Six Handoffs for One Broken Tap
Ask property managers what ruined last week and the answer is maintenance, almost every time. Not because repairs are difficult: because one repair is six handoffs between three parties who cannot see each other.
Does the work of
- Maintenance Coordinator
- Repairs Coordinator
- Property Manager
- Assistant Property Manager
A tenant sends four words about a shower on a Friday afternoon. By the time anybody has worked out which trade to send, whether the owner needs to approve it, and when somebody can be home to let them in, it is the middle of the following week and nothing has been fixed. Nobody in that sequence did anything wrong.
In short
- The problem
- One routine repair is nine steps, six of them waiting on somebody else, across three parties who cannot see each other.
- Why it slips
- The platform records each event faithfully and notices none of the gaps between them, so a job waiting on nothing looks identical to one that is progressing.
- What we would automate
- Structured intake the moment a tenant reports it, possible-urgent escalation, threshold routing, and the chasing of owners, trades and access.
- What stays with a person
- The final urgency call, anything with an injury or safety dimension, insurance claims, spending above the threshold, and a tenant who is upset.
Want the whole thing by email?
This paper, plus the rest of the library. One click to stop them.
The Short Version
A tenant reports a dripping tap in free text. Somebody reads it, decides whether it is urgent, decides whether it is under the owner approval threshold, finds a trade, gets a quote, gets the owner to agree, books it, coordinates access, and then closes the loop with three parties. Every one of those steps is a message to somebody who is not at their desk.
That is why maintenance dominates the week. The individual decisions are easy (most of them are policy, not judgment) but the loop has six waiting states in it, and a property manager carrying a hundred and fifty properties is holding a few dozen of those in their head simultaneously. The platform records each event faithfully and notices none of the gaps between them.
The reframe is that maintenance is not a queue of jobs, it is a set of open loops, each waiting on a named party for a specific thing. Once you write it down that way, most of what a coordinator does is chasing, and chasing is mechanical.
What we would automate is the intake, the chasing, and the closing of the loop. Not the urgency classification where a statutory definition applies, not the spend decision, and not anything with a safety or injury dimension. Section 7 is the boundary and section 8 is the part that is buildable now.
6
Handoffs in one routine repair
Enumerated in section 1: each one a message to somebody not at their desk
3
Counterparties, none of whom can see the others
The tenant, the owner and the trade, each with a different idea of where it is up to
1
Rule that decides most of it
The owner approval threshold in the management agreement is the only real automation input
1. Count the Handoffs
Here is one uncontroversial repair (a leaking mixer tap in a kitchen, no safety issue, a few hundred dollars), written out as the loop it actually is. Nothing in this table is difficult. The column that matters is the middle one.
| Step | Waits on | What usually goes wrong |
|---|---|---|
| The report arrives | Nobody | Free text, no photo, no indication of when it started or whether there is a pet on site |
| Urgency decided | A person reading it | Sits unread over a weekend, which is the failure mode with actual legal consequences |
| Approval path decided | The management agreement | Nobody is sure what this owner’s threshold is, so it gets referred out of caution |
| Trade engaged | The trade replying | A voicemail and a text, then silence for two days. The job has no status because it never started |
| Quote obtained | The trade attending or estimating | Often needs an access visit first, which is a second scheduling problem before any work happens |
| Owner approves | The owner reading an email | The single longest wait in the loop, and the one nobody is chasing because it feels rude |
| Job booked, access arranged | Tenant and trade agreeing a time | Two parties, three attempts, and a manager relaying times between them by text |
| Work done | The trade | Nobody confirms it happened. The tenant assumes you know; the file still says booked |
| Invoice, ledger, close | The invoice arriving | Charged to the wrong property, or the job closes in one system and stays open in another |
The bottom line
Six of those nine steps are waiting on somebody else. That is what makes maintenance feel like drowning while looking, in the system, like nine tidy events. The work is not the repair. The work is being the only party who knows the whole state.
