experian said that they considered 30% of the data held by their companies to be inaccurate on average.
This is not the only issue of trust that affects data democratisation, according to Kevin Hanegan, a founding partner of the Data Literacy Project and chair of its advisory board.
“The number-one thing I hear from the many CEOs and CIOs I talk to is that they don’t feel ready for the democratisation of data, because they don’t trust employees to make the right decisions using it,” he says. “There’s a lot of talk of software being the solution, but technology is the least part of this. Anyone given access to data needs to be able to interpret it. Until then. it’s like giving someone a black box that says ‘the answer is A’, to which their immediate reaction might be ‘why is it A? How would I know?’. There isn’t a tool to bring about this change. It won’t happen overnight.”
Reading the numbers
The lack of trust is part of a bigger weakness with the democratisation concept: the fact that most people struggle to deal with statistics. A recent US study found that 46% of high-school graduates were unable to estimate how many times a flipped coin would probably come up heads in 1,000 tosses. The most common wrong answers were 25, 50 and 250 times.
“The problem is that a lot of statistics are counterintuitive and full of surprises, which became clear during the Covid crisis. That’s especially the case with brains like ours, which are just not built to be good calculators of, say, probability.”
So says Stian Westlake, chief executive of the Royal Statistical Society, which has found a lucrative sideline as a training provider to firms in industries such as pharma and petrochemicals. He adds: “Our brains are very good at seeing patterns, but struggle to see randomness when it’s there. And it’s often the case that [when it comes to data analysis] people don’t know that they don’t know.”
This applies even in professional circles where a good understanding of data might be considered crucial. A study of 492 physicians for the Journal of the American Medical Association this year, for instance, found that their assessments of pre-test data led them to overestimate the likelihood of breast cancer in a patient by 976%.
Small wonder, then, that a 2020 study by data analytics company Qlik found that only 17% of lay employees considered themselves confident in handling data. More than two-thirds (67%) of respondents admitted that they felt overwhelmed by the numbers, while 19% said they had gone so far as to find other ways of completing a task without using data.
The value of data
The danger here is that there will be an expectation that lay employees possess the same understanding of data as its traditional keepers, the analysts. It’s why O’Kane sees a growing role for artificial intelligence systems in filtering data into more comprehensible packages and also in checking all resultant decisions.
“After all”, he notes, “a bank manager doesn’t decide whether or not to approve your loan anymore. A computer does.”
This is also why training is important, according to Sen, whose company gives its employees mandatory training in certain aspects of data analysis.
“Given the volume of data that’s out there, it’s fundamental that any business seeking to democratise data should also educate its people in using that data,” he says. “Without learning how to handle data objectively, we all have biases and will lean into our own experiences when presented with material that challenges us. The training is not sufficient at most companies, but the smart ones are really investing in it.”
But there is an argument that, even before a business addresses data literacy, it must address its data dependency: the belief that all data is valuable, even if this doesn’t help solve any known problem.
“It’s important for any business to recognise the dynamics of complexity in data. And the fact that zooming in on really critical information can sometimes be highly predictive, better so than when using all the other data that might be added,” says Florian Artinger, professor of digital business at Berlin International University of Applied Sciences.
He cites the wildly fluctuating prices of airline tickets over the pandemic – a product of the data on which the industry’s pricing model is based.
“Yes, making data more available can empower an employee’s expertise, but we shouldn’t be lured into thinking that an idea which can be backed by data is necessarily better than one that comes from intuition or experience. After all, data can also be used simply to ‘cover your ass’ and justify inaction or a bad decision,” Artinger says. “What we need first is a business culture that knows not only how but when to use this democratised data – and when not to.”