Skip to Content

From couch potato to 5K: It’s time for the public sector to get active on data

Philipp Fuerst & Liz Henderson
Sep 9, 2024

Data governance and data management are a bit like physical exercise: we know it matters, but it is just too easy to put off. Now that government agencies are looking to data and generative AI to increase their productivity, senior leaders in the public sector need to take the fitness of their data more seriously.

Public sector leaders are facing ongoing budget constraints, shortages of people and skills, and demands for greater efficiency and effectiveness. In this context, many are asking how generative AI (Gen AI) can help. But success with Gen AI rests on data. High-quality data – timely, well-managed and available securely to all parties who need it – is the foundation not only for training AI, but for building a more efficient, cost effective and sustainable public sector in this century.

It is no wonder that data governance and data management are topics public sector agencies can no longer ignore.

Data quality must also be the bedrock of trusted, transparent Gen AI use that protects public sector organizations against public mistrust concerning the use of their data. The old computer science adage “garbage in, garbage out” holds true for all Gen AI applications, regardless of model specification, and compute power.

Why data governance and data management need a personal trainer

How do personal trainers motivate people to start exercising? First, by making clear just how important physical activity is for well-being and avoiding the leading civilizational diseases; and second, by designing easy-to-follow workout routines that lower the threshold to becoming active. Similar tactics are needed when it comes to data quality in the public sector.

For a long time, government agencies had followed the principle of “as much as necessary, as little as possible” when it came to data governance and data management. This meant that overall data quality was not in the best of health.

Essentially, data governance and management were largely restricted to three areas:

  • Data protection to implement legal provisions, such as GDPR
  • Data security to safeguard sensitive organizational and citizen information
  • Data provision for select reports and analysis.

While some agencies have gone further, for example, by managing data to enable process automation, very few have embarked on a full-scale, cross-system and cross-application or even ecosystem approach to data governance and data management.

Gen AI as a catalyst moment

The technical characteristic of this topic, the long-term nature of the challenges it implies, and the wide range of possible applications or use cases for data across the public sector, can mean that the value of data governance and data management are hard to distil into a business case that lands at CxO or ministerial level. Nonetheless, their importance must be conveyed. Data quality needs to become a strategic issue that all public sector organizations must get to grips with from the top down. And it is Gen AI that is giving the impetus to turn the data “couch potato” approach into something fitter and leaner.

With many government agencies moving from Gen AI POCs (Proof of Concepts) towards production, they begin to understand that the value of off-the-shelf large language models (LLMs) that were trained with generic, publicly available data is limited. The true potential of utilizing Gen AI and LLMs – for example to query specific laws, regulations, and directives, provide administrative assistance to case managers, help write government press releases or provide citizens with chatbots for government service – can only be reaped by leveraging a local body of knowledge.

The must-haves of data governance and data management

Local, agency-specific knowledge of data best practices needs to be built from the ground up and continuously updated. Someone at the most senior level within the organization needs to grasp the data nettle. They must decide on the type of data that can go into this local body of knowledge for the respective use case. This involves, of course, compliance with data privacy and security standards – something that the public sector is quite familiar with and capable of.

The newer – and more daunting – task facing those taking responsibility for data is how to take care of data quality in the new world of Gen AI. In a nutshell, unless data errors, inconsistencies, and biases are systematically being taken care of, Gen AI is at risk of producing “insights” that should never see the light of day, much less be used by public sector agencies for decision making and policy setting.

From plan to action – start with the bigger picture

The scale of the technical complexities of data management can be demoralizing. It is therefore critical not to lose sight of the bigger picture. To get started, focus your data needs on a specific use case. We recommend defining a handful of use cases where you can start to experiment with the value and benefits to be gained from using Gen AI. This will help identify your data governance and management needs as gaps in skills, technology and data will undoubtedly surface.

Of course, whatever value opportunities you identify and no matter what productivity increases you’re seeking, unless you get the data quality right, it’s unlikely that you’ll realize the outcomes you’re looking for:

“A number of public sector organisations are already successfully using AI for tasks ranging from fraud detection to answering customer queries. The potential uses for AI in the public sector are significant, but have to be balanced with ethical, fairness and safety considerations.”


