The RMA Process Design Site Record Sheet
What follows covers rma process design worksheet as a maintenance and measurement question, not a hardware question. A lock that has failed on a door is not a return yet. It is a fault report, and between the two sits a decision that belongs to the facility manager: does this unit go back to the factory, or is it fixed, replaced or written off on site? The decision is cheap to get wrong in one direction and expensive in the other, because a return costs freight, a door out of service, and a case that may arrive with nothing to diagnose.
Direct answer. This article is the instrument for making that decision well. The rma process design worksheet set out below is a fill-in record sheet, one page per fault case, capturing the identity of the unit, the symptom as the technician described it, the checks run at the door and what each of them ruled out, the decision with the role that made it, and the action the case closes on. Designed as a process than as a form, it answers three questions in a single pass: what evidence must exist before a unit is authorised to leave a door, who may authorise it, and what happens to the record once the case closes
The document is written as an annotated dialogue between two working voices, the Facilities Lead who owns the site record and the Supplier Support Engineer who receives the return. Every turn is followed by a margin note that converts the exchange into an action the reader can take. The transcript is illustrative: neither voice is a named individual, and no project, building, operator or client is depicted.
A return is a measurement, not a shipment
A unit that comes back from a door is a sample. What makes it worth reading is the record that travels with it: the identity of the hardware, the symptom as observed, the checks run before the decision was taken, and what each of those checks eliminated. Strip all four and the receiving desk is holding an object rather than a case.
That is the working definition of the return-authorisation route worth designing around. An RMA is what is authorised, what is shipped and what comes back. Notice what sits outside it: the freight arrangement, the insurance position and who carries them. A returns process that leaves those unassigned discovers them on the first difficult claim, in the middle of a disagreement, with a door still locked.
So the design has three moving parts, and a programme can get one right while the other two fail:
- The authorisation gate, meaning the minimum evidence that must exist before a unit is allowed to leave a door.
- The decision rights, meaning who may authorise a return, who may accept it, who may close the case, and what happens when the person who normally decides is unavailable.
- The closure loop, meaning what each closed case produces, so that the next occurrence of the same fault is a data point rather than a surprise.
Most returns arrangements have the first of these and little of the others, which is how a programme can be diligent about raising cases and still learn nothing from a year of them.
The two roles and the authority each one holds
The exchange runs between two roles on opposite sides of the returns desk.
| Voice | Stands for | Answers well | Cannot answer |
|---|---|---|---|
| Facilities Lead | The facility manager who owns the door schedule, the site fault log and the maintenance budget | Which door, what the occupants need, what the technician saw, which decisions are theirs | Whether the hardware is repairable, what the failure mode is, whether a group of cases points at one cause |
| Supplier Support Engineer | The technical support engineer who receives the return and works the case | What a returned unit is examined against, what a case file needs to be worked, what the catalogue records about a model | What the door felt like, whether the site power is stable, whether the case is covered, and what any of it costs |
Neither voice asserts a capability, a timescale or a commercial term on the other's behalf. Where the answer depends on the agreement, the sentence in the transcript is a question the buyer asks and the evidence the buyer demands, not a promise in either direction.
Turn 1 — The note that said only "door not working"
Facilities Lead: The note I received this morning said the door does not open. Three words and a door number. My technician went out, found the handle turning but nothing else happening, and the case was raised as a return. The unit is on the loading dock now with a form I did not write.
Supplier Support Engineer: Then you have a fault report and a return request, and they are not the same document. The report is an event. The request is a package: the identity of the unit, the symptom in the words of the technician, the checks already run and what each one ruled out, and a decision with a role against it.
Facilities Lead: The technician ran one check. He turned the handle.
Supplier Support Engineer: That is a check. What it ruled out is the question. A check that eliminates nothing is a repairable case rather than a diagnosed one, and that difference decides whether the return is worth making at all.
Margin note — buyer action: Split intake into two artefacts. The site fault log captures the event and the state of the door. The return request is a separate document that cannot be raised until the identity block and the check record are attached, and the person raising it is recorded by role.
Margin note — buyer action: Put one sentence at the head of the intake form: no unit leaves a door until the record states which checks were run and what each one ruled out.
