Pms integration effort drivers: A stage-by-stage account of what actually drives the cost and effort of a lock and property system integration, and what.

Đăng lúc: Th07 24, 2025

PMS Integration Cost and Effort Drivers Without Quoting Prices

PMS Integration Cost and Effort Drivers Without Quoting Prices

Most quotations for a lock and property-system integration fail long before they fail on money: nobody can say what the seam will ask of them. An integrator asked to estimate pms integration effort drivers is not estimating a product, because no lock decides what a property management system will send at ten past ten in the evening when a guest arrives without a reservation. The same holds for a hotel lock integration estimate: it follows the property management system and lock integration in front of it, not the hardware behind it, and the gap between two systems that somebody else maintains on a release cycle you do not control is measured in decisions, not in hardware.

Direct answer. To put it plainly: pms integration effort drivers has to hold for pms integration effort drivers. pms-integration has to hold at the same time. The sections below describe the order pms integration effort drivers is settled in. A pms integration effort drivers is an integration view. Lock, door, reader and system have to behave together. pms integration effort drivers is where the first interface behind pms integration effort drivers fails.

  • Draw the interface — write down what hands over between lock, door and system at pms integration effort drivers
  • Exercise the handover — record the sequence that proves pms-integration works end to end
  • Archive the record — store the commissioning evidence with the file for estimate counts seams not
  • This is the sequence pms integration effort drivers has to follow, anchored on pms integration.

This document walks that gap in the order an integrator meets it: discovery, mapping, transport, credential lifecycle, records, clocks, degraded mode, enrolment and handover. At each stage it names what consumes the working days, which question has to be answered before the answer can change the work, and who owns the answer. It carries no prices, no hardware counts and no schedules. It carries something more useful for a buyer comparing two bids: a way to make quotations comparable when neither party will publish a figure, and the confirmations that have to be made in writing before a figure can exist at all.

The governing idea is that every unit of integration effort traces back to a question somebody must answer, and every such question needs exactly one owner. Where the answer changes the size of the job, the answer belongs in the quotation as a stated basis rather than in a meeting as a shared understanding. Where the answer is not yet known, the honest line is a condition, a named confirmation and an owner, because an unstated assumption buried inside a price is the most expensive thing on the page.

The stages below are ordered so that each one removes a class of surprise. What survives to the end is a scope that both parties would describe the same way.

The estimate counts seams, not doors

The size of a lock integration is set by how many distinct decisions the seam forces and how many times each decision has to be made. Four counts move that number, and none of them is the number of doors in the building.

Door types, not doors. A property with four hundred doors across six types behaves like six integrations. Each new type — guest room, back-of-house corridor, service entrance, staff door, plant room, parking — brings a new mapping, a new access rule set and a new test. Ten identical guest-room doors are close to one piece of work once the first is built; one unusual door type can cost more than all the guest rooms together.

Credential paths. Card, mobile credential, biometric, mechanical override and time-boxed trade access are five separate lifecycles, each with its own issuing rule, expiry rule and revocation rule. A property that expects three of them to work together has a larger seam than a property expecting one, and the difference is not visible in the door schedule.

Event classes that must leave the lock. Every distinct event someone will later want to prove — arrival, refusal, forced opening, door held open, low battery, administrative override — is a field, a destination and a retention rule that somebody must define.

Organisations whose data is touched. A single-property operator with one property system and one lock platform is a two-party seam. A group with three property systems, a group-level directory and a customer-facing app is a multi-party seam, and the reconciliation work belongs to whichever party is left holding two versions of the truth.

The practical consequence is that an estimate is better expressed as a count of seam units than as a count of doors. A seam unit is one mapping, one rule set, one test case and one piece of documentation for one door type on one credential path. Count the units, name the assumptions behind the count, and the estimate becomes something a second integrator can produce in a comparable shape. The second take on pms integration effort drivers is Security Questions to Ask About PMS Integration. What can be supplied is listed in the smart door locks.

