Written evidence to the House of Commons Public Accounts Committee inquiry on Shared services
1) One-page briefing
● Hleb Buziuk writes in a personal capacity as an independent public policy and governance researcher.
● Three key messages:
○ Since 2022, Cabinet Office has improved governance and reporting, but the published evidence still does not show decision-grade oversight of the portfolio [1][4][5].
○ Cluster delivery is moving, but the main risk now sits in onboarding, sequencing and legacy coexistence, especially where departments or ALBs do not yet have firm dates [1][6][7].
○ Interoperability is now the main value-for-money test, because shared services depends on alignment with functions and adjacent programmes, not on cluster procurement alone [1][3].
● Key findings:
○ In October 2025, 31 of 64 key interdependency datapoints were red and 21 of 64 were missing. That means 81.3% (n=64) of the central dashboard was either incomplete or already signalling material concern [1]. I infer that central reporting improved faster than usable portfolio control [1][4].
○ In Matrix, as at November 2025, five of seven departments had a firm commitment and timeline to onboard, but only 15 of 92 ALBs did so. That is 71.4% for departments (n=7) and 16.3% for ALBs (n=92) [1]. I infer that live schedule risk now sits at the adoption boundary, not only in programme build [1][6].
○ The NAO has reported that the programme originally aimed for departments to be on cloud-based systems by 2028, but its 2026 update records planned departmental onboarding, as at November 2025, extending into 2029 [1][2].
○ Design principles require ‘adopt not adapt’ and interoperability, yet in the NAO’s published ERP extract only two of seven listed functions are common across all five clusters, or 28.6% (n=7) [1][3].
● Key recommendations:
○ By Q4 2026, Cabinet Office should provide a quarterly Portfolio Assurance Pack covering milestones, spend, benefits, onboarding status and interdependencies. How to check: complete raw returns against standard fields.
○ By end 2026, Cabinet Office and cluster SROs should provide dated onboarding plans for each remaining department and ALB. How to check: signed schedules linked to legacy contract end dates.
○ From the next approval round, HM Treasury and Cabinet Office should require an interoperability statement, a design-waiver log and a de-duplicated benefits register. How to check: approval letters and quarterly updates.
○ By autumn 2026, Cabinet Office should designate one DG-level owner for cross-cluster interdependencies. How to check: published terms of reference and a decision log.
● What is new versus official sources:
○ converts broad oversight concerns into reproducible dashboard metrics;
○ separates cluster build from adoption readiness;
○ turns interoperability into observable tests on modules, waivers and interfaces;
○ treats benefit overlap as a portfolio control issue.
● Methods: origin-first public sources only. See section 3, Annex A, section 7 and section 9.
2) About the submitter
I am an independent public policy and governance researcher writing in a personal capacity. I am not submitting on behalf of any department, supplier, cluster or representative body.
3) Methods, sources and limitations
This submission is based on origin or near-origin public sources, led by the National Audit Office, government design documents, HM Treasury responses, a live departmental estimate memorandum and an Accounting Officer assessment [1]-[7]. Committee reports are treated as secondary where an origin source exists.
The synthesis insights in sections 1 and 5 are each anchored in at least two visible sources and are summarised again in Annex A. Same-issuer government responses are used only for what government said it would do, such as developing metrics, a benefits framework or an updated case for change [4][5].
The main headline point that is secondary-only in the current reference set is the earlier 2028 target year. In this draft, that point relies on later NAO descriptions, not the original strategy text, which is not in the current reference set [1][2]. It is therefore softened, marked in Annex A, and paired with a data request in section 7 for the approved milestone baseline and any formal change log.
Where impact is inferred rather than directly shown, I say so and propose a concrete test in section 7.
4) Context and background
Shared services is intended to standardise core corporate services through five clusters, with common platforms, processes and data so government can improve user experience, efficiency and interoperability [2][3]. The 2022 NAO report found major delivery barriers, including weak governance, funding uncertainty and the absence of an overarching business case [2]. The current inquiry asks whether the centre now understands progress, whether clusters are on track, and whether interdependencies can be managed.
