PMS Integration: The Test Plan a Buyer Should Demand
System integrator: Before we approve the pms integration test plan, we need one shared definition of what working means. A connected badge or a successful API call is not enough.
PMS lead: The property system knows that a reservation, room assignment, occupancy status, or departure has changed. It must tell the access system what changed and let the access system return an outcome.
Access-control engineer: The lock and controller manage door access. The reader reports a credential; the controller decides whether that credential is valid for the door.
Buyer’s annotation: This annotated dialogue follows a system integrator through the evaluation of that interface. It treats the property-management and lock sides as two systems with separate responsibilities, separate failure modes, and a shared obligation to produce evidence. It does not assume a particular property system, lock model, credential type, or site topology.
The room is the test object, not just the door
Buyer: What exactly are we integrating?
PMS lead: A guest or resident has a relationship with a room in the property system. That relationship may begin before arrival, change during the stay, and end without the booking simply disappearing.
Access-control engineer: At the access side, the same person or role must be represented by a credential with access rights. We also need the door, controller, and schedule to interpret those rights consistently.
Buyer: So the match is not simply between a booking number and a lock. It is between a lifecycle in one system and an access lifecycle in the other.
PMS lead: Correct. The same room can be occupied by different people, and the same credential can be moved to a different room. A reconciliation report that only compares current assignments can miss both situations.
Access-control engineer: The installed hardware also matters. A valid software rule cannot compensate for a door that does not open, a reader that cannot read the presented credential, or a controller assigned to the wrong opening.
Buyer: Put the physical and logical baselines in the acceptance package, but do not turn them into a general site survey. We need only the facts that affect this interface.
Pms integration test plan is what this piece addresses for the system that has to work on a real door. Buyer’s annotation: Before execution, record the room identifier, door identifier, controller identifier, reader location, applicable opening schedule, credential presentation method, and approved behavior for each connected space. The Office Smart Lock Site Survey and Smart Lock Retrofit Site Survey can support that baseline when those published workflows apply to the project.
Direct answer. In short, pms integration test plan is the subject of this piece. pms integration test plan is where the work on pms integration test plan sits. pms-integration is what has to be evidenced before anyone signs. A pms integration test plan shows what each component must assume about the others. pms integration test plan is the interface that must be satisfied. Then commissioning starts on pms integration test plan.
- Map the interface — write down what hands over between lock, door and system at pms integration test plan
- Test the handover — record the sequence that proves pms-integration works end to end
- Keep the record — store the commissioning evidence with the file for room is test object
- In pms integration test plan, this sequence is what pms integration is checked against.
| Scope item | Decision the buyer must force | Evidence retained |
|---|---|---|
| Property record | Which reservation, occupancy, room-change, and departure events affect access | Approved event list and sample records |
| Access record | Which credential, role, schedule, and door rights are affected | Approved mapping and issued-access snapshot |
| Physical boundary | Which installed device represents each logical room | Door and controller register |
| Operational boundary | Who handles exceptions, duplicate events, and manual access | Named owner and escalation route |
| Acceptance boundary | Which environments and datasets represent production use | Signed test scope |
Buyer: If any one of those rows says “to be decided,” the test is not ready to run.
Turn the request into a question set the lab can execute
PMS lead: Our requirement is usually written as “connect the PMS to the lock system.” That tells a bidder what the project is about, but not what a tester can observe.
Access-control engineer: We need a statement with a trigger, a precondition, an action, an expected outcome, and evidence. If the expected outcome is only that data syncs, two teams can watch opposite screens and claim success.
Buyer: Separate interface behavior from business intent. The business intent might be that a departing guest loses room access. The interface behavior might be that a PMS departure event causes the access platform to remove a named entitlement and return a confirmation.
PMS lead: Then the interface can fail while the intent remains unchanged. That makes the defect easier to diagnose.
Access-control engineer: It also prevents a test from becoming a blanket approval of every lock capability in the project.
