Democratisation of data can lead to organisational problems, says expert panel

Democratising data puts power in the hands of staff instead of managers, which can bring with it a whole host of problems, warn experts at a recent Computing event

Democratising data - making information available to all staff at an organisation - takes power away from managers and risks exposing a misalignment of objectives.

That's the opinion of Paul Fitzpatrick, consultant at Human Longevity, speaking at a recent Computing IT Leaders Forum.

"The alignment of objectves is key to an organisation," said Fitzpatrick. "And where objectives aren't aligned, data can make that situation more transparent. Culturally it's a big leap if everyone has access to the same dashboards. With data comes control, so [democratising data is] putting control in hands of people not executives. If you do that, where there's misalignment of objectives, it will become more obvious," he argued.

Jason Nathan, group MD for data at data analysis firm Dunnhumby explained that people will interpret data in ways favourable to themselves and not necessarily the company, if their objectives aren't properly set.

He used the example of supermarkets, and the complexities inherent even in what seems to outsiders to be trivial data.

"It all boils down to the definition around the data," said Nathan. "It may sound like an extrordinary thing to say, but even something like knowing how much did this product sell by value over this week is really hard. How much time do you allow for returns? What about promotions? How much value do you ascribe to these products from multi-buy packs? People try to game the system in their favour if you allow free reign at a granular level.

"In any job, you see people who act at times in the company's best interests, and at times in their own. At a well managed company people always act in their best interests, which happens to align with the company's. But that's a nirvana which is usually unattainable."

Nathan continued, explaining that when fully democratising data at its most granular level, the interpretation placed upon that data causes a lot of friction.

"As soon as you place a layer of interpretation on top, you're not democratising that data, you're allowing someone else to game it, and you're not allowing unfettered access," he said.

"Around 1832 we had universal suffrage. We need the same thing for data," said Bob Tulloch, technical director at Walnut Medical.

Gopal Sharma, global practice head - strategy and architecture at Liaison explained that organisations need to create an enterprise data layer into which everyone has visibility.

"This will then deskill the roles themselves by establishing business rules and logic around data validation, that's the opportunity. Make that data available for anyone to use, analyse and create algorithms for. That requires a very visible and engaged leadership, and each organisation needs to look inwardly at the blockers they have."

Earlier the panel had argued that Google can sometimes lose interest in some aspects of its cloud offering and not keep them updated as much as they should.