The gate exists because a return carrying nothing to diagnose is the most expensive line in the subject. It consumes freight, a carton and handling, and it takes the door's history off site while the door waits for something to put on it. The unit arrives at the receiving desk with a serial number and no symptom, and the only honest conclusion available there is that no fault could be determined. That is not a diagnosis, it is an absence of evidence, and it is the commonest reason a returns programme produces no improvement across a year of activity. No test at the far end recovers the symptom, the moment it happened or the state of the door at the time. That evidence exists in one place only, and the place is the door it was collected from.
The intake sheet therefore needs four fields before a case can be opened, and every one of them is cheap to capture while the technician is at the door.
| Field | What it records | Why the gate exists |
|---|---|---|
| Door or unit identifier | The identifier from the object schedule, not the room number | A room number changes when a room is re-let; a schedule identifier does not |
| Reported date and time | When the fault was reported and through which channel | It anchors the interval you will compute later from your own records |
| Symptom in the technician's words | The free-text description, unedited | It is the raw material for the symptom vocabulary used later |
| Date and time the door was operable again | Filled when the door is covered, by swap or by repair | It is the second half of the interval pair; without it, this door contributes to no measure |
The last field is the one most often left blank, because the door was covered in a hurry and the form lives in a different system from the keys. A case that closes without it contributes to no duration measure you will later want to compute, and it cannot be reconstructed afterwards.
Turn 2 — The symptom, the observation, and the conditions
Facilities Lead: The technician writes "the handle turns but the motor does nothing". Is that the symptom?
Supplier Support Engineer: That is the symptom, and it is the best kind, because it describes what a person observed rather than naming a component. Keep the sentence exactly as written and put it in a field called symptom. Do not let anyone tidy it into "motor fault", because "motor fault" is a conclusion, and conclusions are what the case is trying to avoid.
Facilities Lead: Then what goes beside it?
Supplier Support Engineer: An observation. What was measured, under what conditions, and what came back. A reading taken at the compartment contacts under load, a check of whether the override moved, the state of the door on arrival. An observation can be re-checked at the far end and a conclusion cannot.
Facilities Lead: What if the technician could not take a reading?
Supplier Support Engineer: The form says "not performed" and why. A blank is read as carelessness. "Not performed, no instrument carried" is read as information, and honest information is worth more than a tidy form.
Margin note — buyer action: Split the symptom into three fields rather than one. Symptom in the technician's words, observation with conditions, and readings taken. A single free-text box collapses all three and destroys the comparison you will want at review.
The pairing works because each half answers a different question. The symptom tells the receiving desk what the door was doing. The observation tells it what the door looked like when somebody looked at it. Neither substitutes for the other.
One more line belongs in every observation, and it is the line teams argue about. Write the conditions: the door closed and latched, the handle released from the rest position, the room unoccupied, the unit in service for a stated number of weeks. A symptom recorded without conditions cannot be reproduced, and a symptom that cannot be reproduced cannot be compared against the next case of the same shape.
Turn 3 — Where a return becomes defensible
Facilities Lead: How long should a technician be at the door before a return is sensible?
Supplier Support Engineer: Long enough to run the sequence, not longer. The sequence is short, every step rules something out, and that is the only reason to run it in order rather than in whatever order the technician happens to prefer.
Facilities Lead: And the sequence is the same for every fault?
Supplier Support Engineer: For the fault that generates most returns, yes. It is a method rather than a diagnosis. The technician runs the steps, each one eliminates a family of causes, and the case stops at the first step that fits. A unit returned before the sequence has been run turns a diagnosis into a freight movement.
Facilities Lead: What is the cluster question?
Supplier Support Engineer: Whether the same fault appears on a second door of the same model and variant. If it does, the fault is no longer a property of one unit. It points at a configuration, a firmware baseline or a supply batch, and the correct response is to stop replacing units and start asking a different question.
The cluster question is the step that separates a repair from a mass replacement, and it is the one most often left unanswered because it needs something the technician does not have to hand: a list of the other doors carrying the same model. A single-door site has to answer it too, and the honest answer is that no second door of that model is installed. That answer is itself a finding, because it removes the possibility of a pattern check.
