Big Data & AI World Recap: Why improving data literacy is about culture and mindset
Fri 6 Nov 2020 | James Orme
October’s Finance & Banking Virtual Summit, hosted by Big Data & AI World, asked a panel of data leaders how they boost data literacy in their organisations (video replay)
How can your company improve its data literacy? With the UK and US ranking 21st and 23rd respectively in global data literacy skills, it’s a question on every organisation’s lips and a challenge data experts tackled head-on in an afternoon panel session at Big Data & AI World’s debut virtual event, dedicated to the financial services sector.
When pressed by KMPG’s director of innovation Maria Shevchenko about the key components of organisational data literacy, data leaders from AXA XL, Open Data Institute, and Chubb emphasised how cultural factors like transparency, communication, and experimentation were just as important as improving technical skills.
“For us, data literacy means three things: data transparency, the ability to do self-service and communication,” said Ashish Haruray, who works in the Office of the CDO at insurer AXA XL. In other words, employees need to be able to independently understand what data means, judge its quality and assess its trustworthiness. “We’ve become accustomed to using data on face value. In reality, nobody can claim data is perfect, but at least we can help [employees] understand imperfections in data.”
Communication from data leaders about models used and their potential biases is crucial to enabling this independence, the panelists said. “We don’t want conversations [that start] ‘I believe’; we want “it’s what the data says’,” said Dante Tellez, Chief Data & Analytics Officer at Chubb, the largest commercial insurer in the US.
Self-sufficiency in turn provides a platform for experimentation, another cultural necessity which if not cultivated limits data literacy, Tellez said. “Once you understand that you’ve got to change your culture, you’ve got to be open to experimentation. 80 percent of experiments fail, you have to be willing to accept failure.”
Violeta Mezeklieva, Data Literacy Specialist at Open Data Institute, an organisation that provides educational tools and advice to help professionals develop data literacy skills, said companies should focus on bridging the gap between soft skills and technical skills and be conscious of their relative blindspots.
“You don’t have to become a data scientist. There are basic things you can learn that can make you more confident using [data],” she said, adding that companies often make the mistake of thinking that hiring more data scientists will remedy data illiteracy.
Haruray claimed that responsibility for enabling data literacy starts with the CDO, while Tellez added other executives like transformation officers, digital officers, innovation officers, and directors of culture must work together to spearhead necessary cultural change across a company and its different roles. Shevchenko said it was useful to view a CDO as more of a “Chief Data Influencer”.
In terms of how CDOs can play this role and effect change, generating interest and excitement around data is crucial, Haruray said. “In order to change data culture [the change] has to have real tangible benefits for those who use data. How [data] can help them really is the message which has to be communicated. It shouldn’t just be a mandate from the top, it should be seen as something that can benefit your folks on the ground”.
AXA, for instance, has a ‘Business Asset of the Week’ program, where employees take turns to inform the wider company about an insurance product they work on, providing colour and describing the data and processes behind it. “You assume these things are well known but actually they are not. It’s the little things that make the difference.”
Tellez said internal resistance to operational change can’t be eliminated but it can be controlled. If data leaders focus on a waterfall approach, instead of making small adjustments showing small benefits, they risk creating too much resistance.
“Don’t change everything at once. When people are very good at doing things in a specific way and you’re trying to change their whole world it’s hard for them. But once they see the benefits it’s natural for them to adopt and accept cultural change,” he said. While Mezeklieva added organisations should start by ‘perfecting the skills that already exist’.
Another benefit of transparency is that the processes needed to enable it also act as a bulwark against unethical data practices, such as the development of discriminatory models and encroachment on employee privacy. Mezeklieva highlighted the current trend of capturing employee health and location data for Covid-19 safety protocols as a particular concern.
“Data literacy in our perspective is not only having technical people in charge of designing systems, but having an entire organisation that is aware of how their actions are complementing the creation of that system,” she added.
While biases can’t be completely eliminated, Haruray said having mechanisms in place to show how information is being processed reduces the risk. “You have to account for what is ethical and what’s not, and that’s complicated, but the more transparent you are, the more likely you will avoid controversies.”
Like most transformational challenges, data literacy can be overcome with the right mindset. But organisations can waste no time in ditching their old ways and embracing change. “Literacy is at the core of transformation,” said Tellez. “But the traditional way isn’t going to work in the future. We have to move faster.”
Written by James Orme Fri 6 Nov 2020