Buyer’s annotation: PMS integration requirements should be written as testable questions. A useful requirement set asks what is sent, who initiates it, what happens when it is delayed or repeated, what response is expected, where evidence appears, and which party is allowed to close the resulting defect.
| Requirement area | Buyer question | Pass condition the team must define |
|---|---|---|
| Arrival | What event indicates that room access is needed? | The expected credential and entitlement appear with traceable identifiers |
| Occupancy change | What event changes the person or role associated with a room? | Only the defined access rights change |
| Room move | How is the old room separated from the new one? | Obsolete rights do not remain active without approval |
| Departure | What event ends access, and what happens if it is repeated? | The defined revocation state is reached and recorded |
| Cancellation | What happens when access was issued before the stay begins? | The credential follows the approved cancellation path |
| Manual access | When may staff override the normal workflow? | Every override has an actor, reason, scope, and evidence |
| Return or failure | What does each side do when no response arrives? | The degraded behavior matches the approved design |
Buyer: Add a no-change control to each scenario. If the room or person does not change, the test should demonstrate that the interface does not create a change by accident.
PMS lead: And every exception should be repeatable. A one-off manual correction does not prove that the workflow is sound.
Build one map of data, commands, and responsibilities
Access-control engineer: We need to know where the integration boundary sits. Is the property system sending an event to a local API, a cloud API, a controller, or another service? We should not describe all of those as one “connection.”
PMS lead: We also need a field dictionary. A room identifier on our side may not be the identifier the access platform stores. A person identifier may not be the credential identifier.
Buyer: Do not accept a field list without direction. For every field, specify the sender, recipient, purpose, expected population, and response behavior.
Access-control engineer: Include the case where the message does not arrive. Then specify the case where it arrives twice, arrives with an unknown room, or arrives before the referenced reservation is available.
PMS lead: And clarify whether the platform receives a command or a fact. “Make this person a guest” asks for an action. “This reservation now belongs to this room” describes a business state. The access system may need both, but it should not confuse them.
Buyer’s annotation: The test map should expose every boundary that can delay, transform, or reject data. It should also identify who can replay, repair, and close each flow. A diagram is useful only when it is tied to the evidence register and defect ownership.
flowchart TD
A[Agree the interface boundary] --> B[Map inbound property events]
A --> C[Map outbound access decisions]
B --> D[Define expected responses]
C --> D
D --> E[Inject missing and repeated messages]
E --> F[Assign evidence and owners]
F --> G[Approve the release decision]
| Direction | Required content | When it does not arrive | Owner of the response |
|---|---|---|---|
| Property system to access platform | Reservation or stay event, person or role reference, room reference, effective time, operation type | The platform applies the approved retry, queue, alert, or manual-repair path | Named on both sides |
| Access platform to property system | Requested action, operation reference, result, reason, completion time | The property system uses its approved stale-state or exception procedure | Named on both sides |
| Controller to access platform | Door state, device identity, event time, source | The platform distinguishes an absent update from a reported closed state | Access-control owner |
| Access platform to controller | Entitlement update or access-state instruction | The controller follows its approved degraded behavior | Controller owner |
| Person or process to both sides | Manual grant, override, correction, or cancellation | The action is not treated as complete until evidence exists | Operational owner |
Buyer: Who owns the clock on each field?
Access-control engineer: That belongs in the map before any test is run. A scheduled right that is interpreted against the wrong clock is not a data problem; it is an authorization problem.
Put one clock and one event vocabulary on the page
PMS lead: Our property events use system time, but the business day can differ from the clock day.
Access-control engineer: The controller and lock also create events. If the property platform, access platform, and controller clocks differ, the same action may appear to occur before or after the permission change.
Buyer: Then the plan needs one time reference, one timezone rule, and a method for showing the source clock. Do not let the tester silently correct timestamps.
Access-control engineer: We should retain the original event time, the receipt time, and the time at which the system acted. That lets us separate late delivery from a delayed entitlement.
PMS lead: We should also decide which event is authoritative for a guest-room schedule: the booking’s arrival time, the recorded check-in, or an access-platform activation time.
Buyer: That is a business rule. Name the rule, its owner, and the evidence that demonstrates it.
