Audit a Green SLA Dashboard Before Accepting It

Why this matters

An SLA dashboard is a reporting instrument, not a certificate of operational truth. A green result may be correct under the agreed formula while still failing to reveal repeated disruption at one location, temporarily restored assets or unresolved complaints. The control problem is therefore not whether the arithmetic is wrong. It is whether the evidence behind the number is sufficient for the decision leaders are being asked to make.

Evidence context

The UK Cabinet Office’s Contract Management Playbook, published 25 March 2026, is government practice guidance. It supports proportionate governance, responsibility mapping, obligation tracking, accurate records, multi-source performance information, change control and planned exit. It does not measure the impact of those practices, and its public-sector rules should not be treated as universal law. [1, pp. 45–49,

53–55, 67]

The Australian National Audit Office published Auditor-General Report No. 25 of 2025–26 on 18 March 2026.

It tested 14 engagements in a diplomatic Security Enhancement Program using records, contract data and procurement information. Four achieved the original scope, timeframe and estimated cost together. The sample is a bounded audit case involving mixed works and services, not an FM-sector benchmark. [2,

summary paras. 11, 19; paras. 1.17–1.18]

The Planning Malaysia study published 3 June 2026 analysed PWD FMM 2016 and interviewed six purposively selected practitioners: five FM contractors and one public client. It reports challenges involving tender information, contract clauses, variations, specifications, work orders and asset data. The small contractor-heavy sample is indicative rather than generalisable; unexplained participant codes also limit respondent-level interpretation. [3, pp. 108–114]

Osmos Global analysis

Separate mathematical compliance from operational acceptance. A response-time target can be met even when restoration is temporary or the same fault returns. The practical consequence is that a reported result must be interpreted in the operating context that produced it.

Select exceptions using risk, recurrence and user impact rather than choosing only convenient closed records.

This changes the management question from whether an activity occurred to whether the agreed condition was achieved and can be demonstrated.

Trace each selected result back to the event, work record, asset history, user consequence and final verification. The control should create enough traceability for another competent reviewer to reproduce the decision without relying on memory.

Record disagreements explicitly. A disputed denominator or missing timestamp is itself an assurance finding, not an invitation to improvise a favourable result. Where evidence is incomplete, the honest result is an exception with an owner—not a confident conclusion assembled from assumptions.

Figure 1. The obligation-to-proof model Original Osmos Global conceptual framework, 2026. No measured dataset. Prepared 1 September 2026.

What this means: A reported result becomes decision-ready only when the proof and acceptance route are explicit.

The practical tool

1. Define the decision that the dashboard will support. Write the purpose and boundary first so the record does not become a generic administrative form. 2. Choose a risk-based sample of green, amber and reopened items. Use information already created by operations where it is reliable, and identify any new collection burden explicitly. 3. Reconstruct each item from primary operational records. Give the action to a named role with authority to resolve the exception or escalate it. 4. Compare the reconstructed outcome with the reported status. Test the step on a difficult real example before making it part of routine governance. 5. Assign corrective action and a date for re-verification. Retain the outcome and the evidence used so that later review can distinguish improvement from a change in reporting.

Illustrative application

A portfolio reports strong monthly response performance, yet one business-critical floor experiences three similar cooling interruptions. The sampled tickets show fast attendance and closure, but two were temporary resets and the permanent component replacement remains open. The dashboard calculation may be accurate; accepting it without an exception review would still misstate reliability. This example is an Osmos Global illustration, not a reported case from the cited sources. Its purpose is to test the decision logic and reveal what evidence would be required.

Questions for the review

• Can the parties demonstrate how they will define the decision that the dashboard will support using a current operating record? • Which evidence would cause the acceptance owner to reject the reported result or impose a condition? • Who has authority to resolve the most consequential exception raised by this contract assurance control? • How will the workflow distinguish a provider-controlled gap from a client, asset, landlord or third-party dependency? • What will be rechecked to establish that the corrective action changed the operational outcome rather than only the report?

