Data analytics can help business leaders and finance professionals make better decisions about sales, financial management, marketing, investments and strategy. But there’s a catch. Success depends on identifying the right data and ensuring it is good quality, experts say.

Yalçın Akçay, director of the Melbourne Business School’s Centre for Business Analytics, says artificial intelligence is making a valuable contribution to business insights, but only when business leaders clearly identify what the business problem is before they start, which can then reveal the sorts of data needed.

“That initial stage, where you properly define and scope the problem, is very critical and will dictate essentially whether you’ll have a successful decision-making process at the end,” he says.

Analytics and data-driven decisions go wrong when we look at data to identify a problem and then look to the same data to find a solution.

“[Some people] take the easy way out and create business problems from the existing data,” says Akçay. “Whereas you should first use your expertise to fully define what business challenge you’re trying to solve and then ask the question: ‘How can data help me make better decisions?’. Then, you go ahead and essentially look for the data that’s critical to addressing that challenge.”

“[Some people] … create business problems from the existing data, whereas you should first use your expertise to fully define what business challenge you’re trying to solve and then ask the question: ‘how can data help me make better decisions?’.”

Combining data

Craig Jones, deputy government statistician with responsibility for data system leadership at Stats NZ, says there is sometimes a tendency among data users to hoover up all the data they can find and “see what it says”.

“In my experience, data just doesn’t work that way,” he says. “You need to invest in its curation. You need to invest in data quality, so that you’re actually getting really robust estimates of what it is that you’re trying to understand. And that takes time.”

Stats NZ produced a briefing paper last year, called Aotearoa New Zealand: Empowered by data. In it, researchers estimated that data-driven innovation already contributes NZ$5 billion a year to the economy and could grow to seven times that amount by 2030, potentially contributing up to 9% of gross domestic product.

Jones says there is value to be gained by combining data from different sources and more data collaboratives between organisations are starting up. These collaboratives facilitate the sharing of data between organisations by setting up governance mechanisms that take account of ethics, privacy and security.

“When you start integrating data with other sources, you start to really derive value from it,” he says.

“When you start integrating data with other sources, you start to really derive value from it.”

Data collaborative case study

For example, New Zealand company The Toha Network has initiated a sustainability-focused online information commons called Calm The Farm. Technologists, scientists, farmers and local regional regulators work together to help farmers transition to regenerative agriculture.

The information commons combines data from many different sources, including individuals, big datasets and remote sensing. Farmers upload daily data about their farm practices, environmental actions and outcomes, while receiving financial rewards and retaining ownership and control of their data.

This data enables scientists to monitor the quality of the farm’s natural environment, such as the health of rivers and the sequestration of carbon into soil. Scientists also feed the insights back to farmers to help inform environmental-health decisions on farms.

Avoiding hand-picked data and bias

Data has great potential to inform decision-making, but if businesses don’t put the right structures, expertise and capabilities in place, it can lead to worse outcomes. “But if you were to ask me, in general terms, if you’re better off having some data even if it’s not perfect – rather than relying on guesswork and good intention – I would say rely on the data,” Jones says.

Akçay quotes the adage “garbage in, garbage out”. Essentially it means that the quality of the insights derived from data depends on the quality of the data. Data users have to ensure they’re not handpicking data to satisfy certain criteria and end up with a selection bias that will alter the insight.

“We know that the more data that you have, the better you can train advanced analytics models and machine learning algorithms. Better trained models will be able to make predictions that are more accurate,” says Akçay.

Data quality over quantity

Where there is incomplete data, Akçay says business leaders should put more weight on their experience and intuition for decision-making than they would when drawing on a full set of quality data.

“But [limited quantities of data] should not be what prevents you from going ahead and taking a dip into analytics and using data,” he says. “When you have a small but representative dataset, you can still build really effective models that can guide you and give you great insights about the data and the business problem.”