Buyer’s annotation: Access control testing becomes unreliable when the teams describe the same event with different words. The test plan should define event names for creation, modification, cancellation, activation, expiry, denial, successful access, and administrative override. It should distinguish an audit log from event memory: the audit log is the retained record with time and source, while event memory is the lock’s local buffer when it cannot reach the server. These are different objects, and the handover package must state where each is visible.
| Event question | Decision to record | Expected evidence |
|---|---|---|
| Time basis | Which system clock defines the effective time? | Configuration snapshot and synchronized clock evidence |
| Time zone | Which zone applies to display and cutoff decisions? | Approved timezone and boundary-case result |
| Event identity | Which reference connects a request, result, and controller action? | Traceable reference in exported records |
| Late arrival | Does an older event overwrite a newer state? | Ordered-event test with final state |
| Duplicate arrival | Is the repeated event harmless or does it trigger another action? | Repeated-delivery result |
| Disconnection | Which events remain in event memory, and which are only held upstream? | Buffer evidence and reconciliation procedure |
| Source record | Where is the audit log retained, and who can read it? | Access list, export sample, and retention record |
Buyer: If two teams give different answers to any row, that is a test-plan defect, not a test-execution defect.
Credential reconciliation is a sequence, not a database match
Buyer: Before we test individual lock functions, can we prove that the access side represents the intended person or role?
PMS lead: The property record may identify a person, an accompanying guest, an employee role, or a service arrangement. It does not automatically create a credential.
Access-control engineer: The access platform may issue a physical card, a phone-held credential, or another form. The person, role, credential, and personalisation data are separate layers.
Buyer: So credential reconciliation must compare the whole chain, not just whether two identifiers are equal.
Access-control engineer: Exactly. The right test asks whether the credential belongs to the expected identity, carries the intended access rights, points to the correct room, begins and ends at the approved times, and is not shared with another person.
PMS lead: It should also detect orphans. A credential may remain active after a cancellation, or a reservation may exist without the access record that should have followed it.
Buyer’s annotation: Credential reconciliation should run in both directions. The forward check asks whether every event requiring access produced the expected credential state. The reverse check asks whether every active credential has a valid business basis. A forward-only report can miss a lingering entitlement; a reverse-only report can miss a missing one.
| Reconciliation scenario | Question under test | Evidence expected |
|---|---|---|
| New reservation | Was access created only for the assigned space and role? | Access-platform record linked to the property event |
| Additional occupant | Did the change add access without disturbing the existing occupant? | Before-and-after entitlement snapshots |
| Room change | Was the former room entitlement removed and the new one added? | Timestamped change history for both rooms |
| Cancellation before arrival | Did provisional access expire or disappear as designed? | Final credential state and cancellation trace |
| Checkout | Did departure remove the intended room rights? | Revocation event and resulting entitlement |
| Repeated departure | Did a duplicate event create an unintended second action? | Duplicate-handling result |
| Reassigned room | Can a stale property event restore old access? | Out-of-order event test |
| Shared role | Is a service credential distinguishable from a personal credential? | Role, owner, and personalisation fields |
| Manual exception | Does an override have a reason and an expiry? | Audit record and later state review |
Buyer: Where is the mismatch repaired?
PMS lead: If the property record is wrong, the property owner corrects it and triggers the defined process.
Access-control engineer: If the credential or entitlement is wrong, the access owner investigates issuance, delivery, mapping, and controller assignment.
Buyer: Name both repairs. The test should not pass merely because someone can eventually make the screen look correct.
Run the stay lifecycle as a series of controlled departures
PMS lead: The dangerous cases are often transitions: a booking changes before anyone arrives, a room is reassigned, or a departure is posted before a late access need has been resolved.
Access-control engineer: The normal path is easy to overtrust because the tester knows what should happen. We need to execute each transition from its actual predecessor state.
Buyer: Build a lifecycle script with one controlled property event at a time. Pause between actions, capture both systems, and only then continue.
PMS lead: That allows us to attribute a mismatch to a particular change rather than to a batch of events.
Access-control engineer: It also reveals whether a repair changes earlier state. If a room move is corrected, does the old entitlement return, remain removed, or enter a review state? The approved answer must be explicit.