The stage map and the gate at each stop

The stages below are not chapters of an essay. Each one ends at a gate, and the gate is a document rather than a decision — a page that says what is now known, what is still open, and who owns each open item. A stage that ends without a gate has not finished, whatever the meeting looked like.

flowchart TD
    A["Stage one: discovery and seam definition"] --> B["Stage two: mapping rooms to doors"]
    B --> C["Stage three: choosing the interface boundary"]
    C --> D["Stage four: credential lifecycle"]
    D --> E["Stage five: records, clocks and retention"]
    E --> F["Stage six: degraded behaviour and approval"]
    F --> G["Stage seven: enrolment and decommissioning"]
    G --> H["Stage eight: handover and ownership"]
    H --> I["Quotation: packages, conditions, exclusions"]

The same sequence read as elapsed work, with the artefact each stop produces attached to it. The gaps between the stops are where the effort actually sits, and they are not evenly distributed: the long gaps are the mapping, the lifecycle and the approvals.

timeline
    title The seam is opened in a fixed order
    section Before an estimate can be written
        Discovery of the property system : written answers
        Definition of the seam : one page
    section While the build is under way
        Mapping of rooms to doors : field map
        Selection of the interface boundary : local or cloud
        Credential lifecycle agreed : issue, expiry, revoke
    section Before the handover
        Records and clocks settled : one authority
        Degraded behaviour approved : signed off
        Enrolment and handover : run book

A stage can be re-entered. A property system release, a change of door type or a change of policy sends the work back to an earlier stage, and the useful discipline is to know which stage it went back to, because the answer changes what has to be re-tested rather than only what has to be re-written.

Stage one: discovery, before any estimate is credible

The first stage rarely produces hardware decisions. It produces answers, or the discovery that the answers are not available. Either way it changes the estimate, which is why an estimate issued before this stage is finished is a guess with a number attached.

The dangerous discovery is not a bad answer. It is an answer nobody owns. Access to the property system's documented interface is a commercial matter held by the property system vendor and the client's own team; if nobody has asked, the integration has no starting point and the first line of effort becomes procurement rather than engineering.

Question the integrator must have answeredWhy the answer moves the estimateWhere the answer comes from
Which property system product and release is in scopeA release change can alter field names, endpoints and entitlementsVendor documentation and the client's own upgrade record
Whether a documented interface entitlement exists for that releaseAn entitlement that has not been bought turns engineering into negotiationThe property system vendor's commercial desk
Where the interface terminatesA local interface and a hosted one produce different network, hosting and support workThe client's IT architecture
Which operations must be real time and which may be scheduledA batch expectation moves work into scheduling and reconciliation designThe operations lead, in their own words
Who is permitted to change the property systemA change freeze turns the integration into a queue positionThe client's change process
Whether the client already holds a door-to-room mapAn unreliable map becomes a survey task that has to be verifiedThe client's property or technical services team
What the lock platform is expected to send backInbound-only scopes and two-way scopes differ by a wide margin in testingBoth parties in writing

Two of these answers have to come from the property system vendor directly, and neither should be taken on a verbal report from the client's operations staff. A vendor's field list and a client's understanding of that field list diverge, and the divergence usually surfaces at the first test rather than at the quotation stage.

The written output of this stage is short: a one-page seam definition naming the two systems, the direction of traffic, the party that hosts the interface, and the party that owns each open question. Everything after it references that page.

Stage two: two vocabularies and one door list

The property system speaks about stays, rooms, guests, reservations and folios. The lock side speaks about doors, credentials, identities, access rights and controllers. Neither side has the other's nouns, and the translation is the first real piece of engineering work in the integration.

The first trap is assuming that a room number is a door identifier. In most properties it is not. The mapping between a room code and a physical door is held in a spreadsheet, a property system field or the head of security's memory, and the three do not always agree. An integration that assumes the room code is the door code will appear to work perfectly in a test on one floor and fail on the floor where the numbering changed during a refurbishment. Mapping is therefore a survey task, and the survey is part of the estimate.

The second trap is treating an identifier as a credential. An identifier in the property system names a person or a role. A credential is the thing that person presents. Personalisation is the per-holder data carried on that credential. The three travel together but they are not interchangeable, and the seam has to carry each one explicitly.