5) Evidence mapped to the Committee’s questions
5.1 How well does Cabinet Office understand progress since 2022?
Cabinet Office understands more than it did in 2022, but the published evidence does not yet show that it can manage the portfolio by exception with confidence. The NAO records revised governance, including a Service and Technical Design Authority, but also reports no single point of ownership and weak escalation across connected boards [1]. It also reports that, in October 2025, 31 of 64 key interdependency datapoints were red and 21 were missing [1]. Government had earlier told Parliament that new portfolio metrics and a benefits framework were being developed, and by May 2024 it said the updated case for change had been implemented [4][5].
Evidence-based observations:
● governance is more structured than in 2022, but ownership and accountability still fragment across boards [1];
● the dashboard metric suggests oversight remains incomplete: 81.3% of datapoints were red or missing (n=64) [1];
● the cited public sources do not yet provide a complete quarterly return that Parliament can trend.
Added value: this compares government’s claimed fixes [4][5] with later audit evidence on whether the resulting information was actually usable [1].
Three points:
● Cabinet Office has more reporting than in 2022, but not yet decision-grade control [1][4][5].
● A dashboard with 52 of 64 datapoints red or missing is a warning indicator, not assurance [1].
● The issue is now less whether the centre has a forum, and more whether it has a complete, comparable return [1].
Counterpoint: central reporting can create burden and false precision. The answer is a thin, standard return tied to clear ownership.
Options:
● Option A: a private quarterly return to the Committee.
● Option B: a standard Portfolio Assurance Pack, public where possible, with a confidential annex if needed.
Preferred option: B, because it improves both accountability and internal management.
5.2 Are clusters on track to deliver the strategy to time and budget, including longer-term improvements?
The evidence suggests uneven progress rather than uniform failure. The NAO says all five clusters have made progress on business cases, procurement and delivery plans [1]. The NAO has also reported that the programme originally aimed for departments to be on cloud-based systems by 2028, while its 2026 update records planned departmental onboarding, as at November 2025, extending into 2029 [1][2]. Programme-level illustrations from Matrix and Synergy suggest that the main live risk now sits in onboarding certainty, integrated planning and legacy coexistence. In Matrix, as at November 2025, five of seven departments had a firm commitment and timeline to onboard, but only 15 of 92 ALBs did so [1]. DSIT’s 2025-26 memorandum records Matrix as Amber, with a whole-life baseline cost of £938 million and an end date of 30 November 2027 [6]. Synergy’s Accounting Officer assessment likewise supports proceeding, but only subject to a fully costed integrated plan and contract extensions through to 2028 for service continuity [7].
Evidence-based observations:
● cluster plans are more mature than in 2022 [1];
● onboarding risk is sharper for ALBs than for core departments in the published Matrix position as at November 2025 [1];
● some clusters still rely on legacy extensions or unresolved future sequencing [1][7].
Added value: this separates programme build from adoption readiness and tests whether delivery dates are supported by signed organisational commitments.
Three points:
● The delivery question is no longer only ‘can the cluster build’, but ‘who has actually signed up to move, and when?’ [1][6].
● A portfolio can be funded on paper and still not be time-controlled if onboarding dates are missing [1].
● Longer-term benefits depend on departments and ALBs crossing the final adoption boundary [1][7].
Counterpoint: not every delay is failure. Some phasing may be rational where departments have modern systems or where value for money is genuinely contested.
Options:
● Option A: ask clusters to provide narrative updates.
● Option B: require a dated onboarding plan, by organisation, with dependencies, funding status and legacy contract end dates.
Preferred option: B, because it makes schedule risk auditable.
5.3 Can government manage interdependencies and interoperability?
This is the sharpest unresolved risk. The design principles say shared service centres and department services should use open technology, consistent processes and data standards, and should ‘adopt not adapt’ to enable convergence and interoperability [3]. Yet the NAO reports inconsistent ERP configuration, weak functional convergence and the absence of a strong technical lead [1]. In its published ERP extract, only Learning and Expenses are common across all five clusters, which means universal functionality coverage is 28.6% (n=7 listed functions) [1]. The NAO also reports that at least 25 other digital change programmes interact with shared services, of which 14 had unknown impact in November 2025 [1]. Clusters further raised concerns that overlapping programmes could count the same benefits more than once, and the ATS reset shows how adjacent programmes can add cost and delay to cluster delivery [1]. I infer that interoperability remains the main value-for-money test because the published evidence does not yet show a single body with clear authority to compel alignment across clusters, functions and linked programmes [1][3].