Margin note — buyer action: The authorisation gate asks one question about the door-side record. Is the sequence attached, was it run in order, and was the cluster question answered. It does not ask the technician to re-diagnose the fault on the form.
| What the gate rejects | Why | What the case needs first |
|---|---|---|
| A symptom with no observation | A description and nothing to check it against | The observation and its conditions |
| A check record with no ruled-out column | A list of actions is not evidence for anyone who was not at the door | The record showing what each check eliminated |
| A sequence that stopped early | The sequence is ordered, and an early stop leaves whole families of cause open | The sequence run, or the reason it stopped |
| An unanswered cluster question | The case may be one unit of a pattern, and a return treats it as a single fault | The second door identifier, or a statement that no comparable door exists |
| An identity block without a variant | A family may carry several versions, and they do not behave alike | The label read and dated, and the version it names |
A check that ruled nothing out is still worth writing down, because it tells the receiving desk which families of cause have been eliminated. A case naming only the checks which produced something has quietly hidden its gaps.
flowchart TD
A[Fault reported against a door] --> B{Intake record complete}
B -->|No| B1[Hold the case at the intake desk]
B1 --> B
B -->|Yes| C{Door-side check record attached}
C -->|No| C1[Send the technician back to the door]
C1 --> C
C -->|Yes| D{Cluster question answered}
D -->|No| D1[Name the second door or record that none exists]
D1 --> D
D -->|Yes| E{Identity block pins model and variant}
E -->|No| E1[Photograph the label and date the catalogue read]
E1 --> E
E -->|Yes| F{Authorisation signed by the named role}
F -->|No| F1[Hold the unit on site and log the reason]
F -->|Yes| G[Pick the branch]
G --> H[Repair at the door]
G --> I[Return the unit for repair]
G --> J[Replace from spares and retain the failed unit]
G --> K[Write off and record the loss]
I --> L[Record both return leg timestamps]
J --> L
K --> L
L --> M[Close the case with exactly one action]
Turn 4 — Pinning the identity of the unit that travels
Facilities Lead: The form asks for model and variant. The technician photographs a label. What else do we need?
Supplier Support Engineer: The date you read the label, and the record you read it against. The catalogue is a dated read of our own product records rather than a live system, and it moves when a model is revised or a variant is added. Write the date beside the model, and keep the copy you worked from.
Facilities Lead: The catalogue lists a lot of fields. What if the one I need is blank?
Supplier Support Engineer: Then the catalogue does not record that capability for that model, and that is the finding. An empty field is not a zero and it is not a promise in either direction. The question cannot be answered from the record you hold, so it has to be put to the supplier and confirmed on the dated product page for the specific SKU.
Margin note — buyer action: Give the identity block a line for every capability field that is not recorded. Write "not recorded in the catalogue row read on [date]" and never a guess, because a guess that turns out wrong is discovered by the customer rather than by you.
| Model | What the catalogue row records | What the case must carry |
|---|---|---|
| SLK-9936S RFID Hotel Card Door Lock | Card and mechanical override access, 4 × AA alkaline batteries, a five-pin profile cylinder, hotel lock management software with a PC-linked card encoder, a standalone lock mode, guest card issuance and unlock records | The fitted finish and door thickness band, and a note that the row records the power supply but no low-battery alert entry, so the alert is confirmed on the dated product page for the specific SKU |
| SLK-019 Stainless Steel Keypad RFID Smart Lock | Card, fingerprint and mechanical override as standard, passcode and mobile app only on a model version, a fingerprint capacity, a stated panel size, a low-battery alert | Which version was installed, because the version decides whether the passcode and the application travel with the unit |
| SLK-1106 Digital Fingerprint Smart Lock | Fingerprint access, with the platform, power, door fit, packing and capacity groups not recorded | The access method as the only recorded capability, and each unrecorded field marked as unrecorded rather than inferred |
Confirm the date on which each row was read on the dated product page for the specific SKU before the reading drives a procurement or refurbishment decision, and keep the reading in the case file. A catalogue row read on one date and a product page checked on another are two pieces of evidence, and a case that does not distinguish them cannot be defended when the two disagree.
