DWP Synergy – Absence & Expenses alpha reassessment
DWP's Synergy – Absence & Expenses alpha reassessment
Synergy – Absence & Expenses
| Assessment date | 28/07/2026 |
| Assessment stage | Alpha |
| Assessment type | Reassessment |
| Service provider | Department for Work and Pensions (DWP), on behalf of the Synergy programme (cross-government shared services) |
| Result | Amber |
Previous assessment reports
Published internally in DWP (intranet)[B(1] [DH2]
Service description
Synergy is a shared services platform that helps 250,000+ colleagues across diverse roles and working environments in DWP, DEFRA, MoJ, Home Office, and Arms-Length Bodies (ALBs) complete core HR and finance tasks through a single Oracle SaaS service, with converged self-service journeys and clear support.
This service aims to solve the problems of convoluted user processes, fragmented quality of service, poor performance, and data oversight by creating a simpler, consistent, and converged cross-departmental service that delivers a better experience for all users, greater efficiency and value for money and standardised processes and data.
Service users
Synergy supports colleagues across Departments and ALBs who need to complete core HR and finance tasks, including:
- employees and line managers completing day-to-day tasks (eg claim expenses, record and manage absence, approvals)
- employees who travel regularly for work, including field sites and remote locations
- employees with atypical working patterns (for example, shift, flexi, compressed hours)
- line managers with large and diverse teams, often with higher variability in roles and scenarios
- colleagues working in controlled environments (for example, prisons and other restricted settings)
- ALB users operating under different local policies and processes (for example, DEFRA ALBs)
- recent joiners who are new to departmental policies and processes.
- specialist and corporate users (or example, Finance, HR, operational support teams)
- users who report a disability or rely on assistive technology
Things the service team has done well:
Since the previous assessment, the service team has made clear progress in addressing the panel’s concerns. They demonstrated a stronger understanding of user needs and provided more evidence of how research, testing and feedback are informing design and delivery decisions.
The service team clarified their main success measures and demonstrated capability to start to learn from data and analytics to improve the service and user experience throughout its development.
1. Understand users and their needs
Decision
The service was rated green for point 1 of the Standard.
Optional advice to help the service team continually improve the service:
User needs and hypotheses
- design hypotheses were clearly stated and iterated based on research evidence. The team spoke about how user needs have been prioritised and evidenced how user needs were part of team planning. They explained how hypotheses are tested, designs iterated and evidence fed into teams for development. However, the team should rework their user needs to reflect the way in which their users would phrase things. E.g. “I need to book my leave so I can go on holiday” as some of the wording in the user needs was quite formal and I am unsure if this is how users would think about what they need to do. The user needs appeared to be focused on the users themselves and were not business or policy needs which was good to see
- the team presented some AI functionality, for example how users could search for guidance using a chat bot. Some user needs and usability testing activity was communicated regarding this in the assessment. The team should ensure that AI functionality is the right way to meet the needs of users, rather than a way to ‘plaster over’ poor guidance, and we would encourage more detailed analysis to support the use of AI in beta
End to end service
- it was good to hear that the team had tested some emails as triggers for initial actions related to the service. However, it would have been even better to have seen more detail on understanding the ‘as is’ user journeys, what was changing for users with regards to these whole journeys and what else had been (or is planned to be) tested
- it was also unclear if the line manager dashboard had been usability tested (for example) or if other screens outside of the happy paths for annual/sick leave requests or expenses claims had been usability tested. All these screens and journeys should be usability tested and iterated before the service is released to private beta users
Real world environments
- user research done with DEFRA and DWP evidenced the difference between users’ needs and contexts. Device testing was also completed to understand how devices changed behaviour and preferences related to this. This was good to hear
- despite a demo not covering these details, it was clear the team understood how design needed to adapt and accommodate these changes across departments. In addition to this, the team clearly communicated how their understanding of the problem and the user needs had led to design changes and plans for measuring success
- the research had not covered real world environments aside from an office setting, but it did appear that the team had gathered evidence about these environments and user needs within these, and that evidence had influenced team decisions. Testing in these real-world environments should be a high priority for private beta
Influencing design
- the team evidenced how they had worked with Oracle to push for changes to the design, this was good to hear. However, there were still some constraints in the system, such as [line manager only] text on field on the sickness absence request for an employee to fill out. The team had made some changes, but it would have been even better to have fixed the underlying issue to fully meet users’ needs and provide a simple service. The team had clearly put a lot of work into changing the core Oracle system to improve the accessibility of the product which was reassuring
- the team should continue to advocate for their users and ensure the service that is delivered is designed with users in mind and meets their needs, and challenging any constraints uncovered
2. Solve a whole problem for users
Decision
The service was rated amber for point 2 of the Standard.
