The Case of the UK’s Government Digital Service: The Professionalisation of a Paradigmatic Public Digital Agency

This case study explores the evolution of the UK’s Government Digital Service (GDS) since its inception in 2011. GDS is seen by many as the gold standard digital agency, copied in multiple instances and praised for its achievements. Drawing on expert interviews with former and current GDS employees, we look at the evolution of GDS through three distinct phases. We argue that at the core of GDS foundation is a coherent and paradigmatic approach to government digital transformation. GDS can be viewed as a conscious effort to create new capabilities within government through a new organisation that questioned dominant routines in government IT and generated new approaches, frameworks, and skills to do digital transformation differently. These new routines have, over time, become part of the central government machinery. GDS has successfully professionalised these new routines beyond its own organisational boundaries. However, in doing so, GDS has itself become a distinctly different organisation that now faces the challenge of how to rejuvenate its original dynamism.


INTRODUCTION
Since it was founded in 2011, the United Kingdom's Government Digital Service ( GDS ) has in many ways become the gold standard public digital agency.It has won awards and praise among its peers and heads of governments, and its blueprint has been copied in numerous countries [Clarke 2019 ].The story of its foundation and success has been told in op-eds, blog posts [Ross 2018 ;Greenway 2020 ], and academic case studies [Birkinshaw 28:2 • R. Kattel and V. Takala and Duncan 2014 ;Eaves and Goldberg 2017 ]. 1 However, GDS has evolved significantly in the last decade.From a small office with a dozen or so people, it has grown into an agency with more than 850 employees; its budget has increased from £37 million in 2012/13 to £455 million for the period from April 2016 to March 2020 [National Audit Office 2017 ].Also, the perception of its success and impact has changed considerably: from a digital gold standard digital agency to "a sad tale", to quote The Economist [ 2020 ].In this article, we are interested in GDS's evolution; that is, how and why it has changed over time.Understanding its evolution sheds light on how digital transformation in government is changing and what kind of organisations can drive it forward.
Drawing on a combined total of 24 interviews with former and current GDS employees, we analyse the evolution of GDS from 2011 to the present.One of the key themes to emerge from the interviews was distinct periods in GDS' development.Accordingly, after briefly introducing the context into which GDS was "born", as well as our research design, the paper is structured around three key periods in the evolution of GDS.We conclude the article by reflecting on the main lessons that other governments can draw from the GDS' story.
In what follows, we aim to show that the evolution of GDS can be viewed as a conscious effort to create new capabilities within government through a new organisation that questioned dominant routines in government IT and generated new approaches, frameworks, and skills to do digital transformation differently.These new routines have, over time, become part of the central government machinery.GDS has successfully professionalised these new routines beyond its own organisational boundaries.However, in doing so, GDS has itself become a distinctly different organisation and now faces the challenges of how to rejuvenate its original dynamism.

CON TEXT: GO VERNMEN T DIGI TAL TRANSFORMATION IN THE UK BEFORE GDS
A wide range of research has explored the history of government IT in the UK, mostly in highly critical terms [see, for example, Dunleavy et al. 2006 ;Perri 2007 ;Clarke 2019 ].A long list of large-scale failures is often attributed to the influence of New Public Management ( NPM ) reforms, which from the 1980s onward led to waves of outsourcing IT functions to the private sector.Comparing a range of countries, Dunleavy et al. [ 2006 : 70] found that countries with the most enthusiastic uptake of NPM had fared particularly poorly in exploiting digitalisation, with the UK emerging as "a world leader in ineffective IT schemes for government".According to Dunleavy et al. [ 2006 ], NPM's emphasis on the principles of disaggregation (chunking-up government hierarchies into smaller organisations), competition (both with private-sector contractors as well as within government) and incentivisation (built on pecuniary motivations instead of professionalism) fragmented administrative institutions and dramatically increased the institutional complexity of policy systems.By hollowing out public sector capabilities and bringing in new contractually based risks and barriers to cross-government policymaking, NPM drastically impaired government IT modernisation.Importantly, however, the NPM reforms also paved a way for a more porous and diverse organisational configurations in government, enabling the creation of non-standardised organisations, hiring and promotion practices [Lapuente and Van de Walle 2020 ].