Buyer’s annotation: The PMS Integration Requirements Template can become the source for these scenarios, but the executable script needs more than requirement text. For every transition, capture the starting state, the property-system event, the access-system request, the controller result, the operator action, the final state, and the evidence location.
| Lifecycle stage | Controlled action | Fault that may be injected | Expected result to define |
|---|---|---|---|
| Before arrival | Create or amend a reservation | Repeat the event | One correct entitlement state with traceable history |
| Credential issue | Bind access to the expected person or role | Delay the issuance response | No false completion and no untracked active credential |
| Arrival | Record occupancy or check-in | Deliver an older arrival event | Current rule remains deterministic |
| Stay | Change room or role | Interrupt the outbound update | Approved incomplete or degraded state is visible |
| Departure | Post checkout or end access | Delay or repeat the event | Revocation follows the approved timing rule |
| Post-departure | Reassign the room | Inject the old event late | New occupant rights are not overwritten |
| Dispute | Correct the property record | Apply the correction manually | Audit evidence links both the original and corrected state |
Buyer: Include a no-access-needed control. Some lifecycle events may need to close one entitlement without creating another. Success is the correct absence, not merely the presence of another record.
The outage rehearsal belongs in the plan
Buyer: What happens when the network, controller, or server is unavailable?
PMS lead: If the property system cannot reach the access platform, a new reservation may exist without access being provisioned.
Access-control engineer: If the controller cannot reach its platform, the door may continue on a local schedule or use another defined state. The test must identify which behavior the design actually provides.
Buyer: Do not turn that into a universal promise. The correct degraded behavior depends on the approved door role and the selected architecture.
PMS lead: We also need to distinguish an unavailable PMS from an unavailable access platform. One may stop new business events while access administration remains available; the other may stop access decisions while the property system continues to change.
Access-control engineer: In each case, we should know whether the event is queued, dropped, blocked, manually recoverable, or visible as stale.
Buyer’s annotation: Degraded mode testing is not an optional resilience demonstration. It is a controlled part of system integration. The offline case is the design, not the exception. The acceptance result should identify the exact architecture tested: centralised or standalone, local API or cloud API, and whether decisions, schedules, or event buffers remain available during the interruption.
| Failure injection | Question under test | Evidence expected | Decision owner |
|---|---|---|---|
| Property system unavailable | Can staff distinguish a new request from an old status? | Interface health state and operator view | Property-system owner |
| Access platform unavailable | Are outstanding property events visible and owned? | Queue or exception report | Integration owner |
| Controller disconnected | Does the installed door retain its defined local behavior? | Controller status and door test | Access-control owner |
| Server unreachable from lock | Which events remain in event memory? | Buffer state and later reconciliation | Access-control owner |
| Delayed event | Does a stale request overwrite a newer entitlement? | Ordered-event result | Integration owner |
| Repeated event | Is the operation idempotent or explicitly non-repeatable? | Repeat-delivery trace | System owner |
| Time source unavailable | Is the effect visible or does a rule silently use a fallback? | Clock-state record | Operations owner |
| Recovery | Are queued actions replayed, reviewed, or cancelled? | Approved recovery log | Named service owner |
Buyer: For every row, write the condition that allows the test to close. “The system recovered” is not a condition. Name the reconciled state, the evidence that proves it, and the person authorized to accept it.
Access-control engineer: And do not approve a door state in the test report. If the interruption leaves the access state unresolved, escalate it to the authority that approved the door’s fail-safe or fail-secure behavior.
Prove the controller boundary, reader boundary, and lock state
Buyer: The test script needs to use the vocabulary precisely. What is each device actually reporting?
Access-control engineer: The reader reports a credential or request. The controller holds the door state and makes the decision. A door-position contact is a switch that reports open or closed; it reports a state, not a cause.
PMS lead: So a door-open event should not be interpreted as proof of who opened it or why it opened.
Access-control engineer: Correct. If access history is needed, the source must be identified and the inference must be separately approved.
Buyer: What about an exit device?
Access-control engineer: Request to exit is the signal that an exit device has been operated from the escape side. It is not a release command. Release on the escape side is the behavior of the exit device, while the RTE signal tells the system that the operation occurred.
Buyer: That distinction belongs in the evidence too. A test must not report an RTE signal as though the access platform instructed the door to release.
