Pms integration hotel case: A hotel PMS integration is judged by what happens when a room and a credential disagree, and by who is authorised to reconcile.

ประกาศเมื่อ: เม.ย. 30, 2024

PMS Integration for a Hotel Property

PMS Integration for a Hotel Property

An integration failure is usually a sequence failure, so pms integration hotel case is traced here from handover to commissioning. A property management system and an electronic lock system are not two halves of one machine. They are software written by two organisations, released on two schedules, each describing the same building to itself in its own vocabulary. One of them knows that room 412 is occupied tonight and that the guest's surname is spelled with two t's. The other knows that a particular credential may open a particular door until a particular moment. The distance between those two statements is the whole of the work.

Direct answer. In brief, this article covers pms integration hotel case from pms integration hotel case to pms-integration. It states what to confirm about pms integration hotel case on the dated product page where a figure is involved. A pms integration hotel case is an integration view. Lock, door, reader and system have to behave together. pms integration hotel case is where the first interface behind pms integration hotel case fails.

  • Outline the interface — write down what hands over between lock, door and system at pms integration hotel case
  • Rehearse the handover — record the sequence that proves pms-integration works end to end
  • Log the exchange — store the commissioning evidence with the file for moment two systems disagree
  • Treat this as the pms integration hotel case sequence, with pms integration as the reference.

In a pms integration hotel case, that distance is where projects are won and lost, and it is not where the contracts are written. Contracts describe interfaces: fields, endpoints, formats, schedules. What they rarely describe is the state in which two systems are allowed to disagree with each other, and what happens in the gap between a decision taken at the front desk and its arrival at a lock. This article is about that gap. It is written for the systems integrator, the IT lead and the access-control engineer who will be asked to make two systems agree in a building that opens its doors to paying guests while people are asleep on the floors above.

The argument it makes is narrow, and worth making in full. An integration that only carries messages is half a design. The half that decides whether the system is still working in its third month is the reconciliation function: the agreed way to notice that a room in the property system and a credential in the lock have drifted apart, and the named party holding the authority to bring them back together.

The situations described below are illustrative composites. No property, guest, operator, supplier or contract named here exists, and nothing in this article states what any particular product does.

The moment the two systems disagree

Picture the ordinary hotel case, because it is less exotic than the integration diagrams usually suggest. A guest arrives late, is walked to a room, and finds that the card the desk handed over will not open the door. Nothing has crashed. The property system shows the reservation as checked in, with a room number; the lock side shows a valid authority on a door; the card carries a credential. Each system is healthy and behaving exactly as configured, and they do not agree about which room is which.

That mismatch has a family of causes and they are not equally serious. The instruction that moved the guest to this room may never have arrived. It may have arrived and been applied to the previous door. The desk may have encoded the credential from a guest profile that still carried the old room number. The room number in the property system may have been re-used by housekeeping while an authority survived the change. In every one of those cases the failure is semantic rather than technical, which is why it survives every diagnostic routine that checks whether messages are flowing. The first-party record behind this is the SLK-002 Keypad RFID Lock for Guesthouses — confirm the SKU on the dated product page.

The mirror-image case is quieter and worse. A guest leaves early, the property system records the departure and withdraws the entitlement, and the withdrawal is delayed, queued or lost. The door continues to honour an authority whose cause has ended. Nobody notices that evening. It is noticed weeks later, in a review of who could have entered which room, or not noticed at all.

These two cases frame the subject of this piece. Both are failures of correspondence, and both are invisible to a message-monitoring dashboard that only reports delivery.

The correspondence both sides have to agree

Strip the interface back and a hotel access system runs on a small set of correspondences. A room in the property system corresponds to a physical door. A reservation corresponds to a stay window. A guest or a role corresponds to a credential. The credential corresponds to an authority on that door, valid for that window. Each correspondence has an owner on each side, and each can break independently.

The room-to-door mapping deserves particular attention because it is the only correspondence that is built from the building rather than from the data. The property system's room list is logical, maintained by the operator, and liable to be renumbered. The door schedule is physical, produced by a survey of the corridor, and it does not change because somebody reorganised a floor plan. The mapping between the two is an artefact the integrator constructs, versions and puts under change control; a room-to-door mapping loaded from an export and never revisited is the most common latent fault in a hotel installation. The survey method for producing a reliable door schedule is set out in Hotel Smart Lock Site Survey.