2. Urgent Is a Legal Category, Not a Judgment Call
This is the part of maintenance that is not about efficiency at all. Australian residential tenancy legislation defines urgent repairs, typically as a list: along the lines of a burst water service, a blocked or broken toilet, a serious roof leak, a gas leak, a dangerous electrical fault, flooding or serious flood damage, serious storm or fire damage, a failure of an essential service such as hot water or cooking facilities, and faults that make premises unsafe. The exact list, and what follows from it, is set by each state and territory.
Two consequences make this the highest-stakes triage decision in the role.
The clock is short and it starts at the report
Urgent repairs carry obligations to act promptly, and in several jurisdictions non-urgent repairs also carry a prescribed timeframe. The clock starts when the tenant told you, not when somebody opened the email, so an unread Friday message is already a compliance problem by Monday.
If the tenant cannot reach you, the decision becomes theirs
Most jurisdictions allow a tenant to arrange urgent repairs themselves, up to a capped amount, and be reimbursed, where the manager or owner cannot be contacted. Being uncontactable does not defer the cost: it transfers the choice of tradesperson and the price to somebody with no reason to shop around.
Misclassifying downwards is the expensive error
Treating an urgent repair as routine risks a tribunal finding, a rent reduction claim and, where somebody is hurt, considerably worse. Treating a routine repair as urgent costs an after-hours callout. The asymmetry should be built into whatever does the sorting.
Minimum standards changed what counts
Several jurisdictions now prescribe minimum standards for rental premises (heating, locks, ventilation and similar), which means some things that were once a request are now an obligation. Any triage rule written before those regimes will quietly under-classify.
The practical design conclusion: a machine should read every report the minute it arrives, ask the questions that establish whether it could be urgent, and route it to a person immediately if it could be, without waiting for anyone to open an inbox. Recognising a possible urgent repair out of hours is worth more than any other automation in this paper, and it is also the one thing that must never be left to decide on its own.
Urgent repair definitions, response timeframes, tenant self-spend caps and minimum standards are set by state and territory legislation and change. Treat the list above as the shape of the obligation and your regulator as the authority on the current rule.
3. The Threshold Is the Only Real Rule
Every management agreement carries a figure: the amount the agency may spend on a repair without asking the owner first. It is the most underused number in property management. Below it, the entire approval handoff (the longest wait in section 1) does not exist. Above it, nothing can move until an owner replies.
Three things are usually true about it, and all three are fixable in an afternoon.
Nobody can see it at the moment of decision
The figure is in a signed agreement in a document store, not on the property record. So the manager guesses, and guessing conservatively means referring a $180 job to an owner who authorised $500 of discretion.
It was set once and never revisited
A threshold agreed four years ago buys a fraction of what it did. Revisiting it across the portfolio is one conversation per owner and it permanently removes work from every future repair.
Standing approvals are rarely asked for
Many owners will happily pre-approve a named trade, or any repair of a given type up to a figure, or all repairs at a property they never want to hear about. Almost nobody asks, so almost nobody has it.
The honest version
Before automating anything, put the approval threshold on the property record where whoever is triaging can see it, and ask every owner whether it is still the right number. That is not a software project and it removes a handoff from a large share of repairs.
4. Where the Loop Actually Stalls
A stalled repair is almost never a repair somebody decided not to do. It is a repair waiting on a party who has not replied, in a system that has no concept of waiting. These are the four states where jobs go quiet, in rough order of how much time they cost.
Waiting on the owner
An approval email sent into an inbox belonging to somebody at work, who will deal with it on the weekend and then forget. Nothing chases it, because chasing an owner feels like pestering a client. Meanwhile the tenant thinks you have gone silent.
Waiting on the trade to quote
The job has been sent and not accepted, or accepted and not scheduled. There is no status for “we asked and heard nothing”, so it looks identical in the system to a job that is progressing.
Waiting on access
Trade and tenant both available, on different days, relayed through a third person by text. This is the handoff that generates the most messages per outcome of anything in the role.
