Improving productivity and the need for service design
I’m still trying to make sense of what government priorities to increase productivity really add up to.
Yesterday, I read about how AI should replace some work of civil servants. This is the latest in a series of UK government announcements about productivity. There were two specific things here that I noted down:
“No person’s substantive time should be spent on a task where digital or AI can do it better, quicker and to the same high quality and standard.”
And:
“If we push forward with the digitisation of government services. There are up to £45bn worth of savings and productivity benefits, ready to be realised.”
There is some nuance here, with that £45bn saving being set out as achievable “even before AI is deployed.”
This made me think back to an interview I heard with Darren Jones, Chief Secretary to the Treasury since July 2024. He was interviewed by Nick Robinson for Radio 4’s Political Thinking last November and said:
“…how we use technology to improve day to day administration and productivity in public services will release tens of billions of pounds that can then be spent on other priorities for people.”
However, it was the next part of the interview that really caught my attention. Specifically, the example he used to demonstrate where productivity could be improved in government. He talked about the spending review process, the formal government procedure that allocates funds to departments and services:
“I was surprised how analogue the [spending review] process still is […] one of the changes I’m making is to digitise it, just so we can use data much more effectively and make better strategic decisions… I write a letter to a secretary of state, they respond with a letter and an Excel spreadsheet.”
What’s being described here is an inefficient analogue set of processes. It’s the need to redesign a workflow that uses data and shared information more effectively. It’s also the potential to use technology to join up a whole process without the need for letters and spreadsheets.
What I’m describing is the work of rethinking the service, or service design. This type of work is most of digital transformation, most of the time in government. If you were still in any doubt, most of government and many services still run on spreadsheets and combinations of email, report writing, and formal meeting structures. I will say here as well that much of the private sector isn’t in much better shape.
In this case of spend review, what Darren Jones is describing is the goal of more real-time collaboration, and the visibility and exchanges needed to reach key policy and spend decisions. It’s then about how you present and communicate those decisions. This is an opportunity to go much further and to think about how the spend review process could be completely redesigned to deliver the right outcomes both effectively and efficiently.
Skipping the first step… service design
Thinking about productivity, AI, and the need for service design, I heard Rachel Coldicutt make this point most clearly last year, saying:
“…if anything is going to unlock people’s time and ability to do more things – it is service design […] In 2024, most things that most people need to get done do not require an astonishingly complex tech stack and state-of-the-art machine learning.”
Service design has to be the first step in understanding how digital transformation can make our government systems more productive.
While I think the intent here is okay, there’s a real danger if we’re talking about automating tasks and using AI without first carefully rethinking and designing underlying systems and processes.
We’re in danger of becoming trapped at a micro-level of optimisation, where we increasingly see people working with AI-tools being a bit more productive… For example, being able to auto-generate the next letter, email, spreadsheet, or report in a manual process that hasn’t changed how ministers make decisions for years. Previous new technologies, like email and spreadsheets, have already been added into what are still analogue processes, over and over again. All of these technologies also promised some level of productivity improvement. But on the flip side, when badly designed or implemented, they may have made productivity worse, creating or expanding parts of processes to become more work.
While individual productivity improvements help people cope better with bureaucratic or complex, rule-based procedures, these processes are the very things that need carefully unravelling. They need questioning and reconfiguring to realise the type of transformation that addresses root causes of inefficiencies. That needs to happen through an understanding and focus on exploring, testing and finding the best way to achieve outcomes with technology – and to Rachel’s point, that tech stack, designed in the right ways, might still be simple, or even harness existing technologies and solutions much more effectively.
With a fixation on realising the benefits of state-of-the art technologies, the danger is AI as a sticking plaster, with many layers of unintended consequences, rather than investing in the careful work required to integrate and use technology in ways that support meaningful transformation.
Productivity gains… to be realised
My bet is that, individually, AI will continue to make us all feel a bit more productive in the coming years, but within the context of our current roles, tasks, and the processes we follow. However, in our workplaces, where these tools are becoming increasingly common and accessible, everyone is going to continue to be busier than ever. I thought it said a lot last week when Google co-founder Sergey Brin described the sweet spot for individual productivity (with AI) as a 60-hour week – reported by the New York Times.
In summary: we need more service design or we’ll all be doing more work within the same inefficient systems of government.
The place service design is most needed is inside government… looking at ways of working, information flows, and especially how we work with data. It’s what I describe as inside-out transformation – if you really want to unlock productivity and cost savings, it’s how you start to deal with internal failure demand – building the foundations for doing more with emerging technologies at the same time as focusing on what the service or system needs to do.
I’ve not got to the point about how savings from productivity can “be spent on other priorities for people”… My view is that the opportunity is helping government to operate and run more efficiently to meet increasing demand and pressures in the short-medium term. That might also mean AI replaces some roles, tasks and processes that enable government to invest elsewhere in other things.
The final consideration here is the test of “where AI can do it better.” There’s lots more to unpack about what is lost when we remove human interactions, decisions, and interventions from the systems we operate. The only conclusion I’ll make for now is that we have to proceed with care, or with the type of careful facilitation, planning, and evaluation that comes with recognising the importance of service design.
This is my blog where I’ve been writing for 20 years. You can follow all of my posts by subscribing to this RSS feed. You can also find me on Bluesky and LinkedIn.