This is amber because:
- the team demonstrated significant progress in understanding user needs across the wider absence and expenses journeys and is addressing user problems beyond the core transaction itself
- several AI-enabled and AI-supported capabilities were presented as part of a wider service response to help address the whole user problem, and the AI capability presented is emerging and exploratory. Evidence presented was largely based on prototypes, planned functionality and future testing approaches, with limited evidence of outcome validation in operational conditions or assessed against longer-term service outcomes. Further analysis and testing are required to understand their impact, effectiveness, risks and contribution to solving the whole user problem
- AI introduces distinct user journeys, error states and failure modes, the team should consider emerging human-centred AI (HCAI) principles and service patterns. This should include consideration of user autonomy, transparency, confidence indicators, feedback loops, correction mechanisms, escalation paths and recovery journeys when AI-assisted interactions do not produce the expected outcome
- AI introduces new service risk scenarios, auditing requirements and accountability considerations, including low-confidence outputs, harmful or incorrect responses, bias, inaccuracy, accessibility impacts, fluctuating user trust and ownership of outcomes, the team should ensure appropriate expertise to support ongoing assessment of risk, governance, assurance, accountability and long-term sustainability of AI-enabled and AI-supported service components within the GDS framework
-
the team should continue to review and apply current GDS guidance relating to AI-enabled services, including the current guidance that services should be prepared to stop using AI technology if needed (see Technology: Using artificial intelligence (AI) in services).
- the team should continue aligning its approach with evolving GDS guidance on AI-enabled services. Consideration should also be given to the implications of working within commercial off-the-shelf (COTS) AI capabilities and the degree to which user needs can be addressed when functionality is predetermined by platform providers. The team demonstrated clear mechanisms for identifying, managing and escalating platform limitations that affect users. However, as the ability to solve some whole problem user needs remain constrained by Oracle platform limitations, user outcomes may continue to be influenced by Oracle delivery decisions and release cycles as well as local service design decisions.
3. Provide a joined-up experience across all channels
Decision
The service was rated amber for point 3 of the Standard.
This is amber because:
-
the team demonstrated a significantly more joined-up view of user journeys across policy guidance, support channels, transactional services and human support than was evidenced at the previous assessment, including consideration of how users move between policy guidance, support channels, Oracle transactions, managers and service support functions. The assessment showed an understanding that users experience absence and expenses journeys across multiple channels rather than through a single transactional interface
-
the team provided limited evidence demonstrating why AI-enabled chat was the most effective solution to the identified user need compared with alternative approaches such as improved navigation, search or contextual guidance. While user testing explored the AI chat concept, there was limited evidence of how AI-supported interactions would operate across channels in a live service environment, including hand-offs to other support mechanisms and the impact on user and service outcomes. Further evidence is required to demonstrate that the proposed channel integration delivers measurable improvements for users
-
the service team and future assessment panels would benefit from clearer evidence demonstrating a comparison of AI chat with non-AI solutions against clear user and service outcomes, including successful policy discovery, answer accuracy, task completion and reduced need for further support. Further evidence may be required during beta to demonstrate that AI is the most effective, inclusive and proportionate solution to the identified user need
-
the evidence presented in the assessment primarily focused on isolated journeys rather than full end to end AI enabled ecosystem journeys. The team should continue to validate end-to-end journeys in functioning service environments, including AI-enabled interactions, to better understand how users move between policy guidance, support channels, transactional services and human assistance. This should include end-to-end accessibility testing and testing of interactions between AI-supported channels and associated support services
-
the team should test the integrated AI feature in the configured service with a representative range of users, including assistive technology users, and demonstrate safe cross-channel hand-offs to authoritative guidance and human support when the AI cannot answer reliably.
-
as AI is an evolving area of government service delivery, the GDS Service Manual, and Service Standard, the team should ensure it has sufficient access to AI expertise appropriate to the service context, to support the design, and evaluation of AI-enabled service journeys and outcomes. This expertise should help assess effectiveness, accessibility, accountability, risk, user trust and the impact of AI on end-to-end and cross-channel service outcomes
4. Make the service simple to use
Decision
The service was rated amber for point 4 of the Standard.
This is amber because:
- the team demonstrated a strong test-and-learn approach, with clear examples of user research informing iterative improvements to journeys, content and interactions
-
the team provided limited evidence of the effectiveness and timeliness with which validated insights are implemented and subsequently evaluated within the configured Oracle environment. As a result, while the team showed a mature approach to identifying usability issues, there is less evidence that improvements have been consistently validated within a working service environment
-
the service team and future assessment panels would benefit from clearer evidence demonstrating traceability from user insight through to design decision, implementation (or mitigation where platform constraints apply), and subsequent validation within the configured service environment
5. Make sure everyone can use the service
Decision
The service was rated amber for point 5 of the Standard.