The identity block also has to say what the unit is holding. Several catalogue rows record unlock records and guest card issuance as functions of the management platform, and rows that record local programming directly on the lock mean the configuration travels with the hardware. Before a unit is packed, the authorisation states what is to be done with the credential state inside it: which cards or codes are to be voided, whether the unit is to be returned to a factory state, and who confirms that it was done.
One further line belongs in the identity block, because it is easy to skip. List the parts removed from the door: escutcheon, mortise case, latch, deadbolt, strike plate, strike keeper, cylinder, spindle, battery compartment. A return that includes the mortise case and the strike keeper tells a very different story from one that includes only the panel, and the receiving desk cannot tell the difference unless the case says which parts travelled.
CONFIRM-WITH-LOCAL-DATA-PROTECTION-RULES: what your own jurisdiction and your own internal policy require before a unit holding issued credentials, access records or resident data is allowed to leave the site, and who signs that decision.
Turn 5 — What has to travel in the box
Facilities Lead: How much do we actually put in the carton?
Supplier Support Engineer: Enough that the case can be worked without a phone call back to the door. The record, the observations, the images and the parts. The unit itself is the smallest part of the package.
Facilities Lead: The technician photographs everything anyway. Does the form need the photographs as well?
Supplier Support Engineer: The photographs belong to the case, not to a phone. A photograph that lives in a personal album on a device that gets replaced is not evidence. The case needs the images attached, in the order the observations were taken, with the door identifier visible in the frame where that is possible.
Margin note — buyer action: Ask for four images: the installed unit in place, the label, the door and frame with the latch and strike visible, and the gap at the door edge. Four images cover most of what a receiving engineer asks for.
The pack holds observations and readings, never conclusions. If a test was not performed, the sheet says so, and the receiving desk would rather know that than receive a confident sentence about a component nobody examined.
The other half of the definition is what comes back, and it is the half most returns processes never specify. A returned unit arrives with the identity block it left with, or it does not. A label that cannot be read cannot be matched to a variant, and a unit that cannot be matched to a variant cannot be worked against the right procedure. So the return also carries the parts the supplier returns, the replacement unit's own identity, the credential state as it was left at the factory, and the written outcome of the examination. Where the outcome is that no fault could be determined, the case closes with that finding recorded rather than with a blank.
CONFIRM-PER-SUPPLIER: what the receiving process does with a returned unit, what it reports back and in what form, what it needs in the box for the case to be worked rather than returned again, and how long the unit is expected to remain off site before the door is covered.
Turn 6 — Who signs, and what the signature has to contain
Facilities Lead: In our building, whoever is on shift raises it and nobody signs anything. It works because the person on shift is also the person who fits the spare.
Supplier Support Engineer: That works until it does not. The moment the case involves freight, a supplier instruction or a credit, the person on shift is making a decision they were never authorised to make, and nobody can tell afterwards who made it.
Facilities Lead: So who should sign?
Supplier Support Engineer: That depends on your agreement, so I cannot tell you, and neither can any sheet. What the sheet can do is name the roles, so that the answers are written down somewhere.
Facilities Lead: Roles rather than names.
Supplier Support Engineer: Roles, with the name recorded against the role on the day. A process that depends on one person being reachable is a process with an invisible queue. Every decision below therefore has a named deputy, and the record shows which of the two signed.
| Decision | Decided by | The record must carry | If the decider is unavailable |
|---|---|---|---|
| Repair at the door within the visit | The attending technician, within documented competence | The part fitted and the door left operable | Anything beyond competence is escalated, not absorbed |
| Authorise a return for repair | The facilities lead, or the supervisor named for that estate | The branch, the evidence relied on, the alternative declined | The named deputy signs, and the substitution is recorded |
| Replace from the spares pool | The facilities lead | The unit fitted and the failed unit retained | The spares pool procedure applies |
| Write off the unit | The facilities lead, with the budget owner where the loss is material | The reason and the asset record update | The case is held, and the hold is visible on the log |
| Accept the return and issue the outcome | The supplier support side | What was found, what was done, what comes back | The case stays open and the door stays on interim cover |
| Approve freight and who carries it | The buyer, under the agreement | The arrangement and the reference it was raised against | The agreement governs, not the case |
| Close the case and raise the action | The facilities lead | The action, its owner and its due date | The case stays open until the action is raised |
An unlogged decline is indistinguishable from a delay. When a return is considered and refused, the record says so, says by whom, and says what evidence would have changed the answer. That single line is usually the difference between a returns process that learns and one that quietly accumulates.
