Workflow evaluation
Reuse the acceptance method. Keep each client's evidence separate.
A shared test can pass while the client workflow fails. Follow one interface change through its acceptance conditions, execution boundary and delivery handoff.
6 minute read By Zof Updated October 26, 2026
This guide presents an evaluation method, not a report of a completed customer engagement.
- Workflow
- A service request passed between two client applications
- Change
- A revised message contract and approval requirement
- Constraint
- Client configuration, access and evidence stay separate
- Decision
- Whether the receiving team can accept this release
The milestone report is green. The interface accepts the request. Yet the receiving application never creates the record that the client needs. The test proved delivery to a queue, while the acceptance agreement required completion of a workflow.
Use the handoff of a changed interface to establish how a reusable method becomes evidence for one client. Connect each acceptance condition to the client configuration, execution record, and observations the receiving team will need.
Start with the promise made to this client
Assume a portal submits an approved service request to a client platform through a queue. A contract change introduces a new approval reference. The receiving platform acknowledges receipt before it validates that reference. A request can therefore receive a successful HTTP response and still be rejected downstream.
The client acceptance condition is specific: one authorized request must create the intended record, preserve its approval reference and expose a traceable final status. The delivery lead needs that evidence before handoff. A successful transport check remains useful, but it answers only the first part of the question.
Separate the reusable question from the client answer
The reusable template asks whether a request reaches its final state with its identity and authorization intact. It names the observations required at submission, receipt, validation and completion. That test intent can travel between programs.
The client's schema, endpoint, identity rules, test records and expected status values cannot be assumed to travel with it. Those belong in a separately reviewed configuration. NIST's Secure Software Development Framework provides a common vocabulary for software producers and purchasers; it does not replace the acceptance conditions of a particular contract.
Follow the workflow
From a shared method to a client acceptance record
- 01
Practice owner
Define the reusable intent
Check submission, authorized processing and final state. Specify required observations without embedding a client endpoint or credential.
- 02
Client architect
Bind the current contract
Select the message version, approval rule, consumer and expected record. Review the configuration against the deployed interface.
- 03
Authorized test team
Run inside the agreed boundary
Submit synthetic requests using client specific permissions. Capture the request identity through receipt, validation and completion.
- 04
Acceptance owner
Transfer a reproducible decision
Review evidence and exceptions for this release. Give the receiving team the method and authorized access to the relevant records.
Reconnect the template to the current system
For a Zof evaluation, the System Graph would connect the changed message contract to the queue, consumer and final record. The delivery team must confirm those relationships against the current deployment. An old diagram or a copied template cannot establish which consumer version is running.
The Verification Engine would exercise the agreed valid and invalid requests within that scope. Each run needs the application version, configuration revision, fixture identity and expected final state. If a dependency has changed, the team reviews the affected assertions before reusing the previous acceptance decision.
Keep execution and evidence inside the client boundary
Keep each client’s test credentials, execution permissions and evidence storage separate. The common library contains the method and synthetic fixture definitions. It contains no client secrets, captured payloads or production identifiers.
Choose the deployment model with the client owner. Deployment on premises may suit a restricted estate; the contract still determines access, retention and permitted exports. Hosting in an EU region alone does not establish compliance. Confirm every required interface during scoping rather than treating a named tool as proof of a supported connector.
An evaluation you can inspect
A reusable acceptance worksheet for the changed contract
Test cases for the changed service request contract. A transport response alone never satisfies the completion assertion.
| Condition | Required behavior | Evidence to inspect |
|---|---|---|
| Valid approval reference | Create the intended record once and retain the approval reference. | Submission, consumer decision and final record share the request identity. |
| Missing approval reference | Reject processing with a defined final status; create no business record. | Validation rejection and a query confirming no resulting record. |
| Repeated delivery | Follow the agreed duplicate rule without creating an unintended second record. | Both deliveries and the resulting record count for the logical request. |
| Unavailable consumer | Show pending or failed status according to policy, never successful completion. | Queue state, recovery attempt and final status after the agreed wait. |
On a small screen, scroll the matrix to compare all columns.
Failure condition to test
Failure condition: receipt passes an incomplete acceptance check
- Condition to exercise
- Submit a request without the newly required approval reference. Check whether the acceptance assertion stops at the receipt acknowledgment and misses the downstream rejection.
- Required behavior
- The acceptance check follows the request to its final state and records the rejection. A receipt acknowledgment cannot produce a passing workflow result.
- Evidence to collect
- The submitted fixture and the contract revision requiring approval.
- A receipt acknowledgment and downstream rejection tied to the same request.
- The original assertion and the final record lookup it failed to perform.
- Release decision
- Revise the acceptance assertion and repeat valid, invalid and duplicate cases before handoff. The client owner reviews the resulting evidence and any unresolved dependency.
Make the receiving team able to repeat the decision
A useful handoff identifies the release, template revision, client configuration, test inputs, results and open exceptions. It explains where an authorized reviewer can retrieve the evidence and how to rerun the check. Credentials remain in the approved secret store, outside the handoff document.
The receiving team should reproduce one passing request and one rejected request. If it cannot explain why those outcomes differ, the handoff is incomplete. The client owner accepts any remaining exception explicitly. A collection of green statuses without the tested conditions cannot carry that decision.
A repeatable method is only valuable when the evidence is specific
The acceptance threshold is that every in scope request has the agreed outcome, another authorized team can reproduce it, and evidence remains attributable to the correct client and release. An unavailable downstream service or a missing approval fixture remains an open limitation, even if the other checks pass.
Measure delivery speed and review effort on comparable changes, including template upkeep and exception handling. For delivery practices, a useful acceptance method must survive the handoff while preserving the evidence specific to each client.
Before you scope the work
Who maintains the shared template?
Name a practice owner for test intent and a client owner for local expectations. Version both. A template update triggers review when it changes the meaning of an acceptance claim.
Does every client need the same deployment?
No. Agree the environment, access and evidence location for each engagement. Reuse of a method does not require sharing execution permissions or client records.
What happens when the new team lacks access?
Treat missing evidence access as a handoff issue. Grant the minimum approved permissions or provide an agreed sanitized record. Do not replace inspection with a verbal assurance.
Can one Sprint define the whole practice model?
A Reliability Sprint can evaluate one client workflow. A wider discussion can address template ownership, contracts and deployment across the practice without pretending that one workflow proves the whole estate.
Reuse the method. Renew the evidence.
A shared template provides a starting question. Acceptance depends on the client configuration, release and observations that answer it. Preserve those connections through every handoff.
Sources and scope
These primary references explain the mechanisms and review practices used in this evaluation.
- Secure Software Development Framework, version 1.1 · NIST
A reference for shared software development practices and supplier communication. Client acceptance conditions remain specific to the contract.
Your workflow. Your evidence.
Put one client workflow through an acceptance review
Bring the interface owner, the acceptance conditions and the release under review. A paid Reliability Sprint scopes one system, typically over seven days, with access, boundaries and price agreed before work begins.
A paid engagement for one critical system. Scope and price are agreed first. Seven days is typical; access and deployment requirements can extend the timeline.
Continue the investigation
Audit, tax, and advisory software
Does an approved client workflow stay approved when its inputs change?
Follow how an amended client artifact changes the evidence required for approval.
Manufacturing and plant software
After the plant connection returns, is the order still in the right state?
See why a constrained client environment needs its own recovery assumptions and acceptance checks.
- Zof for delivery practices
Explore the broader delivery context for repeatable client work.
- Compare deployment choices
Review execution boundaries before agreeing a client configuration.
- Discuss a broader deployment
Consider ownership and operating requirements across multiple engagements.