CorrespondenceProperty system holdsLock side holdsAuthoritative forDrift that follows a disagreement
RoomA room number and its occupancy stateA door identity and its controllerThe lock side for which physical door a number meansThe authority sits on the previous door
Stay windowArrival and departure timestampsThe validity window of an authorityThe property systemAccess ends at the wrong hour
EntitlementWho is entitled, and to whatWhether a presentation is grantedThe property systemA door keeps honouring a departed guest
CredentialA profile and its personalisationA credential identifier and its holderNeither, aloneThe card carries a stale room or name
Group scopeA block of rooms and its master rulesThe doors the master can openThe property systemOne master opens a room it should not
Departure triggerCheckout, no-show, walk-out, cancellationThe moment authority stops being grantedThe property systemA late withdrawal outlives its cause

The pattern worth taking from the table is a division of truth. The property system is the authority on entitlement: who may be inside the building, and until when. The lock side is the authority on physical reality: which door an authority is attached to, and whether it was applied at all. Every reconciliation dispute in practice comes from one side being asked to answer the other side's question.

erDiagram
    RESERVATION ||--|| STAY : "describes"
    STAY ||--|| ROOM : "occupies"
    ROOM ||--|| DOOR : "opens"
    CREDENTIAL ||--|| STAY : "personalises"
    DOOR ||--o{ AUTHORITY : "honours"
    CREDENTIAL ||--o{ AUTHORITY : "carries"
    RESERVATION {
        string reservation_ref
        string arrival_ts
        string departure_ts
    }
    STAY {
        string stay_ref
        string guest_ref
    }
    ROOM {
        string room_number
        string occupancy_state
    }
    DOOR {
        string door_id
        string controller_id
        string power_loss_state
    }
    CREDENTIAL {
        string credential_id
        string holder_ref
        string personalisation
    }
    AUTHORITY {
        string authority_ref
        string granted_at
        string expires_at
        string source_system
    }

The relationship that people forget to draw is the last one. An authority is not a property of the room and not a property of the guest. It is a record with its own identifier, its own validity window, its own grant time and its own source, and it is the unit the reconciliation function has to reason about.

What the property system sends, and what it withholds

The instruction stream runs from the property system to the lock side. Each message carries, at minimum, a reservation or stay reference so that the receiver can recognise the business object rather than a fresh anonymous event; a room, mapped through the room-to-door table to a door; an arrival and a departure; the kind of credential being requested and the personalisation it needs; a flag for group and master scope; the identity of the operator or the desk station that caused it; and a sequence or version marker, so that a retry can be recognised as a retry.

What the property system sends is not the same as what the property system knows. The system also holds the rate, the payment state, the guest profile details, the notes left by the desk and the operational facts of the property — the early departure, the walk-out, the maintenance hold, the room that is out of order this afternoon. Most of that has no business crossing into the access system, and the design should say so plainly, because the alternative is a permanent temptation to send more when a fix is needed and the fix would be easier with one more field.

The clean rule, and the one I would write into the interface specification, is that the property system sends entitlements and the lock side sends events. Anything that is not an entitlement or an event needs a reason that survives a privacy review, which is a conversation the operator has to lead and not the integrator.

That boundary also decides where the degraded mode lives. If the property system calls a local interface on a server inside the building, then a building network fault is an integration fault and your team owns the diagnosis. If the property system calls a cloud interface on somebody else's service, then a third party's availability is now in your critical path, and the escalation route crosses an organisational boundary you do not control. The distinction between a local API and a cloud API is not an architectural preference; it is a statement about who gets telephoned at two in the morning. The shape of the interface on the lock side is covered in Hotel Smart Lock API Integration Guide, and the questions a buyer puts to a vendor in Hotel Smart Lock API Requirements for Integrators.

What comes back, and what the door keeps when the answer does

The return stream is smaller and, in practice, more valuable than the instruction stream. The lock side acknowledges the instruction, and then reports events: a credential presented, a grant, a refusal, a door held open, a door forced, a low-battery condition, a device going offline and returning. The refusal is the event most often discarded and the one that tells you the truth about the correspondence. A granted presentation only tells you that a room and a credential agreed; a refused one tells you precisely which half of the correspondence is wrong, and where.

