Crossing policy’s ‘valley of death’: tips for bridging the gap between development and delivery

There can often be a daunting space between policy development and delivery in government – a space where policy ideas can be undermined as they move from one team to another. Here are Infrastructure and Projects Authority chief executive Tony Meggs’s top tips for closing it


By Tony Meggs

01 May 2018

Photo:Paul Heartfield

The vast majority of government policies are delivered through the implementation of a project or programme of some description. These projects and programmes span a wide range, from capital intensive infrastructure and military equipment projects through to IT projects and major transformation programmes such as Universal Credit or digital courts reform.  Irrespective of their diversity, they have one thing in common: If the projects are not successfully implemented, then the policy objectives are not delivered.

The common causes of failure in major projects are well rehearsed; they are pretty much the same outside of government as within, and across a wide range of project types:

•         Lack of clarity around project objectives.

•         Lack of alignment amongst stakeholders.

•         Unclear governance and accountability.

•         Insufficient resources, whether human or financial.

•         Inexperienced project leadership.

•         Over-ambitious cost and schedule.

Well-designed projects address all these issues, and more, in the vital project initiation phase. Taking time at the beginning to ensure that objectives are crystal clear, stakeholders are aligned, accountabilities are well-defined, and so on, saves enormous pain and heartache later on. Good project initiation maximises the chance of a successful outcome; poor project initiation is a harbinger of failure further down the line.


Unfortunately, it is in this initiation phase that government projects suffer a unique disadvantage. Good project initiation takes time and time is generally in short supply. This is particularly so when governments feel that they must deliver their agenda within the time limits imposed by the electoral cycle. Once an important policy issue is identified, the business of politics demands a ministerial announcement and all too often, and quite understandably, such announcements come in the form of commitments to very specific outcomes by very specific dates.

It is entirely appropriate for government to make commitments and be held to account, but it certainly helps if those commitments can actually be delivered as promised. If the right initiation work has been done, then announcements can be made with confidence and promises delivered upon.

In the world of research and development, it is well understood that turning a successful piece of research, a laboratory discovery, into a usable product or service is a difficult thing to do. That is why the gap between research and successful commercial development is sometimes known as the "Valley of Death", so hard can it be to cross from one side to the other. In a similar vein, we can think of the space between policy development and policy delivery as another "Valley of Death" because it represents the space in which so many policy initiatives are undermined, sometimes fatally, as they are thrown across the valley from the "policy" team to the "delivery" team. 

I believe the single most powerful thing that we can do to improve the successful delivery of government priorities is to establish a seamless flow and inter-connectivity between policy conception, policy development, and policy delivery to ensure that we apply best practice from the earliest phase of policy development.

What does this mean in practice? Here are my top four suggestions:

  1. Always involve delivery expertise in policy development.

As policy is developed, policy teams should continuously consider implementation issues: how will this policy actually get delivered; how can the policy be adjusted to make it easier to deliver; what will it take in terms of resources; how long will it take; what can we learn from similar developments elsewhere? The most reliable way of ensuring that these issues are addressed up front is by including someone with project or other operational expertise from the very start, as part of the policy development team.

  1. Get an independent assessment of deliverability before announcements and commitments are made.

Even when delivery expertise is involved in policy development, there is always a need for independent expert assessment of project plans. This is to counter the inevitable optimism bias that affects any project in its early phases. The IPA can help here, as can departmental assurance teams, peer review teams from other departments, or external experts. The important thing is to calibrate each project or programme against real world experience from outside the project.

  1. Tailor announcements according to the degree of delivery confidence.

Often it is not the announcement itself that proves problematic, but the unnecessary degree of specificity which creates a trap for those who have to make it happen. In particular, dates should be avoided wherever possible as they are frequently hostages to fortune for complex change projects. But it is also important to avoid being overly specific about delivery methods; this allows time and space for proper option analysis to occur as part of the project initiation process. As the project is matured through the appraisal and option selection process, then more detailed and specific announcements can be made.

  1. Understand and embrace uncertainty.

Frequently in government we try to define quite specific outcomes despite having very uncertain circumstances and highly imperfect information. This, especially when combined with optimism bias, can lead to disappointment all around. Policy initiatives, and their associated implementation programmes, often make very broad assumptions as to how the public will react to a policy when it is implemented, and therefore what benefits will be delivered. Unfortunately, the large uncertainty range associated with those assumptions is frequently forgotten as business cases are written and approved, and the benefits banked in advance. There is a strong case for understanding uncertainty in a more systematic and quantitative manner and trying to define it. For example, if the level of uncertainty is large enough to jeopardise the project, could a pilot project be undertaken to narrow the range of uncertainty to within acceptable limits? 

These suggestions are all in service of one objective: to increase the probability of successful policy implementation by making delivery planning an integral part of the policy making process.

The more we can do to operate in a seamless and integrated way from the very beginning, the closer we will come to bridging the Valley of Death and improving the probability of successful delivery of government policy.

For more information about the IPA or to find out about how the IPA can support the early development of your policies or projects, please visit or email

A version of this article appeared in Civil Service Quarterly

Share this page