Property system termLock-side termWhat breaks when the pair is left implicit
Room numberDoor identifierAccess is granted to the wrong door after any renumbering
Guest or staff identifierIdentityRevocation is applied to a record that the credential never referenced
Key or card serialCredentialA replaced card keeps the old holder's access profile
Stay statusAccess rightA late checkout or an early arrival is not reflected at the door
Departure dateCredential expiryAn expired stay stays open until somebody cancels it manually
Reservation note or preferencePersonalisation fieldData that is not on the credential cannot be shown to the person at the door

The third trap is the status enumeration. Pre-arrival, checked in, checked out, late checkout, no-show, cancelled and rebooked each carry a different access consequence, and each consequence has to be written down rather than implied. Every status the property system can emit needs one line in the rule table: what the lock is told, and what it does with it.

The effort driver here is arithmetic rather than difficulty. Mapping work scales with door types multiplied by status transitions, not with door count. Once the first door type of each shape is built and tested, subsequent instances of the same type are configuration. A client who understands this will price the estate by type; a client who does not will price it by door and be surprised twice.

Stage three: the transport, and the two paths people confuse

Two separate data paths run through a building with an integrated lock, and estimates routinely mix them.

The first path runs from the property system to the lock platform: reservations and access changes travel along it. The second path runs from the reader at the door to the controller at the door: presented credentials travel along it. A property system integration does not touch the second path at all, and a reader-to-controller decision does not resolve any of the first path's questions. Keeping them apart is what stops a specification from becoming a paragraph that satisfies nobody.

On the reader-to-controller path, two conventions come up. Wiegand is a legacy wiring convention between a reader and a controller, carrying a data line and usually a power and a ground, and it says nothing about encryption or device identity. OSDP is a specified, addressed, bidirectional protocol with device addressing and a defined message structure. Whether a given lock or controller implements either one is a per-SKU question: confirm on the dated product page for the specific SKU, and write the answer into the door schedule rather than a conversation.

The path a phone credential takes to reach a door is a third question, and it is the one most often answered with a marketing line instead of a written statement. The names that come up in that conversation are Matter, Thread, Zigbee, Z-Wave, Apple Home Key, HomeKit and UWB. They are ecosystem and protocol names, and this document assigns none of them to a particular lock. The integrator's task is to convert a marketing line into a written statement: ask the supplier to confirm, for the exact SKU, what role each named technology plays in the credential path — whether it carries the credential to the door, whether it carries only management traffic, or whether it has no part in the door at all — and ask for the same answer on the dated product page for the specific SKU and in the written response to the enquiry. A supplier who answers in general terms has not answered the question the estimate depends on, so the estimate carries it as a condition rather than as an assumption.

For the property system path, the decision that moves the estimate is where the interface boundary sits — a local API against the lock platform, a cloud API, a scheduled file drop, or a middleware process that translates in both directions.

Boundary choiceWhat it adds to the estimateWhat it takes away
Local APIOn-site hosting, interface service, client network changes, a support path for the clientA dependency on the client's own infrastructure and its availability
Cloud APIA single integration service, fewer client-side componentsA network and outbound dependency on a hosted service and a subscription arrangement
Scheduled file dropSimple transport, easy to audit and to re-runReal-time behaviour, and the reconciliation logic that goes with it
Middleware in betweenMapping once, and a place to hold queues and retriesA component that somebody has to run, monitor and eventually replace

Whichever boundary is chosen, the credential and event questions still have to be answered. A local API does not solve the de-authorisation turnaround; a cloud API does not solve the clock. The boundary also decides who is exposed when the other side ships a release: behind a middleware layer, a property system release breaks one component that can be tested in isolation; written directly against the property system's interface, it breaks the client-facing service at the hour the desk calls.

Stage four: issuing, expiry and the credential lifecycle

The credential lifecycle is where the property system earns its keep and where most of the invisible work sits. Four questions set its size: who issues the credential, what triggers each lifecycle operation, how quickly expiry takes effect, and what happens to a credential that is never returned.

Issuing can happen at the desk with an encoder, on a central server that pushes a card profile out, or through a mobile credential delivered to a phone. Each is a different build with a different support burden. Desk encoding moves work to the front desk and puts the encoder on the critical path of every arrival. Central issuing moves work to the server and into the network. Mobile delivery moves work into the handover path between desk and guest, and into the recovery path when a phone is lost.

The operations themselves are the work: issue, re-issue, extend, cancel, de-authorise, return and replace. For each one, the estimate has to say whether the property system triggers it or a person performs it. Manual cancellation of a departing guest's access is a staffing decision that the integration surfaces as a recurring cost, and it is invisible until somebody counts it.