Two records are created along the way and they are not the same thing. The audit log is a record of an event with a time and a source. The event memory is the buffer a device holds when it cannot reach the server. The difference between them is exactly where events are lost: a device that holds four hundred events during a network fault and delivers them afterwards has behaved correctly, and a device that drops them has failed, and only a staged test distinguishes the two. The published treatment of the trail is in Hotel Smart Lock Audit Log Management.

The behaviour of the door during that fault depends on the architecture, and the architecture has to be declared rather than assumed. Where the arrangement is centralised the authoritative state sits on a server, and cutting the link changes what the door is able to do. Where the arrangement is standalone the device holds that state itself, so the door carries on working and the link matters only for reporting. The comparison of the two arrangements in a hotel is set out in Centralized vs Standalone Hotel Locks, and the choice decides what you promise the night manager. The offline case is the design, not the exception, and it is written before the integration ships rather than discovered in the first week of operation.

The reconciliation window: how long the two sides may differ

Every integration has a degraded mode, and here the degraded mode has a dimension that engineers often miss: time. The two sides are allowed to disagree for a period, and the length of that period is not one number. It differs by the class of event, and it has a direction, because one direction of error is survivable and the other is not.

Event classThe window that mattersThe dangerous directionWhat closes it
ArrivalMinutes, ending when the guest reaches the deskNo authority yet, the guest is standing thereThe instruction lands and is acknowledged
Room changeThe turnaround between the move decision and the guest's arrival at the new doorAuthority still on the old doorThe old authority is withdrawn and the new one granted
ExtensionHours, while the guest remains inside and entitledWindow ends early, the guest is locked out at the doorThe extended expiry replaces the old one
Early departureUntil the guest leaves the buildingAuthority survives its causeThe departure instruction is applied
Group arrivalUntil the coach or group rep reaches the deskMaster issued against the wrong room setThe block's room list is confirmed against the mapping
Night audit and reconciliationUntil the operator's own books are closedTwo systems' totals disagree and nobody is toldThe reconciliation report is produced and read

A single synchronisation target applied to all six is worse than having none, because it conceals the difference between an instruction that is late and an authority that has outlived its reason. Write the window per event class, and write which side of it is dangerous, because that is what tells the front desk which error to chase. The arrival class is the one most often designed around a nightly batch rather than an instruction that exists while the guest is standing at the desk, and the workflow that the batch defeats is described in Hotel Smart Lock Check-In Workflow.

There is a second reason the window has to be explicit. The window is the contract between the two systems and the staff who use them. A design that says "synchronisation is near real time" gives the night manager nothing to hold the vendor to; a design that says "a room change is applied within the turnaround between the move decision and the guest's arrival, and the previous door is refused from that moment" gives the desk a sentence it can use on the telephone.

Six ways a hotel correspondence breaks

The register below is the working set I would expect a pms integration hotel case to carry from the first site walk to the end of the first year of operation. It is a set of states, not a set of defects, and the distinction matters: each entry can be produced by two systems that are both behaving exactly as written.

RefDriftWhat it looks like at the doorWho reconciles itWithin what windowThe artefact that settles it
D1Orphan authorityThe door opens for somebody the desk believes has goneLock platform owner applies the withdrawal, operator confirms the departureUntil the guest leavesA refusal on the old window and an applied withdrawal
D2Missing authorityThe door refuses a guest who is checked inProperty system owner reissues, integrator checks the mappingMinutes, ending at the deskThe instruction log and the applied grant
D3Stale door attachmentThe new room refuses and the old room opensIntegrator corrects the mapping, both authorities are reissuedThe move turnaroundA corrected mapping version and two door events
D4Duplicate authorityTwo credentials open one room, or one master opens a guest's roomProperty system owner withdraws the wider scopeUntil the next stay startsTwo authorities compared against the block's room list
D5Window mismatchAccess ends at the wrong hour in either directionProperty system owner, because entitlement is its truthThe window in the table aboveBoth systems' expiry values for the same authority
D6Identity mismatchThe card carries the old room number or a misspelled nameDesk and integrator, with the property system supplying the profileBefore the credential is handed overThe encoding record and the profile it was built from

Three observations make the register usable rather than decorative. First, D1, D4 and D5 share a direction of error that leaves a door more open than it should be, and D2, D3 and D6 leave it more closed; the two groups need different monitoring and different people on the other end of the phone. Second, only one row — D3 — is genuinely the integrator's to own, because only one row is caused by the artefact the integrator built. Third, every row has a row somewhere in a hotel's own systems that would have caught it earlier, and the reader who wants that catalogue in one place will find it in PMS Integration Failure Patterns and How to Recover.