Portfolio and GCC application

A multi-site dashboard should allow a reader to move from the portfolio result to the building, service and event behind it. For a GCC workplace, that means preserving the affected business function as well as the engineering asset. A short interruption in a training room and the same interruption in an overnight support operation may require different management decisions even when the contractual priority is identical. Do not alter the agreed score retrospectively to express that distinction; show the consequence separately.

The sampling log should include apparently successful cases, not only known failures. Record the selection reason before retrieving the evidence, then retain both confirming and contradictory records. Where a sample deliberately targets recurrence, its findings cannot estimate the error rate of the whole portfolio. The next review should ask whether the specific weakness was corrected, rather than claim that a smaller number of exceptions proves general improvement.

A useful challenge is to ask a reviewer who did not prepare the dashboard to reconstruct one selected result.

If that person needs a private spreadsheet, an undocumented exclusion or the account manager’s memory, the reporting process has a dependency worth correcting. This test examines reproducibility; it does not presume dishonesty or justify withholding payment without contractual authority.

Implementation-quality indicators

• Selected records can be reproduced from source evidence without reconstructing the story after the review starts. • Exceptions carry a named owner, decision route, due date and explicit consequence of delay. • Temporary controls and permanent resolution remain distinguishable in both the operating record and the governance pack. • A later reviewer can see which baseline and definitions applied when the decision was made.

Leadership implications

FM leaders should own the operational meaning of the indicator, not only its calculation. Procurement should preserve access to source records and challenge rights. Providers should be able to reproduce a reported result without assembling an explanation after the review starts. The roles should be adapted to the organisation’s retained capability, site risks and contractual authority.

Practical recommendations

• Start with one service where the current evidence can be reconstructed and where a better decision would matter. • Pilot the tool with client and provider teams, record disagreements in definitions and revise the workflow before scaling. • Report exceptions, uncertainty and overdue verification alongside compliance results. • Review the control after a material incident, portfolio change, system migration or contract variation.

Risks, limitations and unresolved questions

Sampling cannot prove that every record is correct, and a small audit can overemphasise unusual cases. The method should therefore state why items were selected, which evidence was unavailable and what conclusion the sample can—and cannot—support. The article provides an operational framework rather than contract wording, legal advice or a site-specific technical standard.

Source notes

[1] Cabinet Office. The Contract Management Playbook. UK Government, 2026-03-25. March 2026 edition. Printed pp.45–49, 50–55 and 67; PDF page index = printed page + 3. Accessed 1 September 2026. https://www.gov.uk/government/publications/the-contract-management-playbook Contains public sector information licensed under the Open Government Licence v3.0. https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/ [2] Australian National Audit Office. Procurement and Contract Management by the Department of Foreign Affairs and Trade for its Security Enhancement Program. Australian National Audit Office, 2026-03-18. Auditor-General Report No.25 of 2025–26. Summary paras.11, 19; audit scope/method paras.1.17–1.18; recommendation6; chapter4. Accessed 1 September 2026. https://www.anao.gov.au/work/performance-audit/procurement-and-contract-management-by-dfat-for-security-enhancement-program [3] Syarifah Nur Shaqina Syed Shabahar; Haryati Mohd Isa; Nor Suzila Lop; Hussain Ismail. THE CHALLENGES OF FACILITIES MANAGEMENT AND MAINTENANCE CONTRACT DOCUMENT IMPLEMENTATION FOR OFFICE BUILDINGS IN MALAYSIA. Planning Malaysia / Malaysian Institute of Planners, 2026-06-03. DOI 10.21837/pm.v24i42.2033; volume24 issue3, pp.102–116. Methodology pp.108–109; Table2 p.109; findings pp.110– 113; limitations p.114. Accessed 1 September 2026.

Editorial and visual note

This publication is original Osmos Global analysis informed by the cited sources. Reported findings are distinguished from Osmos recommendations and illustrations. Source findings and trademarks remain attributable to their owners. The content is general research and does not replace contract-specific, legal, engineering, safety or other professional advice.