Buyer’s annotation: These boundaries prevent misleading test observations. The lock reports the state; the system decides what the state means. A successful credential presentation, a door-open report, and an exit-side operation are different observations and should be captured in separate fields.
| Observation | What it proves | What it does not prove |
|---|---|---|
| Credential presented | A credential was presented to the reader | That the person was the credential holder |
| Access granted | The controller accepted the presented request for that decision | That the door physically opened |
| Door open | The door-position contact observed the open state | Who opened it or why |
| Request to exit | The exit device signalled operation from the escape side | That the access platform commanded a release |
| Entitlement active | The access platform holds a current right for the credential | That every downstream controller applied it |
| Audit record | A retained event has a time and source | That it is complete without buffer reconciliation |
Buyer: Add a field-proven step for any controller or lock function used in the scenario. A software response can be correct while the installed path is not.
Access-control engineer: And if the project uses a legacy reader wiring convention or an addressed bidirectional protocol, the interface test should confirm the expected message behavior rather than infer it from a successful unlock.
Put privacy and security reviewers in the room before execution
Buyer: The acceptance data may contain names, room assignments, and access events. Who needs to see it, and who decides how it is handled?
PMS lead: We can use synthetic test identities for most functional work, but reviewers still need to know the shape of production records and the fields that will be exposed to each interface.
Access-control engineer: The access side also creates event history. That history may include credential identity, door identity, time, result, and administrative changes.
Buyer: Then privacy is part of the test environment, not a policy attached after the tests pass. Use realistic data shapes without carrying unnecessary personal data into screenshots and sample exports.
PMS lead: We should also test who can retrieve an event and whether the access record explains why a person had access at a particular time.
Access-control engineer: The answer must connect the entitlement decision, the credential, the room mapping, and the event. A log without the decision context is difficult to interpret, but collecting extra context without a defined purpose creates another problem.
CONFIRM-WITH-LOCAL-PRIVACY: Confirm with local privacy counsel which personal-data fields may be collected, exposed to integration partners, retained in test evidence, and used to investigate access events in the deployment jurisdiction.
| Review area | Buyer question | Required decision |
|---|---|---|
| Test data | Can the test use synthetic identities without changing system behavior? | Approved dataset and substitution method |
| Data visibility | Which role can view personal or credential-linked events? | Named access list |
| Evidence handling | Where are screenshots, exports, and logs stored? | Approved location and access control |
| Retention | Who sets and approves retention for each record class? | Record-specific retention rule |
| Investigation | How is a request for access history validated? | Authorized request and release procedure |
| Support access | What diagnostic data can a support team receive? | Minimum-field support package |
| Deletion | How are expired test records and obsolete exports handled? | Recorded disposal procedure |
Buyer: Put the privacy reviewer’s decision in the test pack. If the answer is unresolved, keep the test conditional and state exactly what remains open.
Separate fire, egress, and everyday access decisions
Buyer: This is a property-management interface, but the doors may also participate in building systems. How do we keep those decisions out of an ordinary access test?
PMS lead: The PMS path manages business-driven entitlement. It should not be treated as a substitute for an approved safety interface.
Access-control engineer: Likewise, a fire or egress event must not be validated merely because the property platform shows a room as vacant.
Buyer: Put the interfaces in separate test sections with separate authorities and separate evidence. The test team may observe signals, but it must not invent the release rule.
CONFIRM-WITH-LOCAL-EGRESS: Confirm with the local authority responsible for the installed egress arrangement which observable states, release behavior, and inspection evidence are required for each door.
CONFIRM-WITH-LOCAL-FIRE-INTERFACE: Confirm which authority signs off the fire-alarm interface, how a signal is routed, and what the approved installed-door behavior is for the specific site.
| Boundary | Correct test treatment | Evidence to retain |
|---|---|---|
| Property-driven entitlement | Test arrival, room change, and departure transitions | Event trace and final access state |
| Exit-device operation | Observe the RTE signal without redefining it as a command | Device signal and separate door observation |
| Door-position contact | Record open or closed as a state | Source-labelled contact event |
| Fire-related interface | Execute only under the approved test method and authority | Sign-off, method, and observed result |
| Loss of power | Test the behavior approved for that door’s role | Door-specific approval and observed state |
| Centralized architecture | Identify what depends on server availability | Architecture record and outage result |
| Standalone architecture | Identify what remains locally available | Device configuration and local behavior result |
Buyer: If a test depends on access that may affect free movement, it must not be improvised from a general integration script. Move it into the locally approved method.