Who reconciles, and how much authority the reconciliation carries

Reconciliation is not a single party. Four parties hold different parts of it, and a design that does not say so will discover the point during its first argument.

The property system vendor owns the emission of instructions and the truth about entitlement. The lock platform owner owns the application of an instruction to a door, the withdrawal of an authority, and the events that follow. The integrator owns the configuration between them — the room-to-door mapping, the window boundaries, the group hierarchy — and the health of the connector. The operator owns the business truth: whether the guest actually left, whether the room is fit to sell tonight, whether an override is permitted and who signs it.

The operator's part is the one most often assumed to be automatic. Sometimes reconciliation is a technical action: reissue the instruction, correct the mapping, withdraw the authority. Sometimes it is a decision. A guest who has left the room but is still on the property, a door under override during a leak, a room blocked for maintenance and therefore not sellable — in each case the software can only apply what the business decides, and the business decision has to be recorded somewhere that a reader three months later can find.

DriftWho meets the person at the doorWho changes the dataWho authorisesThe record produced
D1 orphan authorityNight manager or security, not the deskLock platform ownerOperator, on the departure recordA manual withdrawal with its reason and its authoriser
D2 missing authorityDesk, with the encoderProperty system ownerDesk supervisorA reissue instruction and the acknowledgement
D3 stale door attachmentDesk, then the corridor walk-roundIntegratorIntegrator, with the operator informedA new mapping version and a reissued authority
D4 duplicate authorityDeskProperty system ownerOperator, where a group is involvedA scope correction against the block's room list
D5 window mismatchWhoever meets the guestProperty system ownerOperator, where the stay is affectedTwo expiry values compared, then one
D6 identity mismatchDeskDesk, with the integrator for the templateDesk supervisorThe re-encoded credential and the discarded one

Two more practical points about who holds the authority. The front desk is an operational responder, not an authority: a desk override answers the person at the door, and it creates a record that must exist somewhere in the lock system's own history, because a manual release that leaves no trace is indistinguishable from a fault when it is read later. And the size and staffing of the property change the answer. The number of people who can authorise an override at three in the morning is a property fact, and it is not visible in any interface document. The broader commercial division of responsibility between supplier, integrator and operator is the subject of Who Owns PMS Integration: Supplier, Integrator or Client.

flowchart TD
    A["A disagreement is observed"] --> B["Classify it against the drift register"]
    B --> C{"Which side holds the newer truth?"}
    C -->|"Entitlement"| D["Property system owner reissues"]
    C -->|"Physical attachment"| E["Lock platform owner reapplies"]
    C -->|"Mapping"| F["Integrator corrects the mapping"]
    C -->|"Business fact"| G["Operator decides and authorises"]
    D --> H["Both sides read back the same state"]
    E --> H
    F --> H
    G --> H
    H --> I["The exception is written into the shift handover"]

The routing is the practical content of that table. Most disputes are not about who is right but about who was asked to change something, and a dispute at eleven at night is settled fastest when the party whose change will make the two sides agree was agreed in advance.

A move at ten past four in the afternoon

Here is a composite, and the most instructive sequence in a hotel PMS-to-lock integration because almost every step of it is something that can go wrong on its own.

A guest is checked in to room 412. At ten past four, housekeeping reports water damage in 412 and the room is withdrawn from sale. The operator moves the guest to room 407 and records the move in the property system. Four things then have to happen, in order: the property system withdraws the authority attached to 412; it issues an authority attached to 407; the desk re-encodes the guest's credential against the new room; and the old room begins to refuse the credential that used to open it.

Now suppose the second instruction does not arrive. This is the point of the whole exercise, because "does not arrive" is not one outcome. It is four. The instruction may be late, in which case the guest arrives at 407 at six and the door opens. It may be retried and delivered twice, in which case the lock side applies the same grant twice and the two applications disagree about the expiry. It may be queued on the property system's side, in which case it lands at an unpredictable point in the evening. Or it may be dropped, in which case the guest's credential opens 412 and the guest walks into a room that housekeeping has emptied.

