Trust deficit retards predictive analytics adoption in banking, says Qlik

Concerns about trust and regulatory compliance are slowing down the adoption of predictive analytics in financial services.

This is according to research from data analytics firm Qlik, carried out by Censuswide in June and July 2021. Of the 503 financial services professionals surveyed, 200 were in banking, 201 in insurance, and the remainder from accounting, funds and investment, investment, and pensions, according to a Qlik spokesperson.

The report, Unleashing the potential of predictive analytics in financial services, found that 38% of British financial institutions have five or fewer predictive analytics use cases currently in operation, while a possibly leading edge 7% have introduced 50 or more.  

Predictive analytics involves the application of statistical analysis techniques to data to create predictive models to make forecasts. In this research, Qlik was working off a definition of predictive analytics that includes machine learning.

Qlik customer Richard Speigal, business information (BI) centre of excellence lead at Nationwide, said: “Integrating predictive analytics into BI empowers our organisation to harness its benefits for improved employee decision-making.”

Elaborating on predictive analytics use cases at the building society, he said: “I’ve seen Nationwide explore uses for predictive analytics in anti-money laundering, financial crime and risk management. There’s interest in using it to support human decisions, for example by triaging and helping to prioritise staff caseloads. I haven’t yet seen any desire to fully automate in this space though, we don’t have a culture for ‘computer says no’ – and long may it remain so.

“We are very clear about what our customers, our members, mean to us, and that extends to how transparent we are in decision-making processes. We would never want to make a customer feel like decisions were being made about them that couldn’t be explained. A human has to be able to explain those decisions.”

The research found that only half of respondents (50%) trust decisions made by predictive analytics systems to be without bias. Some 44% of them said they feared they could be held personally responsible for decisions automatically triggered by predictive analytics software – rising to 81% among those working in funds and investments. The regulatory burden also weighs heavy on them, with 46% reporting it outweighed the benefit the solution could offer. 

Sizeable numbers of those surveyed cited data management matters as barriers to using predictive analytics. Two-fifths said they faced issues with data quality (40%), data silos (40%) and the speed of data integration (36%). Data privacy (30%) and the use of inaccurate or outdated datasets (30%) were also common concerns. Just over two-fifths (43%) also feared they did not have the skills to implement predictive analytics. 

Another problem revealed by the research was a lack of requisite data literacy in organisations. Three-quarters (76%) of respondents said more data literacy was essential for employees to recognise the limitations of the technology, and it was seen as similarly important in helping them explain to customers and other stakeholders how decisions using predictive analytics are made (77%). 

“The financial services industry is undergoing rapid data transformation,” said Adam Mayer, a senior manager at Qlik. “Predictive analytics will play a key role in empowering employees to take more informed actions, with forecasts helping them consider what might happen, as well as what has happened before, when making decisions. However, our research has shown that many IT leaders are yet to fully trust the insights from predictive analytics and the impact these decisions could have on their customers.”