Turn 7 — The two legs a return adds to every interval you measure
Facilities Lead: You said the process does not publish any figures. How am I supposed to plan against it?
Supplier Support Engineer: You are not planning against mine. You are planning against your own records, and the only thing I can give you is the measurement method, so that your number and my understanding of it are the same number.
Facilities Lead: Then what changes for me when a unit goes back?
Supplier Support Engineer: The interval gets longer, and it gets longer in a way your current record cannot show. That is the practical point of the two extra timestamps on the sheet.
The measures that matter for a returns process are not difficult to define, and none of them can be read off a supplier's letter. They are all computed from the pairs of timestamps your facility team already holds.
| Measure | How it is computed from your records | What it does not tell you |
|---|---|---|
| Response time | From the fault being reported through the agreed channel to a person or system acknowledging it | Whether anybody attended the door |
| On-site attendance time | From that acknowledgement to somebody standing at the door | Whether the door was fixed |
| Resolution time | From the fault being reported to the door being operable again | Why the interval was what it was |
| Mean time to repair | The mean of the resolution intervals across a stated period and population | The tail, or the doors that were never reported |
| Availability | Operable time divided by scheduled time, over a stated population and period | Whether the population is the one that matters to you |
| First-time fix | A visit ending with the door operable and no repeat visit for the same symptom | Whether the root cause was ever found |
| Repeat rate | The share of failures that return within a stated window | The severity of any of them |
| Failed on arrival | A unit that fails before it is put into normal service | Whose fault it was; the boundary belongs in the agreement |
Your target for each of those is your planning assumption, set from your own estate and your own risk position. The process supplies the measurement, and the measurement comes from the case sheet.
A return adds two legs that no door-side measure captures: the interval between the authorisation and the despatch, and the interval between the despatch and the receipt at the receiving desk. Both belong on the case sheet as their own pair of timestamps. If they are not recorded separately, the resolution interval for that door blends the maintenance operation with a logistics chain, and when the estate asks why a floor's doors are slow, the answer cannot be decomposed. A door waiting on a unit in transit and a door waiting on a technician are different problems with different fixes.
Three further costs belong in the buyer's own record rather than in anybody's claim. The first is freight and handling, a commercial arrangement that in most agreements the buyer arranges and sometimes carries. The second is the stock the case consumes when the branch is a replacement from spares. The third is the cost of a second visit, which appears whenever a return arrives without the evidence needed to work it. The third is the hardest to see, because it is booked as routine maintenance rather than as a returns failure.
Turn 8 — The receiving end and the answer that comes back
Facilities Lead: What actually happens to the box when it reaches you?
Supplier Support Engineer: The unit is examined against the symptom in your record. That is the point of sending the record: the symptom you wrote is what gets compared against what is found. If your check record shows that the power path was already excluded at the door, nobody here spends a morning excluding it again.
Facilities Lead: And if the unit looks fine?
Supplier Support Engineer: Then that is the outcome, and it is a meaningful one if the record says what was checked at the door. A unit that passes its examination with no matching symptom behind it usually means the symptom belongs to the door, the frame or the configuration rather than to the unit.
Facilities Lead: So what comes back to me?
Supplier Support Engineer: One of four things: a repaired unit, a replacement unit, an outcome that no fault could be determined, or an observation about a group of units rather than about yours. The last one is why the cluster question matters. Two units from the same batch with the same fingerprint is a different conversation from two units with two different faults, and the case sheet is what makes the difference visible to someone who has never seen your building.
CONFIRM-PER-SUPPLIER: what the receiving process returns to you in writing for each of those four outcomes, what you need to send for each to be worked, and what your own agreement provides for in the interval between the unit leaving the site and a door being covered.