GDS was thus a response to major and widespread dissatisfaction with government IT in the UK.Following a series of high-profile IT failures, the UK Parliament's Public Administration Select Committee published a report entitled "Government and IT-a Recipe for Rip Offs: Time for a New Approach" in 2011 [House of Commons Public Administration Committee 2011 ].As summarised by Clarke [ 2019 : 360] "The report highlighted a dearth of IT expertise; a lack of centralised, horizontal IT governance; and reliance on large-scale, long-term contracting with a small number of large private providers as central culprits driving IT failings in the government."At the same time, the newly elected coalition government appointed the internet entrepreneur Martha Lane Fox as the "UK's Digital Champion", and commissioned her to review the government's online presence.Fox [ 2010 ] recommended The case of the UK's GDS: The Professionalisation of a Paradigmatic Public Digital Agency • 28:3 that there should be a new, central digital team that would have absolute control of the overall user experience across all digital channels, and it should be headed by a CEO reporting directly to the Cabinet Secretary.
The UK government closely followed Fox's recommendations, and in 2011 the UK GDS was established.Mike Bracken, former lead of The Guardian newspaper's digital transition, was selected as the organisation's first executive director.As a result of major early successes and active policy transfer, GDS soon became perceived as a global leader in government digital transformation, mimicked by many countries across the world.

RELATED RESEARCH ON GO VERNMEN T DIGI TAL UNI TS
Despite the active policy transfer that followed GDS' early success, public administration research has remained relatively silent on the strengths and weaknesses of dedicated Government Digital Units ( DGUs ).Notable exceptions include Margetts and Naumann [ 2017 ], who compare the uptake of the "government as a platform model" in Estonia and the UK, and Mergel [ 2019 ], who analyses the origins and functions of a set of prominent DGUs'.In response to the paucity of research, Clarke [ 2019 ] compares six early DGUs and defines DGUs as a coherent set of public administration phenomena "inasmuch as each represents a government unit operating at the center of the state with a shared philosophy, evident in their common commitment to agile, user-centred design, pluralistic procurement and centralized platforms/components, and in their shared rejection of processfirst, hierarchical, formal bureaucratic culture in favor of a tech startup-inspired culture" [Clarke 2019 : 365].Clarke calls for further research on four critical areas which include: appraising the successes of DGUs thus far; comparing the governance structures and powers of DGUs; assessing their long-term sustainability; and attending to their accountability challenges.It is therefore central to analyse how particular DGUs evolve over time, and how they fit in and align with the rest of the public sector.

STUDY DESIGN AND METHODOLOGY
In order to tell the GDS story from its birth in 2011 to the present, we interviewed a combined total of 24 current and former GDS employees.The interviews took place between December 2019 and March 2020.We recruited the former GDS employees (11 in total) through a snowball sampling method, where we initially asked a number of former employees who had worked at GDS during its early days to identify individuals who would be able to talk to its foundations from diverse perspectives.With regards to current employees, GDS provided access as well as suggestions for individuals to be interviewed (14 in total).Interviews were conducted either at the premises of GDS, University College London, or online via video conferencing software, and they lasted between 40 and 90 minutes.Combined, our sample includes individuals in high-level managerial positions, as well as experts from a variety of functions, including design, data science, strategy, product management, and software engineering.The interviews focused on themes such as: "what was the focus of GDS activities and why"; "what value did GDS provide and how?" "how did GDS see its relations with other public organisations and with market actors?"; and "what was the focus of internal organisational practices?"(please see the Appendix for the interview guide and list of people interviewed).Therefore, our analysis is based on the "internal" view from GDS, critically interpreted.Future research should examine the "external" view on GDS from organisations in central government and local authorities that have worked closely with it during the past years.
A key theme to emerge from the interviews was distinct periods in GDS' development.Accordingly, the remainder of the article is structured around three key periods in the evolution of GDS.A summary table (Table 1 ) of the key characteristics of each period is provided at the end of the analysis.