The de-authorisation turnaround deserves its own line. How soon after checkout must the credential stop working — instantly, at a scheduled hour, or at the end of the stay as recorded? The answer is a property policy, and the policy chooses the design. Immediate revocation needs the door to reach a service that can be asked. A scheduled expiry can be carried by the lock itself and keeps working when the network is down. The design consequence is large, and it is invisible in a feature comparison.

Personalisation travels with the credential and raises its own question: which fields, in which format, in which language, and how many of them are actually used at the door. Every field carried is a field that must be mapped, validated, stored and — if it identifies a person — retained under a rule somebody has to set.

Stage five: the record, the buffer and the clock

Three different objects get confused in an integration specification, and each one behaves differently when something fails.

The audit log is the settled store of events, each one carrying the moment it happened and the place it came from. The event memory is the working buffer a lock holds on the device for the stretches when it cannot get through to the server. They are not the same store, and the distance between them is exactly where events quietly disappear. An integration that only specifies the audit log has not specified the offline case at all.

The clock is the second object. The lock, the controller, the property system and the reporting server each keep their own time, and each is set by somebody. Events with two clocks cannot be reconciled, and the reconciliation is done by hand, at cost, usually by the person with the least authority in the project. The questions are short: who owns the authoritative time, what is the maximum accepted difference between the door and the server, and what happens to an event that arrives with a timestamp outside that difference — held, corrected, or discarded?

The record itself needs an owner too. What is written down, where does it live, and who can read it? If nobody can answer, the integration has not been specified. In practice this splits three ways: the lock platform holds the access event, the property system holds the stay record, and the client holds the reporting copy. Each copy can have a different retention period, and a record can outlive the business reason that produced it.

The chain below is the minimum set of objects a pms integration effort drivers estimate has to account for, because each one has its own owner, its own retention rule and its own failure mode. Where a copy exists but nobody maintains it, that copy is where a dispute about an event ends up.

erDiagram
    STAY ||--o{ CREDENTIAL : issues
    STAY {
        string stay_id
        string room_code
    }
    CREDENTIAL {
        string credential_id
        string identity_ref
    }
    DOOR_EVENT {
        string event_id
        string source
        string timestamp
    }
    CREDENTIAL ||--o{ DOOR_EVENT : produces
    DOOR_EVENT }o--|| REPORT_COPY : copied_into
    REPORT_COPY {
        string retention_rule
        string reader_list
    }

This is where personal data enters the record, so it is written here as a question rather than an answer. A record of who opened which door and when is personal data in most jurisdictions, and the retention period, the lawful basis, the access list and the data subject route are decided locally, not by a lock vendor or an integrator.

CONFIRM-WITH-LOCAL-PRIVACY-ADVISOR: who sets the retention period for access records, which authority must be asked about the lawful basis for collecting them, and what must a resident, guest or employee be told about how their movements are recorded?

Stage six: the degraded mode is the design, not the exception

Every integration has a degraded mode and it has to be named. Four separate failures have to be considered, and they are not the same failure.

When the network is unavailable, the door has to keep a decision. When the controller is unavailable, the door has to keep a decision and to say what it kept. When the server is unavailable, the event has to be buffered somewhere and delivered later. When the property system is unavailable, the last known state applies and somebody has to decide whether that is acceptable.

Centralised and standalone architectures answer these differently, and the choice is architectural rather than a feature. Where the lock's state lives on a server, a network failure has consequences at the door. Where the state lives on the device, the door keeps working and the property system loses sight of it until the link returns. Neither is right in the abstract; each has to be chosen per door role, and the choice is recorded per door in the door schedule.

What state the lock enters once the supply is cut is a property of the door's role and of the sign-off that was obtained, not a property of the product. Fail-safe and fail-secure are simply the two names for that outcome, and the choice between them belongs to the person who owns the risk. The important question is not the electrical behaviour but the operational consequence: a corridor that ends up locked during an evacuation is an approval question, not an engineering one, and a door that ends up unlocked in a secure area is a different approval question.

Two signal conventions belong in this stage because they are routinely misread in specifications. A door position contact reports a state, not a cause; it says the door is open and it says nothing about who opened it or why. A request to exit signal is not a release command. Release on the escape side is the default behaviour of the exit device, and the request-to-exit signal tells the system that the exit device was operated — it asks the system to record something, not to unlock something. Any specification that uses a request-to-exit output to drive a lock release is describing a different design, and one that must be checked against the egress assessment for that door.