The supplier side cannot read a fault out of a unit that arrives with no identity, no symptom and no check record. Everything in the pack is therefore not politeness to the receiver; it is the mechanism by which the case is worked at all. A case that returns unworked usually returns unworked because the record was thin, and it usually returns unworked a second time for the same reason, because nobody changed anything in between.
Turn 9 — Closing the case so the next one is not the same one
Facilities Lead: This is where I lose most of it. The unit comes back, we fit it, and then what?
Supplier Support Engineer: Then the case has to close with something, or it has not closed at all. A case that closes with no record of what changed is a case that will be reopened by the next occurrence of the same fault, on a different door, in a different month, and nobody will make the connection.
Facilities Lead: So every case has to produce an action?
Supplier Support Engineer: Every case has to produce exactly one action, and one of the permitted actions is that no change was required. The difference matters. A case with no action recorded is indistinguishable from a case nobody looked at.
| Action type | What it changes | Who owns it | The evidence that it closed |
|---|---|---|---|
| Form change | A field, a wording, or a block that should have been mandatory | The process owner | The new form version and its effective date |
| Door-side procedure change | The sequence, the tools carried, or the escalation point | The maintenance lead | The revised procedure and the briefing note issued against it |
| Supplier question raised | A capability, scope or documentation gap only the supplier can close | The facilities lead | The written answer and the date it was filed against the case |
| Spares mix adjusted | The variant mix held on site, where a variant cannot be covered by the stock | The stores owner | The revised spares list |
| Training note | A competency gap the case exposed | The training owner | The session record against the case |
| Cluster opened | A suspected batch, configuration or firmware pattern | The facilities lead | The cluster record and its outcome |
| No change required | The case was handled correctly and the process is sound | The facilities lead | The note that says so, and the reason |
The review that consumes these actions works on the population rather than on the individual case. Cases are grouped by model and variant, by symptom phrase, by the family of cause that was eliminated, by the branch chosen and by the outcome. The output is a change to the process, a question to the supplier, a training note or an explicit decision that nothing needs to change. It is not a statistic for its own sake, and a review that produces a number and no change has not been run.
The measures you compute from the closed cases feed that review directly. A repeat rate is the share of failures that come back within a stated window, and a rising repeat rate on one model and one symptom is the signal that the process is not solving what it is seeing. Neither number is anybody else's to publish; both are yours to compute, and both become useful only because the case sheet made them computable.
There is a short list of reasons the same case comes round twice, and each one has a different fix.
- The intake record was never completed, so the case exists only as a technician's memory.
- The door-side check record was missing or summary only, so the case could not be worked.
- The identity block was incomplete, so the returned unit could not be matched to a procedure.
- The unit travelled without the part that mattered, so the case had nothing to look at.
- The decision was taken by someone without the authority to take it, and the case stalled.
- The closure was recorded without an action, and the same symptom was written in different words each time.
stateDiagram-v2
[*] --> Reported
Reported --> IntakeComplete
IntakeComplete --> DoorChecked
DoorChecked --> Held
Held --> DoorChecked
DoorChecked --> Authorised
Authorised --> RepairedOnSite
Authorised --> Returned
Returned --> Received
Received --> OutcomeIssued
OutcomeIssued --> Closed
Closed --> ActionOpen
ActionOpen --> [*]
The symptom vocabulary that makes a cluster visible
That last reason is the quietest and the most expensive. A cluster of twenty cases across three floors is invisible if each was written in the words of whichever technician attended, because the review groups cases by wording and the wording differs every time. A controlled vocabulary costs almost nothing and makes the population legible.
| Controlled phrase | The observation that must accompany it | The question it raises next |
|---|---|---|
| The handle turns but the unit does not move | Whether the mechanical override moved, and what the latch did | Whether the drive, the power path or the electronics is at fault |
| The unit answers no credential, including the override | What the indicator showed while the attempt was made | Whether the sequence splits into jam, power or silence |
| The unit answers intermittently | How many attempts were made and how many succeeded | Whether the door position or the environment is involved |
| The reader does not recognise a credential that works on another door | Whether the reader and the cylinder were tried directly | Whether the credential path is at fault rather than the unit |
| The unit reports a low battery condition with cells just fitted | How power was tested and what was read | Whether the alert is a threshold rather than a measurement |
| The unit works on the bench and not on the door | What changed between the bench and the door | Whether the fault belongs to the installation rather than the unit |
The technician still writes freely in the notes field. The controlled phrase is a classification, not a straitjacket, and it exists so that twenty cases can be found together. Where no phrase fits, the case carries the free text and the phrase field carries "unclassified", and the number of unclassified cases is itself worth watching.