GDS: THE CREATION OF A NEW DIGITAL TRANSFORMATION PARADIGM
Low trust in digital identities has perhaps been the key challenge for many Western governments in digitalising public organisations and services [Halperin and Backhouse 2012 ].GDS was path-breaking in that it was able to create a successful digital transformation blueprint that does not rely on (digital) identity.The digital transformation promoted by GDS is focused on information provenance : the main aim is to ensure that the government is digitally able to provide straightforward and truthful answers to questions (that is, online searches) by citizens.Drawing on our interview data, the GDS approach has evolved through three stages that build upon each other in a cumulative fashion, with a major overlap between the last two phases.
First, a "strategy is delivery" phase that introduced new concepts of value (user needs), brought in new skills and new ways of working (service design, working in the open and on the internet, agile practices), and broke existing market structures of IT providers (through procurement and spend controls).According to many of our interviewees, GDS acted in the first phase as start-up up driven by the charisma of its leaders.This phase lasted from 2011 to 2015, ending with the departure of the original leadership team of Mike Bracken, Tom Loosemore, and Ben Terrett.
Second, a government-as-a-platform phase that built on the practices of the first phase but deepened the focus on products that could be used by other government departments, agencies, and local councils (examples of such products are GOV.UK Notify, Pay, Verify).The idea of value shifted towards innovations that enable other public agents to work better and cheaper.This was based on a less combative and more collaborative approach towards partners in government.During this phase, GDS was able to show that it had created significant efficiencies and savings across government.This phase lasted from 2015 to 2017, culminating in the publication of the Government Transformation Strategy in 2017 [Government Digital Service 2017 ].
Finally, a professionalisation phase that builds on the two previous phases, but with the focus shifting ever more towards building and fostering partnerships within the public sector and, above all, on emphasising the importance of government-wide profession (called digital, data and technology profession), unified job descriptions, and career progression pathways.While the focus on delivery still plays an important role, the idea of government-as-a-platform is becoming less central.The value proposition is described through user needs, efficiencies and savings, and increasingly through the notions of trust and trustworthiness (of its main products and as a partner).This, we argue, is the GDS's current phase.
One of our interviewees2 summarised the three-phased evolution of GDS as follows: "I think that GDS has been on sort of three-to-five-year blocks of the journey.And each iteration has re-evaluated what value it is bringing to the government and what its key purposes are.So in the early days, it was to build a thing and shout about it, and then it became about heavily enforcing policy by spending controls and so on.Then, over time, it became more collaborative, and let's talk about this and work with the departments."(Interviewee 1) We describe each phase in more detail below, weaving in major events during each period.

A Unique Ethos and Pull.
In line with Martha Lane Fox's recommendations, the early years of GDS were characterised by the unusual degree of high-level political support it enjoyed.The GDS leadership team, for its part, was astute in translating this support into building a very unusual public organisation.For example, from early on, GDS set out to hire people from both the private sector as well as the civil service.Whilst the practice of employing both contractors and civil servants inevitably led to unclear career progression pathways and inequalities in pay, it also enabled GDS to grow quickly.The hiring of like-minded people worked effectively to exert normative control [Mintzberg 1996 ], as new employees quickly bought into the leadership team's emphasis on user needs, design principles, and the revamping of procurement practices.The BBC proved a particularly good source of talented employees for GDS.However, already in 2014 GDS introduced a new fast stream in the British civil service, the DDaT ( Digital, Data and Technology ) Fast Stream, which marked GDS' early efforts to standardise and professionalise the new skills and novel ways of working that it sought to bring into government.