Access-control engineer: And keep the ordinary PMS test free of unsupported assumptions. A vacant room is a property state, not proof that a lock should change state.
Separate five test layers so a green unit test is not a false release
Buyer: How do we prevent each team from declaring victory inside its own system?
PMS lead: We need a contract test, then an interface test, and then the complete workflow across both systems.
Access-control engineer: The installed door and controller also need verification. A pass in the property platform cannot prove that the credential works at the reader or that the controller applied the entitlement.
Buyer: Divide the work into layers and record each result separately.
PMS lead: That lets us locate a failure. A malformed field can be found at the contract layer without opening a door, while a controller mapping error can be reproduced at the interface layer.
Buyer: Also state which environment each layer uses. Production data, test data, and simulated environments should not be confused in the evidence index.
Buyer’s annotation: Property management system integration should be judged as a chain. The system integration test proves that the two platforms exchange the intended data. Access control testing then proves that the access decision reaches the relevant controller and installed door. The end-to-end layer proves the business outcome. Degraded testing covers the interruption paths, and operational rehearsal proves that the support team can use the evidence.
| Test layer | Objective | Representative environment | Failure can be owned by |
|---|---|---|---|
| Contract and schema | Validate field names, types, identifiers, and response forms | Isolated fixtures and interface logs | Both interface owners |
| Process workflow | Validate property events and resulting access-state changes | Joined test environment | Integration owner |
| Controller and device | Validate the decision path to the installed opening | Controller bench or test door | Access-control owner |
| End-to-end business test | Validate arrival, change, departure, and reconciliation | Representative site setup | Shared service owner |
| Failure and recovery | Validate missing, delayed, repeated, and unavailable conditions | Controlled fault injection | Owner named for each failure mode |
Buyer: A release decision must show the result of every layer. If one layer is not run, mark it as not executed and explain the residual risk. Do not convert an untested layer into an assumed pass.
Write scenarios as executable scripts
Buyer: Give me the shape of an executable scenario, not just a checklist.
PMS lead: Start with the property-side precondition. Which record exists, which event will be sent, and which identifiers are attached?
Access-control engineer: Then capture the expected access request, the controller result, the credential or entitlement state, and the audit evidence. Add the time at which each observation was taken.
Buyer: The script must also say what the tester should do if the result differs. “Investigate” is not an instruction.
PMS lead: It should say whether the tester stops the sequence, retries the event, records a duplicate, or opens a defect. Retrying can destroy evidence if it happens before the first failure is captured.
Access-control engineer: And no personal data should be copied into free-text notes. Use the approved test identity and reference the system record.
Buyer’s annotation: The following script library covers the core lifecycle. Each row expands into a test sheet with precondition, action, expected result, evidence, owner, and disposition. Additional rows should cover the business rules identified for the deployment.
| Scenario | Precondition | Action | Expected result | Evidence |
|---|---|---|---|---|
| Create access | Approved reservation and mapped room exist | Send the create or activation event | The expected role receives only the defined room rights | Property event, access request, final entitlement |
| Duplicate create | Valid access already exists | Send the same operation again | No unintended second right or duplicate event | Before-and-after state and repeat trace |
| Add occupant | One person has access | Add another approved occupant | Existing access remains; new access matches the rule | Two entitlement snapshots |
| Move room | Person is linked to the former room | Send the room-change event | Former rights end and new rights begin as approved | Trace covering both rooms |
| Cancel provisionally issued access | Access exists before arrival | Cancel the reservation | Access follows the approved cancellation timing | Cancellation event and final state |
| Check out | Person is active in the room | Post departure | Intended access ends without affecting unrelated records | Revocation trace and comparison |
| Unknown room | Event carries an unmapped room | Deliver the event | The interface rejects or holds it under the defined rule | Error record and operator outcome |
| Stale event | A newer change is already active | Deliver an older event | The newer state is not overwritten | Ordered-event result |
| Delayed response | Access request reaches the platform | Delay the response | The requester can distinguish pending from failed | Health state and retry record |
| Controller unavailable | An entitlement is ready to apply | Disconnect the controller | No false success is reported | Controller status and pending record |
| Recovery replay | Events were held during interruption | Restore service and reconcile | Each event is applied, reviewed, or cancelled with history | Reconciliation report |
| Reverse reconciliation | Access platform contains active records | Compare against property records | Every active credential has a valid business basis | Orphan and mismatch report |
| Manual override | An exception needs temporary access | Apply an authorized override | Scope, reason, actor, expiry, and later removal are visible | Audit record and final state |
Buyer: Add a negative control to every destructive scenario. The tester should show that unrelated rooms, people, and credentials remain unchanged.