Evidence-based observations:
● 14 of 25 identified programmes had unknown impact on shared services, or 56.0% (n=25) [1];
● functional convergence is uneven: HR did not implement its revised standard in time and Commercial opted out apart from ‘procure to pay’ [1];
● ATS was reset in October 2025 after cluster concerns, with estimated reset costs of £26 million to £38 million, and no published Accounting Officer assessment as of December 2025 [1].
Added value: this turns interoperability from a general aspiration into tests on waivers, interfaces, benefit overlap and cross-programme ownership.
Three points:
● Interoperability is not a technical extra. It is the condition for portfolio value for money [1][3].
● Where impact is unknown for most linked programmes, interdependency management is not yet mature [1].
● Without a design-waiver and interface view, Parliament cannot judge whether divergence is temporary, justified or costly [1][3].
Counterpoint: some variation is justified, especially where security, legal or sector requirements differ. The issue is not zero divergence, but governed divergence with clear cost and expiry.
Options:
● Option A: strengthen SaTDA within the current structure.
● Option B: create a DG-level cross-government board with authority over interdependency escalation, design waivers and benefit overlap.
Preferred option: B, because the current issue is cross-portfolio authority, not only technical advice.
6) Recommendations
Quick wins
Committee levers
Delivery actions
Structural
Committee levers
Delivery actions
7) Quant-lite results and minimum data request pack
Derived metrics
● Interdependency visibility failure rate: 81.3% (n=64). Inputs: 31 red plus 21 missing datapoints from a total of 64 [1]. Steps: add red and missing, then divide by the total. This informs whether the centre has a usable view of cross-programme risk. Caveat: red and missing are different conditions, but both weaken oversight.
● Matrix firm-onboarding coverage, as at November 2025: departments 71.4% (n=7) and ALBs 16.3% (n=92) [1]. Steps: divide firm commitment and timeline by the total in scope. This informs whether schedule risk sits in build or in adoption. Caveat: this is a dated snapshot and scope may change.
● Universal ERP functionality coverage in the NAO extract: 28.6% (n=7 functions) [1]. Steps: count functions present across all five clusters, then divide by the total functions shown. This informs whether convergence is advanced enough to support interoperability. Caveat: the NAO says the extract is illustrative, not exhaustive [1].
Minimum data request pack
● Raw interdependency return: datapoint, owner, RAG, missing-data reason, remedial date. Owner: Cabinet Office. Cadence: quarterly. Enables the Committee to test whether oversight is complete.
● Current scope and milestone baseline: organisations in scope, target dates, approved revisions, change-log reason. Owner: Cabinet Office. Cadence: quarterly. Enables the Committee to test whether the timetable has formally changed.
● Onboarding schedule: total organisations in scope, signed onboarding date, blocker, funding status, legacy contract end date. Owner: cluster SROs. Cadence: quarterly. Enables the Committee to test whether schedule risk is auditable.
● Design-waiver and interface log: waiver, reason, approving authority, expiry, linked interface, estimated cost. Owner: Cabinet Office and cluster design authorities. Cadence: quarterly. Enables the Committee to test whether interoperability is governed.
● Benefits register: benefit ID, owner, baseline, counterfactual, overlap flag, net attribution. Owner: Cabinet Office and HM Treasury. Cadence: quarterly. Enables the Committee to test whether value for money is being double counted.
8) Impact, performance and value-for-money artefact
A minimal committee scorecard, using the metrics and fields in section 7, could track:
● completeness of the interdependency return;
● share of organisations with signed onboarding dates;
● number of open design waivers and unresolved interfaces;
● share of linked programmes with known impact;
● gross benefits and overlap-adjusted net benefits.
A simple pass condition would be: completeness improving, onboarding certainty rising, unknown impacts falling, open waivers reducing or expiring on time, and net benefits reported after de-duplication.