A crucial reason why so many talented programmers and designers wanted to join GDS, despite lucrative offers in the private sector, was their long-standing perception that the government was systematically failing on digital.The widely publicised failings on the part of the NHS to digitise its services (see, for example, Currie and Guah 2007 ; Currie 2012 ) had further increased this perception (Interviewee 7).Many of the early employees shared a vision of value the internet could provide to both the government and society at large: "Actually what was important to GDS was people like Paul Downey, who was old school . . .[in that] he believed in the open web.And so did Tom and so did Mike; they had a vision of how the web should work.And the government actually turned out to be the most relevant place for it to be brought to light at the time when we were also starting to see Facebook and Google and everyone close it down."(Interviewee 7) In addition to the commitment towards an open web, the early leadership team shared a belief in the utility of design thinking in transforming rigid government structures. 3The initial "alpha" version of the first, and even to this day best-known GDS product, GOV.UK, was developed in 12 weeks for £261,000.In a blog post dating to 2011, GDS leadership team member Tom Loosemore was quick to hail the exceptional nature of the feat: "A boundary-pushing experimental prototype (aka a Minimum Viable Product) was delivered by an in-house team working in an open, agile way, placing user needs at the core of the design process.This isn't a new approach, but it's one that's still all too rare across government."[Loosemore 2011 ] The aforementioned elements made GDS into a unique public institution that was able to compete for the most sought-after employees with the private sector.One interviewee, who had a background in private sector consultancy, described the unusual excitement and pull of early GDS: "And I literally walked into the [GDS] office having worked with a lot of government departments during my time at private consultancies, and it was just like, you know, totally different.I just felt like I so want to work here; this is what I want to do.I want to deliver change to government, not just write documents or have strategy meetings about how we might do that.So even just walking in felt so different to anything I've experienced before."(Interviewee 2) The above comment underscores the ability of the early GDS to provide the organisation, and thereby its employees, a sense of unique purpose and mission.
In January 2012, GOV.UK went into beta, and by October it replaced all Directgov websites to create a single domain for accessing both services and information.During that same year, GDS announced a two-year project to update 25 of the largest transactional services in government, with these becoming known as the exemplar projects.The focus on the largest transactional services helped GDS showcase its novel ways of working and, above all, its radical shift in the way it gathered and analysed information for its activities: its relentless focus on users.

Changing the Epistemological Focus of Government.
Governments typically rely on wide-ranging analytical tools from qualitative analysis to complex models to generate knowledge about policy options prior to their implementation and ex-post analysis.Since the early 2000s, governments have been looking to create innovation labs [Tõnurist et al. 2017 ] and one of the key innovations such labs brought to the government was to change the lenses policymakers use in order to gather knowledge about government activities.Starting with Mindlab in Denmark, such labs were able to show that, by focusing on understanding user needs, the effectiveness of policies can be increased through better delivery [Carstensen and Bason 2012 ].In its formative years, GDS based its entire way of working on this epistemological shift.In his first blog post at GDS, Mike Bracken defined service transformation as follows: "So, when we talk about 'transformation', we don't just mean messing about with the hardware and software that makes things happen.We mean thinking about the whole service, getting a multidisciplinary team together, and transforming the experience for users, for the people who are seeking help when they put in a claim.We mean delivering a better experience for them, doing something that makes a genuine difference to their lives."[Bracken 2014 ] In the early days of GDS, this meant taking down many of the pre-existing websites, as they often provided superfluous information, or in some cases were outright misleading.This is perhaps best exemplified by one of the most frequently used examples from that time.A general query on "What is the UK government's policy on Afghanistan?" resulted in a number of policy documents providing different answers to the same question.GDS set out to change this early on (Interviewee 3).
Thus, in the first phase of GDS, the focus on user needs was expressed by two basic principles: First, citizens searching for information (often through search engines such as Google) about public policies and services should easily find a truthful and simple answer; and second, the resulting transactions (filing a claim, changing a license) should be as user friendly as possible.GOV.UK purposely mimics the experience of using commercial online platforms, with the important addition of value neutrality (it is agnostic regarding policy changes).
