Delivering a successful data strategy in government

A well thought out data strategy is essential to enhancing the pace and impact of transformation across government - the challenge for data leaders is delivery. 

For a data strategy to succeed, it requires a significant shift of people and activity from across the business. Government Transformation Magazine asked data leaders from the UK and Canada what it takes to make this happen. 

Getting the strategy right

When it comes to improving the chances of data transformation success, Neil McIvor, Chief Data Officer at the Department for Education (DfE), says the data strategy itself, which can be misunderstood or undervalued, is where he sees mistakes. “There is a lot of variation across organisations on data strategies, from those that have such a high level data strategy that it doesn't really mean anything, to those that are so much in the weeds, it’s more like a delivery plan.”

While data has risen up the agenda, outside of departments’ data experts “there is still insufficient recognition of the importance of data and what the role of the data function needs to be”, McIvor says. 

With departments still at varying levels of  data maturity, the Central Digital and Data Office (CDDO) continues to work closely with the rest of government on their data strategy, putting in place guidelines to ensure that data is trusted and that the Civil Service builds the architecture to be interoperable.

The CDDO recently launched its Data Maturity Assessment (DMA), which enables government to assess their data environments and focus resources and energy on the right data initiatives based on their current capability and effectiveness. 

McIvor believes these initiatives, and the work of bodies like the CDDO more generally, is instrumental in helping keep government on the right track with their data strategies and ensuring they are fit for purpose. 

Eyes on the prize

Stephen Burt, Chief Data Officer for the Government of Canada at the Treasury Board Secretariat, says to get a data strategy off the ground it needs to prove that there is a business case to do something differently with data: “it must be led by needs of the organisation,” he notes.

“There is a tendency for data strategies to get too fascinated by the problems they are trying to solve or the right standards, policies, and frameworks, without actually moving to action.”

He says a focus on specific data deliverables will bring the rest of the organisation along on the journey, hold data teams accountable to timelines and accelerate results. 

Burt has spent the last twelve months mapping out the Canadian Government’s latest Federal Public Service Data Strategy, which he says is a mix of short-term deliverables and stretch goals for the next three years based on four concrete mission areas. Burt says this helps provide a clear view of how data fits into company goals: “when you've got a new programme or policy area that you're developing, it’s important to actually think about what you're going to want at the other end of it and work backwards." 

A failure to clearly communicate the business case for data, and the value it can bring, will see data strategies rapidly lose momentum.

"This is something I sees data teams struggle with," says Caroline Carruthers, Chief Executive at Carruthers and Jackson and co-author of the Chief Data Officer's Playbook. Carruthers advises organisations, including many government departments, on how to build data strategies, and was the first woman to take on the role of CDO in the UK public sector for Network Rail.

“I've seen how easy it is for data enthusiasts to get fixated on solving a problem then quickly move on to the next. But you have got to be able to celebrate the successes - no matter how small - and use this as evidence to demonstrate and articulate to the rest of the organisation what you've done so that you can continue to grow your data strategy.”

This is especially important given the current context data teams are operating in, where the days of waiting two or three years to get value from something are long gone. “If government organisations fail to demonstrate the value of their data strategy within three or four months, they need to go back and look at their strategy again," Carruthers says. 

Part of this means determining what data is valuable and which isn’t - and being prepared to make sacrifices, she adds. “People treat data like it's all the same thing and it's absolutely not. Some of your data is incredibly valuable and should be treated like the crown jewels, while other data is the same as yesterday's newspaper. When implementing a data strategy, the focus needs to be on the crown jewels."

A sustainable legacy

There is a tendency for leaders to want to solve big, complex problems with data. But a legacy should also be sustainable.

Carruthers warns against unrealistic and unattainable expectations: “you can’t boil the ocean. If you go in and try to get 100% accuracy of data or try to entirely fix data quality, the only thing I will guarantee is that you will fail.”

She says there is more value to be had by starting small and building momentum “You don’t have to reach for huge changes and by breaking it down into a smaller set of problems or use cases you can start building a bigger picture around data.”

Finding and focusing on small pockets of success within an organisation is good for morale as well as momentum, Burt says. “It's important to find a few areas where you've already got decent enough datasets. This way you can do something, even if it's small, to start demonstrating results. This will keep your senior executive cadre interested while you continue to build up your data foundations.”

Burt notes there will always be a desire for quick results in government, but at the same time a data strategy needs to be built for the longer term. He recommends having one team and strategy that focuses on the immediate pressing issues and another that is fixed on longer term data building in an expiry date three to five years out. “This forces a strategy refresh to take stock of what was accomplished versus what needs to be re-energised or pivoted to something new.”

