A self-assessment for the contracting authority's team during the operating phase: is each part of the contract management regime in place and evidenced? Ten operating dimensions, one status each, and a suggested band that a named person confirms.
v1.1 — September 2026
Twenty-five questions across ten dimensions, each answered from evidence the team holds. Every dimension gets its own status. The overall band is the worst dimension, never an average, and no band is issued until every question is answered. Questions and the "what to consider" text are ported from the rule families and mechanism checklists of the PPP Contract Review & Operating Map. Paste the contract text and the tool proposes the profile and the evidence notes; nothing is filled in until you accept it.
Stage, financing and revenue model are required before the check runs. Stage and financing decide where "Not applicable" is offered; the revenue model decides which payment questions are asked.
Paste the agreement, schedules included (open the PDF in your reader, select all, copy). The tool looks for the mechanisms each question asks about and proposes the profile and an evidence note per question, each with the term found and where. Nothing is filled in until you accept it, and no answer is ever set for you: the contract shows what the regime should contain, not whether it is being run. The text stays in this browser tab.
| Dimension | Status | Basis |
|---|
The band above is a suggestion computed from the answers. A named person records the decision; an override does not change the suggestion, it sits beside it in the export. The block clears whenever the check is re-run.
The Contract Health Check asks a contracting authority's team whether the machinery for managing a signed PPP contract exists and is being used. It is a screening-level self-assessment for the operating phase. It does not read the contract, it does not judge the private party's performance, and it does not determine any claim, deduction or payment. A separate tool in this series, the Performance Monitoring Maturity Assessment, asks how well the monitoring regime is run; this one asks whether the regime is there at all.
The dimensions, questions and "what to consider" text are ported from PPP Contract Review & Operating Map v1.0.0 (Benchmark Baseline; formerly "PPP Contract Copilot"), a rules-based product built on the Saudi National Center for Privatization & PPP (NCP) Project Agreement Precedent Template. Principally three of its parts are used here:
The status pattern (one status per dimension, never blended; a suggested classification that a named person records) is the product's readiness-dashboard pattern from its benchmark review mode. Its statuses are pass, watch, fail and insufficient data, shown here as In order, Watch, Action needed and Not evidenced. At the dimension level the third status reads "Action needed"; at the band level the same tier reads "Intervention needed".
The payment dimension has two forms. Availability-payment and hybrid contracts answer four questions about the service payment, deductions, persistent failure and disputed amounts. User-pay contracts answer four questions about tariff adjustment, revenue reporting and audit, fees or support payments, and the performance regime. Both forms are listed.
Each question is answered Yes, Partly, No or Don't know. A few questions may also be marked Not applicable, and the option is offered only when the contract's profile justifies it: payment and handback questions before service commencement, and the direct-agreement question where there are no third-party lenders. If the profile changes, the option is withdrawn and the question must be answered again. The answer becomes a question status:
| Answer | Core question | Supporting question |
|---|---|---|
| Yes, in place and evidenced | In order | In order |
| Partly, or applied inconsistently | Watch | Watch |
| No | Action needed | Watch |
| Don't know / not checked | Not evidenced. Holds the dimension at Watch or worse; a dimension answered Don't know throughout reads Not evidenced. | |
| Not applicable | Excluded. A dimension whose questions are all Not applicable reads Not applicable. | |
The dimension takes the worst of its question statuses, in the order Action needed, Watch, Not evidenced, In order. The product's rule is that the worst check wins and that insufficient data counts only when every check lacks data. This tool is stricter on one point, by design: any Don't know holds its dimension at Watch even when every other question passes, because not knowing whether a control exists is itself a management gap. The product's dashboard does not need that rule, since its inputs are measured metrics rather than self-reported answers.
There is no score, no weighting and no average anywhere in the tool. One dimension at Action needed produces Intervention needed however strong the others, because a missing performance bond is not offset by well-run committees. The second gate mirrors the product's cascade, which refuses a classification when too many dimensions lack data. The count of dimensions in each status is shown for orientation only.
The band is a suggestion. The results tab carries a confirmation block where a named person accepts it or records a different band with a reason. Both the suggestion and the recorded decision appear in the export; the override never rewrites the suggestion, and the block clears each time the check is re-run so a decision cannot attach to a result it was not made on.
The optional "Read from the contract text" step on the Contract tab looks for the mechanisms each question asks about. It is a port of the presence tests the product's completeness rules run (plain text patterns, for example Handback Requirements, Performance Bond, Direct Agreement, Warning Notice) together with the term aliases the product recognises for the parties (Procurer, Authority, Grantor; Contractor, Project Co, Concessionaire) and the locators its mechanism registry records. The product runs its tests over operative text only; this step runs over the whole pasted text, ranks hits in the operative provisions ahead of hits in the definitions block, and says when a term appears in the definitions only. Where the question asks the team to know a value (a survey timing, a cure period, a payment period, a bond amount, a committee's meeting frequency), a second pattern lists the phrases of that kind found near the term, up to three, each with its locator, for the team to check which one applies. Phrases inside square brackets (optional or placeholder drafting in a precedent) are not read.
Everything the step finds is a proposal. Each one shows the term found, the nearest preceding clause or schedule heading (inferred, so it reads "near"), and a short quote, and each needs an explicit Accept before it is written anywhere. Accepting a profile proposal sets the revenue model or financing structure, after a confirmation where that would clear answers already entered; accepting a note proposal writes the finding into that question's evidence note, prefixed "Contract text:". No proposal ever sets an answer. The contract can show that a mechanism exists and what its parameters are; whether the committee has met or the bond is in force is a project fact only the team can supply, and the product this tool draws on refuses to merge the two.
Limits of the step, stated plainly. It matches drafting patterns, and drafting varies: a mechanism the contract calls something else will not be found, and a defined term used in a definition only can be found where no regime exists. A question listed under "not found in the text supplied" is exactly that, not evidence that the mechanism is absent. Scanned PDFs yield no text to paste. No accuracy figure is claimed for the patterns; the product they come from makes the same refusal for its own completeness layer, which its authors describe as carrying a material false-positive rate by design.
A contract in its final years (stage Expiry preparation or Handback, or five or fewer years to expiry) with handback planning below In order is named in a data-quality note, because the survey, remediation and retention timetable then runs against a fixed date. Notes are reported, never counted.