This "utilitarian" approach to user needs enabled GDS to instil alternative principles of information (one question should have one answer) that, in turn, necessitated changes in data management (quality of registries), user information analysis, and the tools that the government uses to generate and publish content.While the approach relied on an epistemological shift to gathering and analysing information about citizen needs, it also harboured an even more radical need to reorganise government structure, as expressed in an interview: "... the example that is often used is that if you import a car, for example, you have to deal with six government agencies, vehicles and tax, and they typically make you fill in different bits of paper for each one and get a reference from one to the other.... you could kind of do that on the front of things and behind the scenes, it would still be separate, or you have a decent data architecture where you could literally do that stuff programmatically.And then you started to look at massive changes of efficiency and even potential need for certain departments to exist in the shape that they do today.That was always the big idea, the SimCity style, the diagram of government as a platform underneath."(Interviewee 9) The utilitarian approach is also expressed in the GDS design, which consciously saw itself as continuing the design canon of Frank Lloyd Wright , Joseph Bazalgette, and Norman Foster, with its guiding principles of dignity and simplicity.Creating a standard digital visual outlook for the government followed from the idea that simple form leads to better servicing of user needs. 4However, the fact that the utilitarian approach and the canon itself might be based on a hierarchical (male, white) view of user value was rarely considered.Our interviewees reflected on the limitations of the utilitarian focus on user needs, and its underlying epistemological shift: "I think there is a strong assumption-and I genuinely don't know if this is right or wrong-that people want their interactions with the state to be brief, functional, and done as fast as possible.So, in essence, a very transactional relationship with the state."(Interviewee 4) The case of the UK's GDS: The Professionalisation of a Paradigmatic Public Digital Agency • 28:7 "I think one of the things. . .we did do in GDS was focus too much on user's digital experiences, and not necessarily enough on the broader set of user needs that people have.We were kind of disrespectful to the user, we didn't value the user needs of people across all the channels or who might not be able to use digital channels, we were very much focused on it being a digital experience."(Interviewee 2) 5.1.3Revamping Procurement.Prior to GDS, the way that the government structured information was tightly connected to the way it developed digital infrastructure, which was mostly by outsourcing it through large procurement contracts (see Section 4 ).As one interviewee (6) explained, this created a system of "failure by design", whereby a failed public IT project eventually led to a new large-scale procurement contract.Even the simplest website management was outsourced at huge costs, as one interviewee recollected: "You know, when we arrived, there was a site called Business Link . . .I think the contract had been all the procurement to be done with BT, who had outsourced it to IBM.And it was a website with maybe 500 static pages, and it was costing 50 million pounds a year.We ended up funding quite a lot of GDS from the money we saved in closing down Business Link.If you wanted to change a page on that, you had to pay IBM 10,000 pounds.You know, it was basically just the government being ripped off because it didn't understand and again that's the stuff that doesn't really get talked about, partly because it's not unusual.It's happening ever y where all the time."(Interviewee 7) The poor digital infrastructure, in turn, generated a massive failure demand (demand caused by failure to do something right for the users), increasing the strain on actual service provision and IT units in public organisations.Hence, increasing the success rate of IT projects became the second cornerstone of GDS's value proposition (next to user needs), as well as one of its key metrics.In 2013, GDS launched the Digital Services Framework, which laid the ground rules for digital procurement, and in 2014 this was supplemented by new rules for public sector IT contracts. 5he latter rules also formed the basis for spend controls exercised by GDS on all larger government digital projects.In the same year, GDS' Digital Marketplace replaced the G-Cloud CloudStore.
The radical revamping of government digital procurement-effectively a market reshaping and creation of common public sector standards for (agile) digital procurement with effective enforcement mechanism through spend controls on public digital projects-was met with huge resistance from the large IT suppliers: "... IBM had a seat on the board of the DVLA [Driver and Vehicle Licensing Agency].So arm's-length bodies had companies on their boards.IBM was the largest single supplier to DVLA and they had a seat on the board.So we spent. . .nine months to a year getting IBM off the board of the DVLA. . . .And so we used to talk about it all the time in presentations; you know, no more big IT was one of the things we said and we talked about the oligopoly."(Interviewee 7) It can be argued that the first phase of GDS ended in August 2015 with Mike Bracken announcing his departure amid rumours of a wider shake-up.Soon, other key figures such as Tom Loosemore and Ben Terrett also left.However, in a surprising turn of events, the 2015 spending review, instead of slashing GDS's funding, increased it to £450 million over four years until 2020. 6Amid the shakeup, in 2016, the UK claimed the top spot in the United Nations' e-government index, showing the increasing international visibility and impact of GDS.In the GDS context, GaaP was mentioned already in 2013 by Liam Maxwell, at that time the UK Government's chief technology officer, who suggested that "we will implement government as a platform, providing departments with common business functionality that can be reused by multiple users in multiple service areas" [Maxwell 2013 ].