UK Government Digital Services (Office for Artificial Intelligence)

Considerations such as these pertain as much to the ‘what’ as to the ‘how’ you use data and Gen AI. The difference between data governance and data management is bit like the difference between your training plan and the actual physical workout. Data governance is your plan. It lays out the key principles of what needs and can be done with data. In effect setting the guardrails to work within, for safe and secure collection, storage, and use of the data, while protecting the organization and the data from potential risks. Data management is what happens next – how you navigate your approach to data, including AI-generated data, within those governance guardrails.

What can Chief Data Officers and other leaders do next?

We recommend a six-step approach to creating leaner and fitter data governance and management practices in order to leverage the power of Gen AI for greater efficiency and cost effectiveness:

  • Step 1: Establish the value case – why you’re doing this – and build support for it across the organization. Is it for regulatory compliance, or are there opportunities for efficiency gains or perhaps prospects for innovation to enable a better citizen experience?
  • Step 2: Consider your approach and initial use cases. Who are your data producers, owners and consumers?  In thinking about Gen AI opportunities, is there a departmental or multiple-department process that needs to be documented? As you document the process you can identify the potential activities where Gen AI might add value and increase productivity.
  • Step 3: Look at the type and quality of data needed. Consider what level of variety you need to teach the LLM so that the outcome is both fair and representative. Explore whether you have the required data and if it is at the right quality level. If your data is falling short, ask what you can do without the full set of trusted data and what steps you need to develop to bridge the missing data gap.
  • Step 4: Establish your data governance regime. Assess what data governance you need to protect the data you already have or need to obtain. A well-formulated data governance regime should be informed by the organization’s business goals and have direct implications on how the data needs to be practically used and managed.
  • Step 5: Draw up and communicate effective data management practices, including determining who is supposed and allowed to do what with specific pieces of data.
  • Step 6: Monitor the way in which your data is collected, stored, accessed and reused to ensure optimum data quality (e.g., accuracy, consistency, reliability).

Building momentum with a compelling value case, gaining stakeholder support, and arrival at a common organizational agenda will give you the initial building blocks on which to transform your data management and governance activities, one brick, aka data product, at a time.

It’s not a sprint – but not a marathon, either

Think of it like middle-distance running – it will take a bit of time and effort, not too much, but not too little, to get to the required fitness level. As Gen AI becomes increasingly mainstream in the delivery of public sector services, government agencies need to tackle data governance and management if they want to leverage its power. And beyond just Gen AI, there are other ways to leverage the data dividend, with better analyses, forecasts of future trends, process automation, etc.

The key is to do it right and do it comprehensively. Returning to our sports analogy: a good basic fitness level enables you to become better in all kinds of sports. Or, in other words, getting your data governance and data management approach up to speed will enable you to transform your data regime from “couch potato to 5K” for better citizen, societal and economic outcomes.  

Authors

Dr. Philipp Fuerst

VP Data-Driven Government & Offer Leader, Global Public Sector
To unlock the value of their data, governments need to make organizational changes and meet new technology requirements. Yet, the many examples of public sector agencies that have already successfully embarked on the journey to become data-driven organizations show that these hurdles can be overcome. Their gains in decision making, operational efficiency and citizen experience are tangible and significant. Our clients believe the benefits they have reaped are well worth the effort.

Liz Henderson

Executive advisor, Capgemini
I am an experienced leader in digital, data, and AI, dedicated to empowering organisations unlock the full potential of their data. By fostering a culture of Capability, Confidence, and Curiosity, to drive improved decision-making, innovation, and competitiveness—ultimately leading to greater success and growth. With my strong background in advising on transformation initiatives, I work closely with leadership teams to develop data strategies that recognise data as a valuable asset. These strategies are aligned with business objectives, enabling organisations to leverage their data for strategic success. Recognised as a leader in the data industry, I am known for my strategic acumen, innovative thinking, and ability to navigate complex challenges. My global experience spans Europe, Asia, and the Americas, and I am respected for delivering long-term solutions that create lasting value.