PMS lead: That is especially important after a room move or a cancellation, where a broad update could look successful while reaching beyond the intended person.
Make evidence readable by another team
Buyer: What will the buyer receive at the end of the run?
Access-control engineer: Not a folder of screenshots with unexplained filenames. The evidence must connect each scenario to its result.
PMS lead: Include the input used, the observed output, relevant logs, and the person who ran the test. Redact or substitute personal data according to the approved handling rule.
Buyer: Distinguish raw evidence from interpretation. Let the log show what the system recorded, then let the test note state whether it met the requirement.
Access-control engineer: If a defect is open, the evidence should point to the defect record without embedding a second, conflicting explanation in the log.
Buyer’s annotation: Handover documentation should preserve the chain from requirement to result. Use the Documenting PMS Integration for Handover and Audit workflow to structure the package. The purpose is not to create a large archive; it is to let an authorized reviewer reproduce a decision without relying on the tester’s memory.
| Evidence artifact | Required content | Retention rule |
|---|---|---|
| Approved scope | Included rooms, systems, scenarios, exclusions, and owners | Match the approved release under review |
| Interface map | Direction, identifiers, fields, responses, and boundaries | Current revision with change history |
| Test script | Preconditions, actions, expected results, and fault injections | Locked before execution except through recorded change |
| Run record | Environment, tester, execution time, result, and attachments | Stored with the corresponding scenario |
| Raw logs | Unedited system output relevant to the observation | Protected and access-controlled |
| Defect record | Expected versus actual result, severity, owner, and status | Active until accepted closure or superseded |
| Reconciliation report | Forward and reverse credential comparison | Generated from the approved test population |
| Release decision | Passed, failed, conditional, or deferred items with rationale | Signed by the named decision authority |
Buyer: If the evidence cannot be mapped to a scenario, place it in a general appendix and do not count it as proof of that scenario.
Define failure severity and retest rules
Buyer: Some failures will be trivial; others may make the interface unsafe to operate. How do we distinguish them without arguing after the fact?
PMS lead: A mismatch that leaves one entitlement active after departure may have a different consequence from a formatting issue that blocks a new reservation.
Access-control engineer: The classification must include the behavior at the door, not just the inconvenience in the interface.
Buyer: Agree the severity meanings before testing. Then link each defect to the person empowered to accept it.
PMS lead: A waiver should identify the affected scope, the interim procedure, and the condition for closure. An undocumented exception becomes normal operation without anyone deciding that it is acceptable.
Access-control engineer: A retest should verify the corrected behavior and check the side effects. Fixing a room-change event may inadvertently alter cancellation or checkout handling.
Buyer’s annotation: The PMS Integration Failure Patterns and How to Recover reference is useful for recognizing recurring symptoms, but the acceptance rule remains local. The Who Owns PMS Integration: Supplier, Integrator or Client decision must be resolved before a defect can have an accountable closer.
It is worth pairing this with the smart lock environmental testing. That is the same question pms integration test plan has to answer here.
| Severity class | Definition for this test plan | Release treatment | Retest requirement |
|---|---|---|---|
| Release-blocking | The workflow can grant unintended access, remove required access unpredictably, or conceal a material mismatch | Release cannot proceed for the affected scope | Full affected workflow plus adjacent regression set |
| Major | The property and access states diverge and require manual intervention | Proceed only with an approved exception and defined monitoring | Corrected scenario and the manual process used during the exception |
| Minor | A non-critical mapping, display, or evidence issue does not change the authorization result | May proceed with an owner and closure condition | Targeted scenario and review of related evidence |
| Observation | A behavior is outside scope but may affect future use | Record for design review; do not disguise it as a pass | No retest unless scope changes |
Buyer: The retest rule is simple: a corrected result closes the defect only when the evidence shows both the expected outcome and the absence of the original side effect.