In the period from 2015 to 2017, GaaP arguably became the dominant focus of GDS.Its Sprint 15 7 was dedicated to GaaP; in 2016, GDS launched GO V.UK Pay, GO V.UK Notify, and GO V.UK Verify.In 2016, Principles for Government as a Platform products were published. 8It is important to note, and in comparison to the GOV.UK intervention, that none of these platforms are mandatory and public organisations in the UK (and eventually elsewhere) are free to opt-in or out of using them.By November 2020, 2.02 billion notifications had been sent using Notify, 10.7 million had been made using Pay, and 22 services were using Verify with 7.5 million users. 9 While the uptake of GaaP at GDS was initially strongly influenced by the vision set out by O'Reilly [ 2011 ], the actual application of the idea at GDS was rather different.One interviewee explained to us that, from early on, GDS approached the idea of GaaP with its own specific focus on: ". . .failure demand and getting the government to work together to build common tools for government . . . the stuff they did were much more about how you could decrease failure demand ... we can stop different departments duplicating each other's work while fighting over the same resources and actually pull that stuff together."(Interviewee 5) Thus, from the outset, GaaP became a way for GDS to build on its idea of reducing failure demand, combined with the focus on user needs, as a central tenet of its work and, in doing so, establishing shared building blocks for public organisations to use for their services.
Our interviewees explained that the real value and potential in GaaP lies not in the cost-savings that it might generate (more on this below), but in its enormous potential for unlocking service innovation in the public sector by, in essence, creating unused resources.Thus, "the value in platforms is not necessarily the direct cost savings, although there are some, but in what it unlocks because it makes experimentation cheaper, and iteration cheaper" (Interviewee 8).
"GaaP potentially provides public agencies with more agility as "it isolates certain risks into certain individual pots . . .there's one hopefully well-funded, well designed, robust way of doing it, which means that the risk is in one place and can be much more easily controlled in that space . . .It also means that we can change direction much more easily on a government level . . ." (Interviewee 1).

GaaP Leads to Deeper Partnerships with Other Departments.
Although GDS published a Government Transformation Strategy in 2017, which solidifies the focus on cross-governmental platforms, 10 it has not launched 7 For further details on GDS's 'sprints', and this one in particular, see https://gds.blog.gov.uk/2015/02/04/sprint-15-round-up/ 8 See here for details: https://governmentasaplatform.blog.gov.uk/2016/02/29/governmentasaplatform-foundations/ 9 All of the GOV.UK platforms publish their usage data on respective dashboards.any other platforms to date.The wide consensus is that Pay and Notify work very well, but that Verify is a weaker platform, as underlined by its smaller uptake.As digital competencies of other government departments and agencies have increased, and as competition from the private sector has accelerated, the GaaP approach has started to lose momentum at GDS. "I think maybe there's also been a bit of scepticism towards not necessarily the government as a platform concept, but how the products fit within the market and what they potentially do to the market as well.And some, a little bit of kind of fear . . .I guess that's the fundamental question of, are we actually building something to cover a market failure and should the government be doing that.Should we just look to the market anyway to buy this stuff?"(Interviewee 10) In the GaaP phase, GDS has seen itself less and less as a market shaper and has been increasingly aware of the need to cooperate both with other public organisations as well as with private companies.Furthermore, it has been under ever-increasing pressure to show the cost savings of its platforms and demonstrate that it is not disrupting markets in any negative ways.
GDS was indeed able to show enormous savings during its first three years of activities: it had allegedly saved up to £3.56 billion by 2015. 11With the increased funding after the 2015 spending review, GDS was obligated to show how much it can save for the rest of the government (Interviewee 9).
During this period, GDS' impact on the wider government has increasingly taken place through the spreading of skills in other departments, forming and maintaining communities of practice, and building the professional standards of its core capabilities around digital and design.For many departments, GDS was the catalyst for change: "The thing I would say, though, that we haven't sort of mentioned is that GDS was an effective catalyst, and provided cover for the departments to set up their own teams internally and some departments are much further ahead than ...And so there was a successful effort to introduce skilled leaders and give them a mandate to build their teams and do stuff.And that was quite successful."(Interviewee 12) Importantly, other government departments have caught up with GDS in terms of their digital capabilities: "Departments are smarter consumers of these things, because their capability has grown.All of them always had some degree of capability . . .We have communities around them [GaaP products] as well.We cultivate these communities through things like slack channels where people can ask questions, kind of to us but what we end up seeing is more often than not other teams have gone through the same thing and they share ideas or templates or patterns or whatever."(Interviewee 9) Thus, the success of GDS products inevitably led to shifts in its ways of working across government.Somewhat counterintuitively, the early success of the GDS and increasing spread of digital capabilities across government departments, often driven by GDS, meant that GDS itself became less relevant.