The record sheet, block by block
The sheet is the artefact. It is designed to print on one side of A4 for the first six blocks, with the last two blocks filed separately because they are completed at different times by different people: the closure blocks are filled when the outcome comes back, not at the door.
The blanks below are the fields, and the note beside each one says what it is for, so the sheet can be issued to a team that has never seen it. Where the note says "from the object schedule", the schedule is the only acceptable source. A room number is not an identifier for a unit.
Block A — the case header
| Field | What to write |
|---|---|
| Case reference | Issued when the intake record is opened; every later document carries it |
| Door or unit identifier | From the object schedule, not the room number |
| Reported date and time | With the channel it came through and who reported it |
| Door operable again date and time | Filled when the door is covered, not when the case is opened |
| Interim cover | Spare swap, use of the override, or the door held under a watch condition |
| Priority, and who set it | A priority nobody can explain becomes a queue nobody can read |
Block B — the identity of the unit
| Field | What to write |
|---|---|
| Model as written on the label | Transcribed, not inferred from the purchase order or the spares list |
| Variant or version | Some catalogue rows record access methods as available only on a model version |
| Date the label was read | And by whom |
| Catalogue row consulted | With the date of that read, and the copy kept |
| Dated product page checked | Confirm on the dated product page for the specific SKU before the reading is relied on |
| Capability field not recorded | Written as "not recorded in the catalogue row read on [date]", never as a guess |
| Parts removed from the door | Escutcheon, mortise case, latch, deadbolt, strike plate, strike keeper, cylinder, spindle, battery compartment |
| Physical condition | Scratches, water marks, impact marks, corrosion, anything unusual about the finish |
Block C — the symptom and the observation
| Field | What to write |
|---|---|
| Controlled symptom phrase | One phrase from the vocabulary, or "unclassified" |
| The technician's own words | Unedited, in the notes field |
| Observation | What was seen, read or checked, with the conditions it was taken under |
| Readings taken | The reading, the instrument and where it was taken |
| Checks not performed | And why, so that a gap is visible rather than assumed absent |
Block D — the door-side check record
| Field | What to write |
|---|---|
| Check record attached | Identify the sheet by its reference |
| Sequence run in order | Yes or no, with the reason if not |
| What each check ruled out | The summary that travels with the case |
| Cluster question answered | The second door identifier, or the statement that no comparable door is installed |
| Authorised by | The role, the name against the role, and the date |
Block E — the branch and the decision
| Field | What to write |
|---|---|
| Branch chosen | Repair at the door, return for repair, replace from spares, or write off |
| Evidence relied on | Which lines of the check record supported the decision |
| Alternative declined | And why it was declined, because an unlogged decline is a delay |
| Cost owner | The budget the case is charged to |
| Freight arrangement | Who arranges it, who carries it and who pays, as the agreement provides |
Block F — the return
| Field | What to write |
|---|---|
| Contents of the box | The unit, the parts, the record and the images, listed |
| Cell state | Cells removed, or the reason they remain fitted |
| Credential state at despatch | Cards voided, codes expired, unit returned to a factory state, and who confirmed it |
| Override key | Which key, who holds it, where it is recorded |
| Despatch date and time | The first leg of the return |
| Received date and time | The second leg, taken from the receiving docket |
| Receiving reference | So the box and the case can be joined at the far end |
Block G — the outcome
| Field | What to write |
|---|---|
| Outcome recorded by the receiving side | Repaired and returned, replaced, no fault determined, or a cluster observation |
| What came back | The unit, the parts, or the replacement, with its own identity block |
| Fitted on the door | With the date the door was operable again |
| Credential state after | Confirmed against the door that is actually in service |
| Case closed on | Date and role |
Block H — the review
| Field | What to write |
|---|---|
| Cluster the case fell into | Model, variant and symptom phrase |
| Action raised | One of the permitted actions, or an explicit no change required |
| Action owner and due date | Named by role |
| Form version used | So a later case can be compared against what was in force at the time |
| What changed | And when it took effect |
| Next review date | The date the population is read again |
Four rules make the sheet work. A blank field is a defect in the sheet rather than in the technician, so the sheet carries lines for "not performed" and "not available" to keep the two apart. The sheet is versioned, because a worksheet edited by email without a version number is not a record. One case is one sheet, and the population review reads many. And the sheet is retained after the case closes for as long as your estate retains its fault history, a decision your own policy and your own insurer's position should inform.