The four outcomes have to be distinguishable from the outside, because the front desk's response is different for each. A late instruction is a wait. A duplicate is a duplicate-application defect and needs an engineering answer. A queued instruction is a wait with a longer horizon. A dropped instruction is a manual override and a record. The way to tell them apart is to observe the acknowledgement and the event stream, which is why the acknowledgement has to carry the business reference rather than only the transport reference.

sequenceDiagram
    participant PMS
    participant Connector
    participant LockPlatform
    participant Door
    participant Desk
    PMS->>Connector: withdraw authority for room 412
    Connector->>LockPlatform: withdraw authority for room 412
    LockPlatform->>Door: remove authority from door 412
    Door-->>LockPlatform: door event authority removed
    PMS->>Connector: issue authority for room 407
    Connector->>LockPlatform: issue authority for room 407
    LockPlatform->>Door: grant authority on door 407
    Door-->>LockPlatform: door event grant applied
    LockPlatform-->>PMS: acknowledgement with business reference
    Desk->>PMS: re-encode guest credential against room 407
    Note over Door: a guest presenting at 412 must now be refused

The same scenario contains the harder failure. Suppose the move instruction arrives perfectly and is applied to 407 — but the room-to-door mapping loaded in the connector still points 407 at a different physical door, because the corridor was renumbered after the survey. Nothing in either system is wrong. The authority exists, the grant applied, the acknowledgement returned, and the guest is standing in front of the wrong door. Only a physical walk-round, or a mapping version check against the current door schedule, finds it. This is the failure that belongs to the integrator and to nobody else, and it is the one that a message-monitoring dashboard will never report.

The way the property system expresses a room is not the way the door is identified. Room numbers are renumbered, floors are reassigned, and a wing is closed for refurbishment. The mapping is a versioned artefact with an owner, and a version change to it is a change to the access system even though no message crossed the interface that day.

Two clocks, and one record that has to survive the argument

Events with two clocks cannot be reconciled. A hotel integration usually has more than two: the property system server, the lock platform server, the controller on the floor and the device itself. Each stamps the moment it observed something, and those moments are different facts. When a dispute arrives at the door — was the guest inside when the withdrawal ran, or was the withdrawal late — somebody has to be able to say which number answers the question and which number merely records when the record was written.

The specification should name an authoritative clock for the record, state a drift allowance, and state what happens when the allowance is exceeded. The allowance is a decision the integrator and the operator make together, based on what the reconciliation has to survive. Ordering a check-in against a checkout needs a tight tolerance; establishing that a room was unoccupied at a given moment needs a tolerance that will withstand an argument, and the two are not the same number. Outside the allowance the event should be flagged rather than quietly accepted, because a flagged event is a question and an unflagged one is a number nobody will revisit.

The second requirement is that the record can be produced. A trail that exists but cannot be exported to somebody who was not present when the event happened is not a trail; it is a hope. That means the export format, the retention control and the reader list are part of the integration, not part of the operations team's preferences — and the duration of the retention is the operator's decision, made under whatever regime applies in the jurisdiction where the building stands.

CONFIRM-WITH-LOCAL-DATA-PROTECTION-REVIEW — whether records of who presented a credential at which door at which time attract a defined retention period at a hospitality property in this jurisdiction, who inside the operator may read them, and what notice is owed to guests. A door event log contains personal data about identified people, and the duration and the reader list are set locally rather than by the lock.

stateDiagram-v2
    [*] --> NoAuthority
    NoAuthority --> AuthorityPending: instruction received
    AuthorityPending --> AuthorityActive: grant applied at the door
    AuthorityPending --> InstructionLost: no acknowledgement in the window
    InstructionLost --> AuthorityPending: resent by the property system
    AuthorityActive --> AuthorityActive: replaced after a room change
    AuthorityActive --> AuthorityClosed: departure processed
    AuthorityActive --> AuthorityStale: withdrawal did not arrive
    AuthorityStale --> AuthorityClosed: reconciled by the platform owner
    AuthorityClosed --> [*]

Two states in that diagram are where the money and the arguments are. AuthorityPending is the interval in which the property system believes a guest is entitled and the door does not yet agree. AuthorityStale is the interval in which the door agrees and the property system does not. They have opposite consequences, and a single status light showing "in sync" cannot represent both of them, which is why the reconciliation duty has to be written down as a process and not left to a light on a wall.

When the link goes quiet

Three failures, three different doors and three different people to call.