A Focus on Partnerships and Standardisation.
While the 2015 spending review secured GDS a significant funding bump, the political support it enjoyed was much more fickle in the post-Brexit vote era.In 2019, Parliament's Science and Technology Select Committee's Digital Government report (2019, para.7) concluded that there was a slowing down in government's digital momentum and recommended that GDS's purpose should be twofold: "to provide advice to departments when needed, but also to devise and enforce minimum standards consistently across Government digital services."This was foreshadowed in 2018 when the then-GDS Director General Kevin Cunnington wrote in a blog post (2018, para.10) about the progress of the government's digital strategy, focusing on the role of GDS across the government: "As well as giving government the tools and resources to transform the way they operate and build better services, we're also working with departments to create the right structure and support for transformation." Accordingly, since 2018, GDS has focused less on GaaP and more on cross-governmental cooperation, standards, and professionalisation of civil servants in the technology and design jobs.Today, GDS is in charge of digital and technology function in the civil service and this is perhaps the key change in the narrative of what GDS's role is.A long-standing GDS employee outlined the organisation's current role within government: "I think fundamentally professionalising the civil service is the big strand of work, sort of the functional agenda that our chief executive has led, and I think over the last three years there's been a maturing of that.GDS is also head of the Digital, Data and Technology Function across government, which is 17,000 people.And I think it's part of that just enabled us to have a slightly different relationship with departments."(Interviewee 11) However, some have noted a certain shallowness of the uptake in other departments.Its impact has also been rather uneven, leading to a certain shallowness in engagement, as demonstrated in the quote below: "I think when you talk about the content, it wasn't really deep service transformation ... and part of GDS legacy is, unfortunately, loads of people using all of these terms like discovery, alpha and beta, and things like experimentation, prototyping, as a way of working without that vision or goal of why or what they're trying to achieve.So we've almost turned it into a process that they just simply follow.And we've made it really easy for people to prototype infinitely.It's almost like because you've lowered those barriers to being able to do this stuff, because prototyping is quite easy.We're doing more and more of it, but not necessarily delivering better services because of it."(Interviewee 2) Next to its main GaaP products (Notify, Pay, Verify), Digital Marketplace and leading on digital and technology function across the government, GDS is still in charge of spend controls of most government digital projects, which gives it an important vehicle to ensure its service and capability standards are being followed by other departments and agencies (Interviewee 13).However, there is a sense that GDS is not able to go beyond its initial focus on user needs (providing better transactions) in a significant way: "The real goal is to be changing the whole organisation and make it into a digital one.On that maturity curve, we're still fixing transactions all these years later, eight years later, and you might have expected it to take a while, but there's very few examples anywhere in government of anyone even trying to make their organisation into a digital organisation ..." (Interviewee 12) Table 1 summarises the features of GDS's three-phase evolution.

CONCLUSION AND MAIN LESSONS
In the last few years, GDS has become the standard-bearer for the digital and design profession within the UK government.To use an analogy here, the hackers have become the new mandarins, or at least the mandarins have co-opted some of the capabilities brought in by GDS.Initial new capabilities and ways of working introduced by GDS have increasingly become part of routine skills that many departments and agencies should have in-house, rather than relying on a central agency.In this sense, and somewhat counterintuitively, GDS has become a victim of its own success.It does not function anymore as a charismatic start-up but rather as a professional public digital agency.However, as other government departments have internalised their most valuable capabilities, GDS has found it difficult to develop new cross-government platforms, and its cost-based metrics put it in a defensive rather than a proactive situation.Perhaps unsurprisingly, institutionalisation has put constraints on the original dynamism and agility of the organisation.Initially, the capabilities that GDS brought into government formed around a redefinition of the value proposition (a focus on user needs and the need to diminish failure demand), accompanied by an influx of people who could deliver towards this new notion of value.It is worth noting that many of the people hired by GDS outside the civil service did not come from the private sector, but from the BBC and other similar public or third-sector organisations.In its initial years, GDS was managed by normative control functions, such as hiring like-minded people, that allowed it to grow quickly and scale its activities.The value system underlying GDS was inspired as much by post-war British modernism, with its focus on public space, as by the open web movement and hacking in a positive sense.GDS was a charismatic and agile organisation that was able to jump-start digital transformation through a single website (GOV.UK) and revamp government IT contracts through service frameworks and spending controls.Its market-shaping abilities were formidable.However, the utilitarian approach to user needs (simplicity and authenticity of transactions) also made it susceptible to market-based logics of austerity (savings as the main metric), which, it can be argued, has undermined GDS's ability to build platforms.