The fire interface sits in the same family and carries the same discipline: release on a fire signal is not the lock's decision to take.

CONFIRM-WITH-LOCAL-FIRE-AUTHORITY: which authority signs off the fire-alarm interface for each door category, and what is the documented release sequence for the doors in this property?

Stage seven: enrolment, decommissioning and the address book

Provisioning means bringing a device into service with its identity, its door and its owner. Decommissioning means removing it from service, and it is a separate, recorded operation — not the undo of provisioning, and not free.

The enrolment effort is driven by how a device finds its own address. Where the platform knows which device sits on which door because the installer wrote it down, the work is simple and hard to audit later. Where the device announces itself and the binding between device, door and room is made afterwards, the estate scales and the audit gets harder. Either way somebody maintains the address book: the property team after a renumbering, the installer after a swap, the platform owner when a group consolidates sites.

The estimate should count three things here that are usually missing: the initial binding work per site, the recurring maintenance of that binding, and the decommissioning at the end. That last one is the item most often priced at zero, and it is the one that appears in the client's operating budget rather than the project's.

Access for trade contractors is the same lifecycle with a short life. A credential issued to a contractor for a defined period has to expire on its own, and the person who closes the job must not have to remember to cancel it.

Stage eight: handover, first-line support and the boundary

The last stage is where an integration either holds or becomes a phone call that lasts a year. The fifth question every integration has to answer is which side owns the fix, and Who Owns PMS Integration: Supplier, Integrator or Client sets out that boundary in full.

An ownership boundary is drawn around three things: who can change the property system's mapping, who can change the lock platform's rules, and who holds the environment in between — the middleware, the interface service, the certificate, the credentials that let two systems talk to each other.

Where that environment is a third party's, the support contract has to say what happens when that party is slow, because the client experiences the delay and the integrator absorbs the escalation. Where it is the integrator's, the handover has to carry the operational knowledge to run it, which is a documentation cost that belongs in the estimate rather than in goodwill.

First-line support needs a triage table. A technician at the door can decide some things without escalation — a dead battery, a mechanical fault, a credential that was never issued, a card that was re-encoded — and must escalate others immediately: a door that reports the wrong state, a credential that works on one door and not another, an event that is missing from the record, a lock that has lost its network path. The table names, for each symptom, the first check, the escalation trigger and the owning party. Handover is also the moment the clock question becomes visible: if the reporting depends on buffered events, the handover has to say who watches the buffer and who tells the client that a gap exists.

The driver register: what each driver changes and who answers

The register below is the working list behind a pms integration effort drivers estimate. Each row names what the driver changes, how it is measured before the quotation is written, and who has to answer. A driver with no answer is not a small driver; it is an unpriced one.

Not every driver deserves the same attention. The two that matter most are the ones that are expensive and hard to see, because they arrive late and force rework in several stages at once. The plot below places the familiar drivers on effort against visibility, and it is worth noting that the visible-but-expensive drivers are usually scheduled for and the invisible-but-cheap ones are usually skipped.