The network between the two systems goes. The instruction stream may stop in either direction, and the door's behaviour depends on whether the arrangement is centralised or standalone, as set out earlier. What the specification must state is how long a pending instruction is held, who is told that it is being held, and what the front desk does in the meantime. A property that has told its staff "the system is offline, keys are issued manually and logged in the book" has designed its degraded mode. A property that has not will find out at the desk.

The controller goes. The door stops being a controlled door for as long as it lasts, and the question becomes which state it holds. Fail-safe and fail-secure describe which state a lock takes when power is removed, and the decision belongs to the door's role and to the approval for the building, not to the product and not to the integrator. A door on an escape route, a door protecting a server room and a door to a corridor of empty rooms can all reasonably answer differently. What the design must do is refuse to leave the question open: for each door class, the intended state on loss of power is stated, someone is named as the approver of that statement, and the behaviour is observed rather than assumed.

The server goes, or the connector between the two systems goes. Both systems may be perfectly healthy while the thing that ties them together is not. This is the failure that crosses an organisational boundary, and the specification should say who is notified, by what means, and within what window. A connector that fails silently for a night is more damaging than one that fails loudly, because the drift accumulates quietly and is discovered by a guest.

Two hardware signals sit close to this discussion and are often misdescribed. A door position contact reports open or closed; it reports a state, not a cause, and a log line that concludes the door was forced has inferred more than the contact knows. A request to exit is the indication an exit device sends when the door is worked from the escape side. That indication is not a release instruction: the door lets go on the escape side because of how the exit device is built, and the signal only reports that the operation took place. And the rule that governs the fire interface is that release is not the lock's decision; what a fire signal does to an opening is a matter of the fire strategy and the approval for that building.

CONFIRM-WITH-LOCAL-FIRE-AUTHORITY — which authority accepts the fire-alarm interface for each opening in this building, whether that interface is commissioned and tested as part of the fire system rather than the access system, and what evidence the commissioning must produce. A per-building and per-opening question.

CONFIRM-WITH-LOCAL-BUILDING-CONTROL-BODY — whether the release on the escape side and the state each opening takes when power is lost appear in the approved classification for the building, and which document is the record of that classification.

The reader-to-controller path is a separate seam inside the same project, with its own vocabulary. Wiegand names an older reader-to-controller wiring convention, the kind that runs a data line and usually brings a power line and a ground with it, and the name carries no promise about encryption or about telling one device from another. OSDP is a written specification with addressed, two-way messages and a defined structure, and whether any particular product implements it has to be confirmed for the model in question on the dated product page for the specific SKU. Either way the reader reports and the controller holds the door's state and makes the decision.

The handover that carries the truth between shifts

A hotel runs on handovers, and the reconciliation function has a natural home in one. The shift handover sheet is where the drift between the two systems becomes human knowledge rather than software state, and it is worth designing deliberately rather than accepting whatever the duty manager writes on the day.

The items that belong on it are specific. Rooms where an instruction was applied manually. Doors under override, with the reason and the person who authorised it. Credentials issued outside the normal path, including those issued to a guest whose instruction had not yet landed. Reservations that arrived while the connector was unavailable. Authorities extended by hand at the end of a shift. In a property with a well-run system the sheet is short and boring, and that is the design target — a long sheet is evidence that the integration is not doing its job.

The reason this matters more in a hotel than in most buildings is turnover. Staff at the desk are frequently new to the process, working a night shift, and handling a door they have never seen. The handover sheet is the only mechanism that carries the truth about exceptions from one set of hands to another. Where an exception is not written down, it does not survive the shift; it survives as a rumour, and the rumour becomes a false record the moment somebody writes it into a log.

This is also where the physical walk-round belongs. A small number of doors — one per floor, one on a shared controller, one carrying a recent manual override — checked physically against what the two systems say, is a cheap way to catch the correspondence failures that no interface test will find. It costs an hour of a duty manager's time and it catches the class of error that a register of known drift types cannot see, because both systems agree with each other and are both wrong about the building. The record the sheet and the walk-round produce is the same record the operator will be asked for later, and the format for building it is described in Documenting PMS Integration for Handover and Audit.

Putting the seam in writing before the door is drilled

The thing that reliably fails in hotel PMS integration is not the technology. It is the written record of what was actually agreed, which is why the questions below are asked before the hardware arrives rather than after the first complaint.