CONFIRM-WITH-LOCAL-RECORD-KEEPING-RULES: how long your own jurisdiction, your own building policy and your own insurance arrangement require you to keep fault records, images of door hardware and credential state records once a unit has left the site.
erDiagram
RMA_CASE {
string case_ref
string door_unit_id
string symptom_phrase
string branch_outcome
}
DOOR_UNIT {
string model_variant
string label_read_date
string catalogue_read_date
}
DOOR_UNIT ||--|| RMA_CASE : "is recorded against"
CHECK_RECORD ||--|| RMA_CASE : "is attached to"
EVIDENCE_ITEM ||--o{ RMA_CASE : "supports"
DECISION ||--|| RMA_CASE : "settles"
RETURN_SHIPMENT ||--o| RMA_CASE : "travels as"
CLOSURE_ACTION ||--|| RMA_CASE : "closes with"
What this sheet will not settle
It does not find the fault. The door-side sequence finds the fault, or narrows the field until somebody with a bench can. The sheet carries that work forward; it does not replace it.
It does not set your targets. Every measure in this subject is computed from your records and against targets you set, and no figure in this article is anybody's commitment.
It does not tell you whether a given model is repairable in the field, how long a repair takes, or what a replacement costs. Those are questions to put to the supplier in writing, model by model, and to file the answers in your own records.
It cannot turn an unrecorded capability into a capability. Where a catalogue row leaves a field empty, the record says so and the question goes to the supplier.
Where the next step sits
Five moves, and none of them is a purchase.
- Write the authorisation gate down as a rule. One sentence: no unit leaves a door until the identity block is complete, the check record is attached with what each check ruled out, the cluster question is answered, and a named role has signed.
- Issue one sheet per case and version it. Eight blocks, one page for the first six, with the closure blocks filed separately when the outcome comes back.
- Fix the symptom vocabulary before you need it. A page of phrases now is cheaper than a cluster that stays invisible for a year.
- Name the deciders and their deputies. By role, with the substitution recorded on the case.
- Set the review date and the agenda. Cases grouped by model, variant and symptom phrase, with one output per case and one change per review.
The sourcing-side preparation around this work is worth doing in parallel. The Smart Lock Supplier Due Diligence Checklist and the Smart Lock Factory Audit Checklist for International Buyers are the documents to put in front of a supplier before the returns arrangement is agreed, and the Smart Lock RFQ Guide for International Buyers is where evidence expectations get written into the enquiry instead of discovered afterwards. On the operational side, the Hotel Smart Lock Maintenance Guide covers the routine work that should keep cases from arising, the Hotel Smart Lock Site Survey is where the door schedule supplying the identifier comes from, and the Smart Lock Audit Logs explains the event records that sit alongside this sheet on the credential side.
Companion pieces cover the parts of the returns cycle this sheet deliberately does not duplicate: A Field Decision Tree for RMA Process Design sets out the branches in full, RMA Process Design in a Multi-Site Operation covers what changes when the estate spans several buildings, RMA Process Design Documentation Set lists the documents the process assumes, and Warranty Claim Evidence: A Field Worksheet to Complete covers the narrower case where a claim follows a fault.
The freight terms, the evidence the supplier needs in order to work a case and the way a return is settled are things to settle in writing, and the request a quote route is where a scoped enquiry starts that conversation. Product-side questions are best settled on a sample before any of it is scaled up, and order samples is the route for that. The process a returned sample then moves through is described under support and warranty.