9) Committee may wish to ask
Cabinet Office
● What exact quarterly return tells you whether the strategy is on track to time, budget and interoperability?
● Who has authority to resolve a conflict between a cluster, a function and another central programme?
● What formal change log records any revision to the original timetable?
HM Treasury
● What evidence do you now require to show that shared-services benefits are not also being claimed elsewhere?
● When departments or ALBs do not sign onboarding plans, what happens to the value-for-money case?
Cluster leads
● Which remaining organisations still lack a firm onboarding date, and what is blocking each one?
10) Risks, further work, references, annex list and annexes
Risks and further work
● some denominators are dated snapshots rather than live rolling measures;
● the ERP functionality figure is an extract and should be treated as an indicator;
● government response documents are same-issuer statements of intent, not independent outcome evidence;
● the earlier 2028 target is secondary-only in the current reference set and should be checked against the approved milestone baseline.
References
[1] National Audit Office, Update on government shared services, 6 March 2026, https://www.nao.org.uk/wp-content/uploads/2026/03/update-on-government-shared-services.pdf
[2] National Audit Office, Government shared services, 30 November 2022, https://www.nao.org.uk/wp-content/uploads/2022/11/government-shared-services.pdf
[3] Cabinet Office and Government People Group, Design Principles: Summary, updated 18 August 2023, https://www.gov.uk/government/publications/a-shared-services-strategy-for-government/design-principles-summary
[4] HM Treasury, Treasury Minutes, CP 902, July 2023, https://assets.publishing.service.gov.uk/media/64b1676348826b000d3a9ea0/E02946343_HMT_Treasury_Minutes_CP_902_v01_PRINT.pdf
[5] HM Treasury, Treasury Minutes Progress Report, CP 1102, 10 May 2024, https://assets.publishing.service.gov.uk/media/6656dd027b792ffff71a847f/E03139041_CP_1102_Treasury_Minutes_Progress_Report_May_24_Accessible.pdf
[6] Department for Science, Innovation and Technology, Supplementary Estimates memorandum 2025 to 2026, 10 February 2026, https://www.gov.uk/government/publications/dsit-supplementary-estimates-memorandum-2025-to-2026/dsit-supplementary-estimates-memorandum-2025-to-2026
[7] Department for Work and Pensions, Synergy Programme Accounting Officer Assessment, 8 May 2025, https://www.gov.uk/government/publications/dwp-accounting-officer-assessments-for-the-government-major-projects-portfolio/synergy-programme-accounting-officer-assessment-may-2025
Annex list
● Annex A: Evidence Ledger.
Annex A: Evidence Ledger
● Claim: Cabinet Office has better governance than in 2022 but not yet decision-grade oversight. Claim type: SYNTHESIS. Source A: [1]. Source B: [4][5]. Source B role: Same issuer. Status: INPUTS-ANCHORED. Confidence: High, because the later audit tests the earlier stated fixes. What would change my mind: complete quarterly returns with trend. Verification owner: Cabinet Office and NAO.
● Claim: Delivery risk now sits chiefly in onboarding and adoption rather than procurement alone. Claim type: SYNTHESIS. Source A: [1]. Source B: [6][7]. Source B role: Independent corroboration at programme level. Status: INPUTS-ANCHORED. Confidence: Medium. What would change my mind: signed plans for remaining organisations. Verification owner: Cabinet Office and cluster SROs.
● Claim: The timetable now appears to extend beyond the earlier 2028 target year. Claim type: RESTATEMENT. Source A: [1]. Source B: [2]. Source B role: Context only. Status: SECONDARY-ONLY. Confidence: Medium, because the current reference set does not include the original strategy text. What would change my mind: approved milestone baseline and change log. Verification owner: Cabinet Office.
● Claim: Interoperability is the main value-for-money control risk. Claim type: SYNTHESIS. Source A: [1]. Source B: [3]. Source B role: Context only. Status: INPUTS-ANCHORED. Confidence: Medium, because the control gap is clear but the full cost effect is not yet published. What would change my mind: waiver log and complete impact mapping. Verification owner: Cabinet Office and HM Treasury.