Some products in this category are described in the market in terms of consumer ecosystems and short-range radio standards — Matter, Thread, Zigbee, Z-Wave, Apple Home Key, HomeKit, UWB — and none of those names may enter a specification on the strength of a brochure. For every SKU in scope, confirm on the dated product page for the specific SKU and obtain the supplier's written statement naming that SKU and that interface before the row is treated as an available path for this property. Where a phone-held credential is offered as the answer to the room-to-door correspondence, the same discipline applies, because the credential is issued, personalised, delivered, returned or revoked by a different set of parties than a card handed across a desk.

What to demandAcceptable form of answerWhat disqualifies the answer
The interface typeA statement of local API or cloud API for the named platform, with the failure behaviour of each"We integrate with most systems"
Protocol implementation per SKUA written statement naming the SKU and the interface, datedA family-level statement that covers every model
Event and audit exportA field list, a format and a delivery method, with a sample exportA list of features described in prose
Credential encoding surfaceThe commands available at the desk and their error responsesA statement that the desk "writes" the card
Instruction identityThe business reference carried, and how a retry is recognisedAn incrementing transport counter alone
Degraded behaviourThe stated behaviour of the door and the instruction queue when the link failsSilence, or a statement that it is never offline
Update and rollbackWho may change the interface, when, and how a bad change is withdrawnA note that the supplier manages it

The first row does more work than the others: a local API and a cloud API place the failure boundary in different organisations, and the degraded mode follows from that single decision. Written answers of this kind belong in the requirements set, collected in PMS Integration Requirements Document Template, and the cases that prove the behaviour afterwards are set out in PMS Integration: The Test Plan a Buyer Should Demand.

Where this editorial stops

Some boundaries are worth stating plainly so that the article is used for what it is good at.

It makes no claim about what any product does. Every statement about a specific product's capability, interface, protocol support or conformity route is a question to be answered in writing, on a dated page, for the specific SKU. A design decision cannot be taken against an answer that has not arrived.

It answers nothing that belongs to the jurisdiction where the building stands. Retention periods, the treatment of door events as personal data, the fire interface, the escape classification and the operable force at a door are all decided locally, and this article carries them as questions with an owner rather than as answers it invented.

It does not describe a property, a guest, an operator or a contract. The scenarios are composites, written so that a reader can recognise a situation from their own building, and their inclusion is not a claim that any named hotel operates this way.

It does not replace the design work. The reconciliation function described here has to be specified against a real door schedule, a real desk procedure and a real night-shift staffing pattern, and none of those can be borrowed from an article.

The questions to put in the first workshop

If this article is used, it is used in a meeting with the property system vendor and the lock supplier in the room, before the mapping is loaded. The questions are below in the order that usually gets the fastest answers, and each one has a follow-up that exposes whether the answer is real.

  1. Which room-to-door mapping is in force today, who owns its version, and what happens when a room is renumbered?
  2. What is the authoritative clock for a door event, and what is the permitted drift between the two systems?
  3. For each event class — arrival, room change, extension, departure, group arrival — what is the permitted interval between the decision and its effect at the door?
  4. Which direction of error is dangerous in each of those classes, and who is told when it happens?
  5. What does the door do when the connector is unavailable, and what does the front desk do in the same ten minutes?
  6. Which state does each door class hold on loss of power, and who approved that state?
  7. When the two systems disagree, which one is asked to change its data, and who signs that decision?
  8. What is recorded when the desk overrides a door manually, and where does that record live?
  9. Which report answers the occupancy question, and does every figure in it name the system and the clock that produced it?
  10. Who holds the retention decision for door records, and where is that decision written down?

The tenth question is the one that tends to be deferred longest and regretted soonest, because door records are personal data about identified people and the period, the reader list and the handling of a request are all set by the operator's own position in the jurisdiction of the building. The earlier questions are engineering; that one is not, and it belongs to the operator from the beginning.

Where the written answer belongs next

The general shape of the seam between a property system and a lock system is set out in the Hotel Smart Lock PMS Integration Guide, and the room-to-door mapping in particular is treated in Hotel Smart Lock Room Assignment. Neither replaces the mapping you build for your own building. When the mapping, the clock and the reconciliation windows are written down, the commercial conversation belongs at Request a Quote — and it will be a shorter conversation, because the requirements are already specific.

ตะกร้าสินค้าของคุณ
การใช้คุกกี้จะช่วยปรับปรุงประสบการณ์การใช้งานเว็บไซต์นี้ของคุณให้ดียิ่งขึ้น นโยบายคุกกี้
Chat on WhatsApp