Data proved a vital tool for helping public sector organisations respond to the pandemic. Now, public sector data leaders are turning their attention to longer-term priorities, such as empowering business users while ensuring data quality and nurturing a data culture. These were the topics of discussion at a recent Tech Monitor webinar, supported by Alteryx, in which UK public sector data leaders shared their lessons from the past two years and agenda for the future.
The pandemic put government bodies on an emergency footing. The first reaction for many was to spin up new data dashboards. Warwickshire County Council‘s data team integrated external information such as population profiles and the prevalence of health conditions in order to support its response.
“At the time they were trying to identify households that might need additional support and there was a list issued by government of those households,” recalled data management specialist Nina Bobe.
In many cases, these new dashboards were followed by more sophisticated analysis. Oxfordshire County Council, for example, simulated the spread of Covid-19 through its population so that it could respond to real outbreaks more effectively.
“We ran a few thousand simulations of what community transmission might look like,” said John Courouble, the council’s head of data analytics, analytics and visualisation. “We were able to say at any given moment, ‘Does this look like community transmission or does this look like there is … a behavioural factor that is driving particular vectors of transmission of Covid?’, and therefore how might we deal with that in terms of preventative advice and enforcement.”
These are just a few of the many use cases for analytics that emerged during the pandemic, said Alan Jacobson, chief data and analytic officer at Alteryx. “We saw organisations … building automation to be able to handle the load of how to distribute benefits,” he said. “We saw workforce analysis of which offices needed to be closed versus opened [and] things like how to distribute supplies so that the supplies got to the right places.”
Data in the UK public sector: the post-pandemic agenda
Now, the agenda is turning to strategic priorities. For Oxfordshire, an important objective is integrating its systems with the NHS and the police. “There’s an interesting cultural difference,” Courouble explained. “Data sharing work with the police is a bit more like trying to get 100 kites up in the air and when you only get ten of them, you’ve still got ten kites up in the air. Data sharing with the NHS is a bit more like trying to get a hot air balloon in the air. And it’s going to be beautiful and impressive when you’ve got the whole hot air balloon in the air. But there’s no such thing as getting 10% of a hot air balloon in the air.”
Warwickshire is rolling out a new analytics platform, and addressing data quality. “One of the tools we’re looking at is master data management,” said Bobe, “to combine data from different areas of the business at the resident level, and hopefully being able to answer new and interesting questions.”
The Department for International Trade is also turning its attention to data quality. “Making sure that the data is collected consistently across different countries and business units allows us to do data science much more effectively,” explained Anthony Coyne, head of data science at the department.
One challenge to this objective is the widespread use of Excel for data analysis, said Coyne. “Lots of people are doing really interesting analysis on Excel and storing it in their own area. But we want to get that information into a centralised place where everyone can see it, and have a shared version of what’s going on because that’s what really allows data science to happen really effectively.”
Data democratisation in the UK public sector
Coyne’s comments reflect a tension between empowering users to conduct their own analysis and maintaining data quality.
“In the public and private sector, we see this real drive towards democratising analytics, having more people able to go on this journey,” said Jacobson. “The reality is there are more data-driven questions coming in every day than there are people at any of your agencies that necessarily have either the right tech sitting on their desk or the right skills to be able to provide that data-driven answer. So we see a big effort at upskilling the workforce and providing technologies that make it easier for them to go on this journey to get more data-driven answers.”
This impetus to democratise data is in tension with the desire by data teams to ensure quality in both data and analysis, said Couroble. Business users are becoming increasingly familiar with data technologies, but they may not have the accompanying statistical knowledge.
“There’s a huge risk when people are good with data technologies that if they’re either bad at statistical understanding or they are bad at understanding the subject matter, they’re going to come up with beautiful, wrong answers.”
Dobe added that empowering users with analytics needs to be accompanied with a shared responsibility for data quality. Senior managers need to recognise the importance of collecting good quality data from the outset,” she said. “You’re only going to get it out what you’ve put in.”
Jacobson warned data leaders against policing users too rigidly, however. “We have to trust that our domain experts, when they do their calculations and do their analysis, that they know how to check their work and that they are going to do the appropriate due diligence,” he said. “It cannot be that you’re responsible for checking all their work.”
Data culture in the UK public sector
Developing a strong ‘data culture’, in which the whole organisation, understands the value and responsibilities of data, is often mentioned as a vital component of success.
The UK government already has a strong culture of producing and using traditional analysis, explained Coyne. Where its data culture may need improvement concerns the use of data science.
“With data science, people either think it’s going to be like Skynet and take their jobs or that it’s complete rubbish and it’s been massively oversold,” he said. An effective data culture requires an understanding that “data science can help in particular, narrow domains and really help automate certain things,” he said.
How can public organisations develop a data culture? For Courouble, it requires celebrating successful projects. “The way to build a data culture is to do good things, shout about those good things, and cause more people to want to do good things of the same kind,” he said.
For Jacobson, an effective data culture is one that is inclusive and open minded. Not everyone will understand natural language processing or other advanced techniques, but “those things tend to work best in environments where people are inquisitive, curious and open to learning,” he said. “organisations that have those kind of learning mindsets and open and inclusive cultures tend to do best on, on accelerating on these analytic journeys.”