Expect the unexpected

A good data strategy will carve out a path, but they are by no means set in stone. Momentum depends on the ability to be flexible, to cope with changing regulatory environments or fundings pressure, and adapt at pace to the unexpected.

“One of the biggest mistakes is when data strategies become too static and never change,” says Carruthers. “It should be a living, breathing document that is aspirational and just out of reach so you are constantly moving towards it."

She adds: "It is not a linear process and there is nothing wrong with taking a few steps left or right, or changing your behaviour to ensure you have an easier journey.”

McIvor agrees that a data strategy needs to be scalable so that it can adapt to continuously evolving scenarios and keep moving forward. “This is also one of the biggest challenges for data leaders," he notes. "It has to be detailed enough not to be too loose, but  flexible enough to move with the times.”

At DfE, McIvor and his team built an interactive dashboard to help schools and local authorities improve levels of pupil attendance. During the teacher strikes earlier this year, the flexibility built into the design enabled McIvor to automatically pull attendance registers every hour throughout the day - some 76.5 million records -  to build a bigger picture around the impact of the strike. “We were able to take advantage of the flexibility of the data strategy to scale it and move to accommodate new changes quickly,” he says.

Tom Wilkinson, Chief Data Officer at the Scottish Government, says the delivery of a data strategy is dependent on the agility of an organisation, which means ensuring that it is structured and set up to support the vision of data.

“In practice, this translates into how well connected the department is and how easy it is to collaborate between traditionally siloed teams. This means developing digital tools to help both sharing information and keeping track of how much collaboration there is across boundaries."

Data empowerment 

Richard Davis, Chief Data Officer at Ofcom, stresses the importance of a data strategy that is driven by the people using it.

Davis and his team are in the process of developing data products that deliver maximum value for Ofcom. “Vital to achieving that goal means looking at how to empower the people who are not data professionals to understand what they need to get data to work for them,” he says.

“It is essential not to have a top down diktat where you are telling people what the data strategy looks like. It needs to be driven by people from across the organisation who are actually at the coalface doing the work. We've been listening to them and working with them to capture what they see in their requirements, and use those to build up and be able to identify where there are both challenges and opportunities - whether it's a data professional development or the IT platforms.”

Wilkinson says empowering people to think about and use data differently is key to maximising data transformation within government. He adds that a lack of communication with the public can be fatal when building trust around data. “The Scottish Government’s Data Transformation Framework (DTF) has sought to address this by improving and enabling data reuse in the Scottish Public Sector through data maturity assessments that identify pathways to increase knowledge and skills.”

There is a sense among most data leaders that the biggest obstacles to delivering a data strategy in government are cultural not technical.

Carruthers says she sees organisations frequently fail to distinguish between the two, especially when they get too bogged down by the technology that is meant to manage the data. “There's no piece of software in the history of the world that's made life better on its own. While technology is a really powerful tool in our arsenal to be able to use data better, people have to know why they are using it and where they are going with it.”

Striking a balance between caution and innovation 

Delivering a data strategy is a balancing act: data leaders must walk the line between innovation and caution. On one hand, they must maintain leadership buy-in by delivering the desired results and keeping the organisation aligned. On the other hand, embracing innovation requires a certain level of risk and trepidation. 

Carruthers says data leaders have to address the pendulum: “One half is all your risk-averse activities, like legislation, governance and security, while the other half is driving value in big, exciting projects.”

How much focus you should give each side of the pendulum will depend on how mature the organisation is and how ready they are for the change, she adds. 

McIvor acknowledges that it can be a hard line to walk and striking the right balance means innovation sometimes has to be done “a little bit under the bonnet…you can’t always wait to be asked to do something.”

This also allows data to be seen by the wider organisation as an enabler of change rather than an obstacle, he notes. 

Another valuable lesson for McIvor during DfE’s own data transformation journey has been prioritisation. “To deliver a data strategy properly, you have to get everybody excited and on board - the problem is, everybody then wants everything. You need to have some really hard priorities because you just don't have all the resources to be able to deliver everything everyone wants.” 

Diversity of thought 

Embracing diversity of thought is a critical element to ensuring that a data strategy is supporting those that it is intended to help.

“Multidisciplinary teams are where the magic happens," Burt says. "Not only in terms of domain expertise, but also in terms of ethnicity, gender and socio economic background. Having people with different lenses makes a really big difference to how effective and sustainable your data strategy can be.”

The primary example of this taking shape is in the AI and automated decision-making space, where the focus is on ensuring that algorithms are created without bias. 

When solving a problem, consulting a range of different perspectives is more likely to lead to  a solution that is less obvious and perhaps more innovative. McIvor points out that data transformation should be approached in much the same way: “when you’re building and executing a data strategy, you've got to get a mixture of different ways of thinking involved otherwise you'll go into your own little world.”