quadrantChart
    x-axis Effort added low --> Effort added high
    y-axis Visibility low --> Visibility high
    quadrant-1 Large and visible
    quadrant-2 Large and hidden
    quadrant-3 Small and hidden
    quadrant-4 Small and visible
    "Door-to-room map accuracy": [0.78, 0.72]
    "Property system release drift": [0.62, 0.28]
    "Credential path count": [0.66, 0.55]
    "Radio protocol confirmation": [0.18, 0.12]
    "Retention rule change": [0.4, 0.3]
    "Degraded behaviour approval": [0.52, 0.86]
    "Ownership boundary": [0.34, 0.9]
    "Clock disagreement": [0.46, 0.22]
DriverWhat it changesHow it is measured beforehandWho answersWhat the quotation must state
Property system product and releaseInterface surface and field namesNamed release and the vendor's interface documentProperty system vendorProduct, release and interface behind the estimate
Interface entitlementWhether the work starts or waitsWritten confirmation of entitlementProperty system vendor, clientEntitlement carried as a client condition
Number of door typesMapping, rules and test volumeDoor schedule grouped by typeClient technical servicesDoor types and the per-type unit of work
Door-to-room map accuracyWhether a survey is neededSample check against the physical doorClient property teamSurvey duty and who signs the map
Credential paths in scopeIssuing, expiry and revocation workCount of paths with their rulesClient operationsEach path, its issuing point and expiry rule
De-authorisation turnaroundWhether the lock must be online to revokePolicy for checkout and departureClient operationsTurnaround rule and the design it implies
Interface boundaryHosting, network, support, release exposureChosen boundary and its ownerClient IT, integratorBoundary, host and the component to monitor
Event classes requiredField, destination and retention per eventEvent list agreed with the reporting ownerClient reporting ownerEvent list and the destination for each
Clock authorityReconciliation effort and disputesNamed owner and accepted differenceClient ITAuthoritative clock and tolerance
Retention and access to recordsPrivacy exposure and storage workRetention rule and access listClient legal adviserRetention basis as a client confirmation
Degraded behaviour per door typeHardware, configuration, approvalDoor-by-door role and approved stateClient risk ownerApproved state per door category
Fire and egress interfaceThe design of the release pathInterface list per door categoryClient fire strategy authorInterface and the approving authority
Firmware and platform releasesWho patches, when, and recoverySupport model and release channelLock platform supplierRelease policy and recovery path
Enrolment and binding methodInitial and recurring setupBinding method and its maintainerLock platform supplierBinding work per site and its upkeep
DecommissioningWork at the end of a site's lifePlanned removals over the contractClient, integratorDecommissioning as a separate operation
Ownership of the fixSupport cost and escalation timeSupport matrix with named partiesClient, integrator, supplierEscalation path and first-line scope
Radio path statementHow the credential reaches the doorWritten role for each named technologyLock platform supplierRole of each named technology, per SKU

Rows with a client answer in the last column are conditions. A condition in a quotation is honest; a silent assumption in a quotation is a dispute waiting for a date.

What a quotation must state before two bids

Two quotations for the same property can differ by a wide margin and describe the same work, or differ slightly and describe completely different work. The clauses below are the ones that make the difference visible without disclosing a price, and the PMS Integration Requirements Document Template turns them into a specification that can be issued.

Clause the quotation should carryThe question it answers
Named systems and releasesWhat was the estimate written against
The interface boundary and its ownerWho runs the piece between the two systems
The seam definition pageWhere the two systems meet and what crosses it
Door types in scope, with the per-type unitWhat a unit of work buys
Credential paths in scopeHow many lifecycles the estimate covers
Event list and destinationWhat leaves the lock and where it lands
Clock authority and toleranceHow a disputed timestamp is handled
Approved degraded behaviour per door categoryWhat happens when a link drops
Conditions and open questions, with ownersWhat is not yet known and who will say
ExclusionsWhat was deliberately left out
Support boundary and first-line scopeWho is called, and by whom
Deliverables and acceptanceHow the buyer knows the work is finished
Change mechanismHow a change in the property system is priced

The exclusions line is the one that saves the most time later. An estimate that does not say what it leaves out is a claim that it covers everything, and every exclusion discovered on site is a variation.

The five questions, written the way the integrator asks them

Every integration article has to answer five questions, and an estimate has to answer them as well. They are written here as they should be put to the parties on either side of the seam.

What does each side send, and what happens when it does not arrive? Ask for the message list, not the feature list, and then ask what the door does for each message that never arrives. The answer has to name a degraded mode. "It retries" is not an answer; "the lock uses its last known state and the controller buffers" is.

Who owns the clock? Ask which device is authoritative for the timestamp on an access event, what the accepted difference is, and who reconciles an event that falls outside it. Events with two clocks cannot be reconciled later by an argument.

What is written down, where does it live, and who can read it? Ask for the record list, the system of record for each, the retention period and the access list. If nobody can answer, the integration has not been specified, and the quotation cannot be compared.

What happens at the moment the network, the controller or the server is unavailable? Ask for the answer per door category, not per product. If the answer is that the door fails locked, that is an approval question to be signed by the person who owns the risk, and it should appear in the quotation as a confirmation rather than a design note.

Which side owns the fix? Ask who changes what on each side, who holds the environment in between, and what happens when the environment is a third party's. A boundary that is not drawn becomes a phone call that lasts a year.

