Six Severity-1 Defects and a Monday Go-Live
Kestrel Health has engaged an integration vendor to build a claims interface between its new patient system and an external clearinghouse. The contract's acceptance clause reads: "System is accepted when user acceptance testing is complete with no Severity-1 defects open and defect leakage below 5%." The vendor has declared the build complete and requested sign-off. UAT has executed 320 test cases and logged 42 defects, of which 6 are Severity-1 (claims rejected or duplicated). A further 17 test cases could not be executed because test data was not loaded. The vendor's project manager argues the six Severity-1 items are "configuration differences, not defects," and that the 17 blocked cases are "out of scope for this release." The sponsor is under pressure from the clinical operations director to switch the interface on next Monday, because the manual workaround is costing overtime every day.
1. Should the project manager sign the acceptance? Justify the answer.
Read the worked answer
No. The decision is not a matter of opinion - it is tested against the clause that was signed.
- Criterion 1 - "no Severity-1 defects open": six are open. The criterion fails on its face, regardless of how they are characterised. Whether they are "configuration differences" is a classification question that has not yet been resolved; an unresolved question cannot be counted as a pass.
- Criterion 2 - "defect leakage below 5%": leakage cannot be computed when 17 of 320 planned cases (5.3% of the suite) were never executed. The acceptance metric is unmeasurable in its current state, which is not the same as met.
What sign-off would actually do: trigger payment and start the warranty clock, transferring risk for a claims interface that is currently rejecting and duplicating claims to the client. In a healthcare claims context, duplicate payments carry recovery cost and compliance exposure that dwarf the overtime being avoided.
What to do instead: reply in writing that the project does not accept the deliverable, cite the specific clause and the specific evidence (6 open Sev-1, 17 unexecuted cases), and request a joint triage within a defined number of working days. Acceptance is a gate, not a mood.
2. How should the project manager resolve the vendor's two claims - that the Sev-1 items are "configuration differences" and that the 17 blocked cases are "out of scope"?
Read the worked answer
Both claims are disagreements about facts that the contract already contains; resolve them against documents, not against opinion.
Configuration vs defect:
- Take each of the six items to the requirements traceability matrix and the signed scope statement. If the documented requirement says the field validates in a certain way and it does not, it is a defect. If the requirement was never specified, it is a gap to be dispositioned - but a gap that produces rejected or duplicated claims is still a Severity-1 business impact and still blocks acceptance.
- Convene a triage meeting with both project managers and both technical leads. Every item gets a classification, an owner, and a fix date, agreed in the minutes.
"Out of scope for this release":
- The 17 blocked cases were in the agreed test plan. Something either removed them or failed to supply the data; find out which. If the vendor believes they are out of scope, ask them to point to the scope statement that excludes them.
- If the blocker is client-side test data, that is the client's action item and should be owned and dated by the client immediately - it is not a reason to leave the metric unmeasured.
The judgement: do not argue about labels. Convert every disputed item into a documented requirement reference plus an owner plus a date, and if the vendor will not agree, invoke the contract's dispute-escalation path rather than continuing a bilateral debate.
3. The clinical operations director insists the interface goes live Monday regardless. What options should the project manager put forward, and what must be true for any of them to proceed?
Read the worked answer
Reframe the question from "accept or refuse" to "what risk can the business knowingly carry, and who owns it."
Options to put on the table:
- Conditional go-live with a documented workaround. Switch on only the claim types unaffected by the six Sev-1 items; route rejected/duplicated claim categories through the existing manual process. This captures most of the overtime saving while containing the failure mode.
- Short stabilisation window. Go live Monday with a daily defect review and a named rollback trigger and rollback plan. Specify the trigger in advance - not "if it looks bad."
- Delay and complete UAT. Quantify the cost: overtime per day versus the cost of duplicate payments and recovery. Put both numbers side by side so the decision is made on comparison, not pressure.
What must be true for any of the above:
- The accountable business owner - not the project manager - signs a written risk acceptance naming the specific defects being carried.
- The vendor's warranty and defect-fix obligations survive the go-live decision; do not let a conditional go-live be read as acceptance under the contract.
- The risk register and issue log are updated, and the sponsor is told the acceptance criteria have not been met.
The judgement being tested: schedule pressure may legitimately change the decision, but it must never change the record. The project manager who writes "accepted" to end an argument has shifted the risk onto the client permanently.