Waiting on nothing at all
The work was done a fortnight ago. The trade has not invoiced, nobody confirmed completion, and the file still says booked. These are the jobs that turn up in a rent roll audit and the ones an owner discovers before you do.
This is a general problem rather than a property one: software records events, and every state above is an absence. We wrote it up from the construction side, where the same structure produces a register of outstanding items living in one person’s head, in waiting is not a status. The fix is the same in both industries: give every job an expected next event with a date and an owner, and treat the absence of that event as the thing worth alerting on.
5. Triage Happens at the Report, Not After
Most of the elapsed days in a maintenance loop are spent asking questions that could have been asked at the moment the tenant was paying attention. “The shower is broken” generates a reply asking what exactly is wrong, then a two-day wait, then a photograph that reveals it is a different trade entirely.
A structured intake is the cheapest large win available here, and it is worth being specific about what it must capture.
- 1
What, exactly, and where in the property
Enough to pick the right trade first time. The difference between a leaking tap and a leaking pipe under the slab is the difference between a plumber for an hour and an insurance conversation.
- 2
When it started, and whether it is getting worse
This is the urgency signal, and tenants volunteer it readily when asked and almost never otherwise. A three-week-old drip and a three-hour-old drip are different jobs.
- 3
A photograph or a short video
Halves the diagnostic guessing, prevents the wrong trade being sent, and becomes evidence if the cause is later disputed at exit.
- 4
Safety and habitability flags
Water near power, gas smell, no hot water, no cooking facilities, no heating, a door or window that will not lock, anyone in the household at particular risk. Asked as explicit questions, not inferred from prose.
- 5
Access reality
Working hours, pets on site, whether the tenant consents to the trade holding keys or entering unaccompanied. Collecting this once removes the longest relay in section 4.
- 6
Whether it has been reported before
A repeat report about the same thing is a different problem: a failed repair, an owner declining work, or a defect. It should never re-enter the queue as if it were new.
Notice that none of those questions requires a property manager to ask them. They require somebody to ask them immediately, in whatever channel the tenant used, and to keep asking until the answers are complete. That is a conversation with a checklist behind it, which is the most automatable thing in this paper and the reason intake is where we would start.
6. The Trade Is the Third Party, and They Have the Same Problem
It is worth remembering what the other end of that unanswered call looks like. The plumber you are chasing is on a roof, missing their own inbound calls, and running the same relay in the opposite direction, which is the subject of most of our trades library. Two parties each waiting for the other to be free is not a responsiveness problem, it is a structural one.
Which suggests some unglamorous conclusions about the trade side of the loop.
A small panel beats a long list
Three trades per discipline who know your properties, your paperwork and your approval process will out-perform twenty who each need explaining to. Response time is mostly a function of familiarity.
Send a job, not a question
A message containing the property, the access arrangement, the fault, a photograph and the approved spend can be accepted in one reply. “Are you free this week?” cannot, and costs three exchanges before anything is scheduled.
Ask for the compliance documents once, and track their expiry
Licence and insurance currency for every trade on the panel, with the dates held somewhere that raises its hand. This belongs in the same queue as everything else in the compliance calendar paper.
Confirm completion as a separate event
Invoicing is not evidence the work was done, and a trade's job being closed is not the same as your loop being closed. A one-line confirmation from the tenant closes it properly and costs nothing.
7. What Stays With a Person
We build this kind of automation and we would still keep the following out of it. In each case a machine deciding is worse than the delay it removes.
The final urgency call
A system can recognise that a report might be an urgent repair, ask the clarifying questions and escalate it within minutes. It should not be the thing that decides a burst pipe can wait until Tuesday.
Anything with an injury or safety dimension
Electrical faults, gas, structural movement, a fall risk, a report that mentions somebody being hurt. Straight to a person, at any hour, with the after-hours path already agreed.
Insurance claims and anything that might become one
Water damage, storm damage, malicious damage. Whether something is a repair or a claim changes who pays and what evidence has to exist, and getting that wrong early is difficult to undo.