GDS now faces the difficult question of what its value proposition is to the government (and users), beyond providing efficiency gains and professionalisation of digital skills.What value proposition can it develop beyond utilitarian notions of user needs (transactions) and failure demand?With the ever-increasing scope of the private platform providers, such as Amazon and Google, that offer hosting and other services to governments and thereby play a prominent role in shaping digital marketplaces, GDS faces questions about its role in the wider landscape of (digital) industrial policy.In other words, GDS needs to find new sources of agility, and purpose and vision, as the context within government machinery and external market actors within which it functions continue to change at an ever-increasing pace.The lessons from the GDS case pertain to both how to build and develop public sector digital agencies, and what the key capabilities and ways of working of such agencies are.First, the GDS started with a clear value proposition: how to significantly enhance the user experience of public digital services.This focus led, second, to the creation of new civil service roles and career tracks in data and digital technologies, which allowed to bring in skills that were at the time relatively rare in government.Third, the new digital skills were deployed agilely to considerably shorten delivery and feedback cycles.Fourth, to ensure that the new skills and focus on user experience diffuse across main government departments, GDS created spend controls of digital procurement across the government.And fifth, to ensure access by SMEs to government digital procurement, GDS created a digital marketplace.
Bringing all these elements of the blueprint together-which is arguably still unsurpassed in the West, at least-the GDS case shows why and how central agile agencies can jump-start digital transformation efforts and why user-centered design skills are essential.Its evolution in the past decade shows why initial focus on radically new ways of working (agile practices and focus on user experience) -at least for the UK context-can bring rapid wins and thus enhance the legitimacy of such agile agencies.Similarly, the drive to professionalise and diffuse the new ways of working and skills across government departments shows how such agile organisations can and indeed need to connect and collaborate with existing policy and service departments.Furthermore, the case demonstrates the importance of having a development pathway that includes strong efforts to professionalise agile and design skills across the civil service.
Simultaneously, the GDS case illustrates centralised government digital agencies' potential limitations and pitfalls.Whilst the charistmatic top-down leadership model of GDS led to major early successes in recruitment, delivery and cost-savings, it also created resistance and resentment amongst government departments subjected to its control.Whereas the private sector inspired agile, delivery first ethos resulted in impressive feats of engineering in the short term, its effectiveness in taking on the more complex organisational reforms and legacy service transformations that are required next to advance government digital transformation is far from clear.Based on GDS's story, we can argue that agile organisations need to evolve into more routine service organisations in the public sector.Such pathways may include political contestation and bureaucratic rivalry, as the case shows, thus such agile organisations in government should plan for a long-term strategy in which, next to the scaling of solutions, also diffusion and professionalisation of practices play key roles.In sum, the long-term effectiveness and sustainability of the GDS model is unclear, and further comparative research on the relative advantages and disadvantages of different digital transformation pathways is urgently needed [Clarke 2019 ].GDS offers a blueprint, but not one that should be followed blindly.

5. 2
Phase Two: Government as a Platform (2016-2018) 5.2.1 A Limited Uptake of an Original Idea.The idea of government as a platform ( GaaP ) has captured the imagination of many governments in recent years, including the UK.Originally conceived by O'Reilly [ 2011 ], GaaP envisages the transformation of the coordination among public administration from closed, structured, and formalised hierarchical relationships into open, flat, and unstructured relationships.A further ambition of the framework is that shared software, data, and services can help to open up public service production processes to actors previously excluded from them [for an overview, see Pope 2019 ].
The case of the UK's GDS: The Professionalisation of a Paradigmatic Public Digital Agency • 28:11