Make acceptance a visible decision, not an email trail
Buyer: Who says the interface is accepted?
PMS lead: The property owner should confirm that the lifecycle rules match the business. The access owner should confirm the authorization and controller behavior.
Access-control engineer: The integration owner should confirm that failures, reconciliation, and evidence work as designed. The person operating the system should confirm that exceptions can be handled.
Buyer: Those are different approvals. One team cannot sign for all of them, even if the same person attends the meeting.
PMS lead: Any conditional pass should state what the operator does while the remaining defect exists.
Buyer: And any deferred item should have an owner, a boundary, and a condition for review. “Follow up later” is not a disposition.
Buyer’s annotation: Use a scorecard that records evidence rather than subjective confidence. A scorecard supports the decision; it does not replace it. Critical workflow categories should have no hidden exceptions, while conditional categories must be visible to operations.
| Acceptance dimension | Evidence required | Pass basis | Decision owner |
|---|---|---|---|
| Lifecycle behavior | Arrival, change, cancellation, and departure results | Every required transition matches the approved rule | Property and access owners |
| Credential reconciliation | Forward and reverse mismatch reports | No unexplained active entitlement remains | Access owner |
| Data contract | Field, identifier, and response validation | No undefined mapping or silent coercion | Integration owner |
| Time and event meaning | Clock reference, event vocabulary, and ordered-event results | Events can be placed and interpreted consistently | Both system owners |
| Failure behavior | Missing, repeated, delayed, and unavailable tests | Each interruption follows the approved degraded design | Service owner |
| Installed access path | Controller and door evidence | The intended decision reaches the installed opening | Access-control owner |
| Security and privacy | Reviewer decisions and data-handling record | Required controls and approvals are documented | Security and privacy reviewers |
| Egress and fire interfaces | Local authority method and sign-off where applicable | Testing follows the approved boundary | Named approving authority |
| Operations | Exception procedure and support rehearsal | An operator can identify, contain, and escalate the condition | Operations owner |
| Evidence and handover | Indexed results and signed release record | Another authorized reviewer can trace every decision | Integration owner |
Buyer: Put the table into the release record. State the final disposition plainly: accepted, accepted with conditions, or not accepted for the defined scope.
The buyer’s final review: questions with no assumed answers
Buyer: Before I sign, what must each side send?
PMS lead: The approved event set, identifiers, effective times, operation references, and responses. The plan must also say what happens if one of them does not arrive.
Buyer: Who owns the clock?
Access-control engineer: The plan names the reference, timezone rule, source timestamps, and behavior when the clock is wrong or unavailable.
Buyer: What is written down, where does it live, and who can read it?
PMS lead: The property event, the access request, the decision, and any repair are separate records. Their locations and access rights must be named.
Buyer: What happens when the network, controller, or server is unavailable?
Access-control engineer: The answer belongs to the approved architecture and door role. The test must show the actual degraded path, not an assumed universal one.
Buyer: Which side owns the fix?
PMS lead: The interface owner coordinates, but the party that controls the faulty record or device owns the correction.
Buyer: Those are the five questions. If any answer is vague, the release decision is not ready.
Buyer’s annotation: Use this final review as a gate over the complete test package rather than as a verbal walkthrough. Every answer should point to a named requirement, scenario, evidence artifact, and owner. The pms integration test plan is complete when another authorized reviewer can trace the business request through the interface, the access decision, the installed door, the failure path, and the retained evidence.
CONFIRM-PER-SUPPLIER: Obtain written confirmation of every proposed lock, controller, credential, interface, and model-specific capability; confirm on the dated product page for the specific SKU before approval.
Buyer: Send the completed scope, interface map, scripts, evidence index, defect register, and approval record with the equipment request. Ask the supplier to identify every item that still needs written confirmation rather than inferring a capability from a general product description.
System integrator: And keep unresolved items visible until their owners close them. A signature should confirm the tested behavior and the stated boundary, not transfer an unknown risk into operations.
Once the requirement is written down, request a quote is the move.