Two further questions belong to the same list because they change the size of the work rather than the shape of it: how long a credential stays valid after the stay it belongs to has ended, and how many credential paths the property expects in the first year of operation.

Three illustrative situations

The following are illustrations of how the same driver set produces different shapes of estimate. They are not accounts of projects and carry no figures from any customer.

An illustrative small hotel with a standalone lock system. The property system is on the operator's own server and the locks hold their own state, with a card encoder at the desk. The door types are few and the mapping is a spreadsheet. The effort sits in the survey and in the de-authorisation rule: because the locks work offline, the natural design carries the expiry on the credential itself, which drops a network dependency and moves policy into the card-encoding step. The estimate is shaped by the map, not by the interface.

An illustrative serviced-apartment operator moving to a hosted property system mid-contract. The operator has running locks in several buildings and a property system that is changing. The driver here is the migration, not the connection: a credential issued before the change must still be understood by the platform that receives it afterwards, and the address book has to survive a renumbering. The estimate grows at handover rather than at build, which is why a quote written for the steady state is the wrong quote.

An illustrative group with three properties running two different lock architectures. One property is a hotel with cards at the desk; another a residential block with mobile credentials; the third a serviced property in another country with a different property system and a different privacy regime. The driver is the number of parties in the seam and the number of record copies each keeps. The estimate is dominated by reconciliation and by the support boundary, because the group is the only party that sees all three.

In each illustration the seam is the same shape. What changes is how many door types, credential paths, event classes and organisations are in it — which is exactly what an estimate should be counting.

Presenting an estimate without publishing a price

An integrator working under a no-price constraint does not have less to say. The constraint removes the number and leaves the basis, and the basis is what a buyer comparing bids actually needs.

Work the estimate in packages rather than in hours or currency. A package is a named outcome — the door map, the field mapping, the issuing path, the event route, the degraded-mode configuration, the handover pack — and each package carries its assumptions, its preconditions and its exclusions. A buyer can read a package list and judge whether the scope is right; a buyer cannot read a bare figure and judge anything.

Inside each package, state the drivers that would change it and the size of the change, expressed as a proportion of the package rather than as a sum. The door map package moves when the room-to-door map is unreliable; the issuing package moves when a second credential path is added; the event package moves when a new event class is required.

The conditions list travels with the packages. Each condition is a question, an owner and the consequence of an adverse answer, expressed in scope terms: "if the property system does not publish a documented interface for the named release, the translation work is re-planned as a file-based exchange." The client can then decide whether to buy the answer, find it elsewhere, or accept the changed scope.

Two disciplines make the difference between an estimate that survives a tender and one that does not. First, name what the estimate excludes, because an estimate that claims completeness is guaranteed to discover an exclusion on site. Second, state which parts are the integrator's own assumptions and which are the client's answers, so that a later change of assumption is visibly a change of assumption and not a disagreement about scope.

Where this document stops

The seam has been walked once, stage by stage, and each stage has produced the same output: a question, an owner, and a line in a quotation. What has been left out is any figure, because figures belong to a specific property, a specific property system release and a specific set of answers.

The next documents take that output and turn it into other things: a test plan that proves the seam behaves as written, a catalogue of what goes wrong and how it is recovered — see PMS Integration Failure Patterns and How to Recover — a requirements template that captures the conditions as clauses, and a handover record that lets a building be maintained by more than one team. The published library covers the surrounding ground in more detail — the seam itself in Hotel Smart Lock PMS Integration Guide, the interface boundary in Hotel Smart Lock API Integration Guide, the architecture choice in Standalone vs Centralized Hotel Lock Systems: Which Architecture Fits Your Property, and the record side in Hotel Smart Lock Audit Log Management.

If a quotation needs a basis for a lock side to answer in writing rather than in a meeting, the supplier side of the seam starts with Smart Lock System Integrator Procurement Guide and continues with PMS Integration: The Test Plan a Buyer Should Demand. To put an integration estimate in front of a supplier, request a quote with the seam definition page and the door type list attached — the answer that comes back is then comparable with another answer, which is the purpose of writing the seam down first.

Giỏ hàng của bạn
Trải nghiệm của bạn trên trang web này sẽ được cải thiện bằng cách cho phép cookie Chính sách cookie
Chat on WhatsApp