Skip to content

How do public sector innovation units work in practice? 

11 April 2018

News and media


Front cover of the Mapping public sector innovation units report

Public sector innovation (PSI) units have spread rapidly across government departments in Australia and New Zealand in recent years, and an ANZSOG-funded research project is providing the first analysis of their structures and operations.

The Mapping Public Sector Innovation Units in Australia and New Zealand report found that the emerging landscape of PSI units in Australia and New Zealand was yet to be fully documented, with little understanding of the areas they were working in, or the distinct capabilities and approaches that they were using. The report identified at least 26 PSI units based in various levels of government in Australia and New Zealand alone, along with many examples of independent innovation units that frequently work with governments on policy and public sector reform

The report features results from a survey, conducted this year by Dr Michael McGann, Professor Jenny M Lewis, and Dr Emma Blomkamp, from Melbourne University’s Policy Lab, which show that while there is a wide range PSI structures:

  • they are generally small
  • reliant on contractors and consultants
  • located within a single department
  • focused on the early stages of the policy cycle.

The survey results revealed that, internationally, the rise of PSI units has been framed as a response by governments to increasingly complex challenges, particularly in the field of social policy. In Australia and New Zealand, PSIs are focused on areas of social policy and services, rather than areas such as energy and taxation.

Only four PSI units reported that their establishment was an initiative of an elected official or member of government. This suggests that the emergence of PSI units within government in Australia and New Zealand is being driven by public managers and administrators, rather than by politicians or elected officials.

The majority of PSIs employ five or fewer people, the outcome being that many have to regularly draw in external expertise to carry out work. Government-based PSI units tend to recruit primarily from within the public sector. While these PSI units tend to recruit at least some staff from outside government, including community-sector organisations and design agencies, only about a third report recruiting ‘many’ of their staff from outside government.

PSIs have a range of remits, from policy development and reform; involving, identifying or scoping problems; consulting with stakeholders; scaling and spreading new approaches; supporting and developing partnerships; developing policy proposals and reforms, and working on systemic change.

The data suggests that PSI units and teams are very heavily involved at what might be termed the earlier stages of the policy cycle – through identifying problems, generating ideas and piloting solutions. Conversely, far fewer units reported that they were involved at the evaluative and scaling end of the policy cycle. For example, the proportion of PSI units who reported that they ‘very frequently’ worked on ‘evaluating programs/trials/pilots’ was just under 30 per cent.

Surprisingly, the data suggests that one of the gold standards of evidence-based policymaking, randomised control trials (RCTs), is seldom used by PSI units. Only five PSI units reported using RCTs with any degree of frequency, although more than 17 units reported drawing on ‘behavioural insights’—a methodology closely associated with RCTs— either ‘quite’ or ‘very frequently’ in their work.  This suggests that PSI units may be relying more on the findings of research conducted by other organisations and applying these in practice than initiating experimental research of their own.

PSIs also had a strong tendency to use human-centred design – associated with the use of ’interviews and/or empathy conversations’, ‘focus groups’, ‘ethnographic methods’, ‘user testing/prototyping’ and ‘systems thinking or mapping’ – alongside other approaches.

The survey raised the concern that there were possible discrepancies in the methods used by PSIs, such as using behavioural insights without randomised control trials, scoring high in ‘evaluation and systems improvement’ but not in ‘evidence-based methods’, or favouring human-centred design methods without much contact with citizens or community groups.

The survey represents one phase in a broader research project that the Policy Lab is conducting supported by an ANZSOG research grant. The Policy Lab’s research team will continue this project and build on the survey results by carrying out five case studies of PSI units working on various policy and innovation domains at different levels of government. The case studies will provide richer insights into the governance and operations of the selected PSI units, their relationships with other actors and institutions, and the methodological approaches they apply. The results of this research will be available towards the end of 2018 and a public summary will be shared on the project web page: http://go.unimelb.edu.au/ix86.