This is amber because:
-
the service team has made substantial progress since the previous assessment and demonstrated a significantly stronger understanding of accessibility and inclusion. Accessibility considerations are now influencing user research, design decisions and platform change discussions, and the team provided evidence of engagement with users with a range of accessibility needs
-
the evidence presented focused primarily on accessibility improvements identified and tested through Figma prototypes. The team acknowledged that assistive technologies could not be fully used within those testing environments, limiting confidence that findings and improvements will translate into the configured Oracle service. While accessibility issues have been identified and mechanisms exist for tracking and escalating platform-level issues, some improvements remain dependent on Oracle delivery processes and release cycles. As a result, further evidence is required to demonstrate accessibility outcomes within a working service environment and to provide confidence that identified barriers are being effectively remediated over time
-
the team should prioritise the monitoring and remediation of accessibility issues within configured Oracle environments. Future alpha assessments would benefit from evidence of accessibility backlog size, remediation progress, numbers and types of defects resolved, known constraints and the effectiveness of mitigation measures where platform limitations prevent full remediation
-
the team should undertake end-to-end accessibility validation in a configured Oracle environment using assistive technologies and a diverse range of users to provide confidence that accessibility improvements demonstrated in prototypes are realised in the working service and can be validated during Beta
6. Have a multidisciplinary team
Decision
The service was rated green for point 6 of the Standard.
Optional advice to help the service team continually improve the service:
- the team have several embedded content designers, which addresses the previous concern on this standard. Given the use of artificial intelligence (AI) in the service, going forward the team should include, or have readily available to them, someone who fully understands how these capabilities work and the impact of using them
7. Use agile ways of working
Decision
The service was rated amber for point 7 of the Standard.
This is amber because:
-
the team showed several examples of user research informing design decisions and prototyping. They also demonstrated a clear route from user needs to backlog priorities. While agile delivery is still being constrained by the Oracle service request resolution process, the team showed how decisions are prioritised, governed and progressed through established change routes
-
the team brought additional AI-enabled capabilities, including Oracle Guided Learning, automated workflows and chatbot functionality into scope. These capabilities were not reviewed in sufficient depth during this reassessment. It was not clear how the assumptions and risks associated with AI would be quantified, tested with users and improved. It was also unclear what governance and guard rails are in place to support good data protection, ethics and security practices. The panel recommends a workshop with DWP AI specialists within the next 3 months. This should help the team develop a clearer approach to testing, governance and risk management against all standards before the planned Alpha assessment for public-facing services and before entering Beta
8. Iterate and improve frequently
Decision
The service was rated amber for point 8 of the Standard.
This is amber because:
-
the team showed evidence that user research and testing are informing documented user needs and service improvements. It was good to see examples of changes made in response to user feedback, including in changes to the Oracle Guided Learning widget. When initial changes did not improve user experience as expected, the team carried out further testing and identified the need for additional changes
-
the evidence for testing and iteration across different environments and with different user groups was less developed. While the team described a range of users and contexts, constraints on access and testing with these groups meant there were fewer examples of how different user needs had informed changes to the service. Testing with a wider range of users and operational contexts should be a priority for the team in private Beta
-
steps have been taken to facilitate frequent internal prototyping and iteration, which is good to see. It was still unclear though how quickly these prototypes are released into a live environment, nor how easily iteration is possible if many significant changes require Oracle change requests. Some requests raised with Oracle over 6 months ago are still not live in a production instance, so frequent iteration of the service is constrained by the Oracle quarterly release cadence. This is a risk that the team needs to mitigate as they broaden their testing and feedback strategy to support more varied user needs and contexts
9. Create a secure service which protects users’ privacy
Assessed by: Tech assessor with performance analyst input when relevant
Decision
The service was rated green point 9 of the Standard.
10. Define what success looks like and publish performance data
Decision
The service was rated green for point 10 of the Standard.
Optional advice to help the service team continually improve the service:
- continue to build a full performance framework, overseen by a performance analyst to clearly define success and failure, connecting business objectives, user needs and the most meaningful measures
-
establish baselines and benchmarks to give richer context to analytics and work towards testing hypotheses statistically, especially to inform iterations and feature development
-
demonstrate how you’ve iterated using data at beta assessment
-
consider how metrics will need to be used and be interoperable across the whole service, so unify your data strategy and approach programme-wide
-
document all methodology, insights and what you learn, especially important for failure, to continuously improve and avoid others repeating any mistakes – vital part of knowledge transfer, continuity and maintaining standards
11. Choose the right tools and technology
Decision
The service was rated amber for point 11 of the Standard.
This is amber because:
- the Payment Gateway integration within Oracle Fusion does not satisfy the required level of compliance for the service. The service roadmap needs to show how this critical functionality will be sourced
12. Make new source code open
Decision
The service was rated green for point 12 of the Standard.
Optional advice to help the service team continually improve the service:
- while it is not possible to make any of the source code open due to proprietary and business model restrictions, the team should continue to be transparent about how the platform is being used and configured.
13. Use and contribute to open standards, common components and patterns
Decision
The service was rated amber for point 13 of the Standard.
This is amber because:
- the team did not show how patterns will remain consistent when Oracle updates, or how design decisions will be governed across four departments. For future assessments, they’d be expected to develop and explain how they would ensure shared design governance and show how this would work in practice to keep things consistent and aligned to developing research and user needs
14. Operate a reliable service
Decision
The service was rated green for point 14 of the Standard.
Next Steps
This service can now move into a private beta phase, subject to addressing the amber points within three months time and GDS spend approval.
To get the service ready to launch on GOV.UK the team needs to:
- get a GOV.UK service domain name
- work with the GOV.UK content team on any changes required to GOV.UK content