What happens when data goes wrong? BAT's CMO discusses
- 27 August, 2015 09:16
Analytics vendors are keen to describe how they can divine insights from the world’s vast troves of data. But in the eyes of BAT head marketing intelligent and deployment, Paul Ormonde-James, even the smallest slice of the wrong data can through all predictions out of kilter.
Speaking at the Chief Data Officer Forum held recently in Melbourne by Corinium, Ormonde-James described how the current focus on Big Data has led organisations – and particularly their IT groups – to horde data.
“They dump anything they can find into this space,” Ormonde-James said. “Because they are not users of the data, they don’t understand the problem – they are just interpreting an instruction. Then people like me have to find the needle in the haystack.”
The result is higher storage costs and a more lengthy sifting process, while the greater number of data sets used increases the chances that one will incorporate incorrect data can quickly infect the other sets it is matched against.
“The key to it is to embed a strict governance, of making sure before you bring in a data set that it is thoroughly analysed,” Ormonde-James said. “I would prefer accurate small data to big bad data. What percentage of bad data do you need to pollute the rest of the data? Not a lot.”
Ormonde-James said this approach was crucial in his work at the World Bank, which coincided with the Global Financial Crisis.
“Stuff was moving at a dynamic rate,” Ormonde-James added. “The key thing was to check if the data was right – was it coming in right, was it telling us the right things.”
As a result, Ormonde-James insists that the rules that determine what is loaded into a data warehouse be controlled by a chief data officer or chief intelligence officer.
"People that are looking at what the problem is and working out if the data can actually solve it,” he added.
Once the data is loaded and analysed correctly, insights should flow. But this can be just the start of the battle that chief data officers will face.
“Now what sometimes happens is these insights show you that the organisation needs to change, and do things differently,” he said. “And here comes the hurdle. Does the senior management support that? Because you could be saying things to them that are totally different to their paradigm. Now there is more data to challenge those urban myths.”
According to Ormonde-James, that immediately leads to the credibility of the chief data officer being challenged.
“If your senior management team are not willing to be open and to grasp these new concepts, what happens is you lose credibility, and get fired,” Ormonde-James explained. “So this new world is sometimes quite dangerous, because some companies don’t want to know. It’s called plausible deniability.”
Ormonde-James said his main determinant when taking a role is the culture of the organisation he is joining, and its willingness to change. He has counselled peers to leave roles where the organisational culture proves intractable.
“What I look for is someone really senior who gets it,” Ormonde-James said. “And that’s the white knight. And then I work to prove the value of getting this right and the headway we’re making, and give them the tools and processes and people to drive performance. And then they go out and they spread the word. You have to find others that are willing to take on the journey, support you, and together you drive change. They must be the voice.
“I don’t recruit the smartest people, I recruit people with passion. People that understand there is a journey. I look for people that can get things done and are true to their word, and can manage expectation, and make sure they get the credit for it.”
Follow CMO on Twitter: @CMOAustralia, take part in the CMO Australia conversation on LinkedIn: CMO Australia, join us on Facebook: https://www.facebook.com/CMOAustralia, or check us out on Google+: google.com/+CmoAu