Spending above the threshold, and any spend an owner has declined
The threshold conversation is the agency's authority to act. Where an owner has said no to work you think is necessary, that is a professional and sometimes a legal conversation, not a workflow step.
A tenant who is upset, or reporting the same thing again
Repeat reports and unhappy tenants are signals that the process already failed once. Both deserve somebody who can apologise and make a decision.
8. What Can Be Automated Today
What is left after section 7 is still the large majority of the elapsed time in a maintenance loop, because almost all of that time is spent waiting and chasing rather than deciding.
| The work | What a machine does with it | What it hands back |
|---|---|---|
| Intake, at any hour | Asks the six questions in section 5 in the channel the tenant used, until they are answered | A complete report with photos, access details and safety flags, before anyone opens an inbox |
| Possible-urgent detection | Matches the report against the urgent and habitability categories and escalates on suspicion | A phone call to the on-call person in minutes, not a queue position |
| Threshold routing | Reads the approval limit and any standing approval on the property record | Either a job that proceeds without asking, or a specific request to a named owner |
| Owner approval chasing | Follows up an unanswered approval on a schedule, by email then by voice | A recorded decision with a date, or a documented failure to obtain one |
| Trade engagement and chasing | Sends the complete job, then chases acceptance and scheduling instead of waiting | An accepted job with a date, or a reassignment before a week is lost |
| Access coordination | Negotiates a time between tenant and trade directly, both channels, one thread | A confirmed appointment, without a manager relaying times by text |
| Completion confirmation | Asks the tenant whether the work was done and whether it fixed the problem | A closed loop, or an early warning that the repair failed |
| The waiting report | Tracks every job by what it is waiting on and for how long | One list of everything gone quiet, which is the artefact nobody currently has |
Where we come at this
We build in this gap, so read the table above as interested rather than neutral. What we make is the layer underneath the conversation: Australian voice, the compliance that decides whether a call may be placed at all before it is placed, memory that persists across calls so the third conversation knows what the first one said, and a single thread across voice, SMS and messaging rather than three channels that cannot see each other. Maintenance is a good fit for it precisely because the hard part is three counterparties and a lot of chasing rather than a difficult decision. It is also genuinely custom work: the conversation on top is specific to a rent roll, and the write-back has to land in your platform rather than in a second system. That is a scoped build, not a subscription, and anyone who tells you otherwise has not looked at your approval thresholds.
9. Getting Started
- 1
Time one loop end to end, twice
Take a completed routine repair and a completed one that needed owner approval. Record elapsed days and the number of separate messages sent by anyone. The touch count is what a tool can move; the elapsed days are what the owner is judging.
- 2
Split last month's jobs by the threshold
How many were under the approval limit? If the answer is most of them and they were referred to owners anyway, you have found free time before buying anything.
- 3
Find the jobs that are waiting on nothing
Every open maintenance item more than three weeks old with no event on it. Read what each one is waiting on. This list is the argument for everything in section 8, and it is uncomfortable reading.
- 4
Check your after-hours path against section 2
If an urgent repair is reported at 6pm Friday, what happens? Answer it honestly, including who is actually looking. This is the one item here that is about liability rather than efficiency.
- 5
Automate intake first
Structured intake plus possible-urgent escalation, before any of the chasing. It is the cheapest piece, it improves every downstream step, and it is the one nobody regrets.
Urgent repair definitions, response timeframes, tenant self-spend caps and minimum standards obligations vary by state and territory and change regularly. Nothing here is legal advice: confirm the current obligation with your regulator or a property law adviser. The handoff sequence in section 1 describes a common Australian residential workflow as at September 2026 rather than any particular agency.
Before you build anything custom
Two of the three counterparties in this loop have a turnkey option already. For the agency’s own money (management fees, letting fees, its own invoices), Office Voice chases on live Xero data and sends a payment link mid-call. For the trades on your panel, the same product answers the phone they keep missing, which is why they are not calling you back; that is what our trades page covers. The rent roll maintenance loop itself is the part that is genuinely a build, worth trialling the first before commissioning the second.
Not ready to talk yet?
Leave your email and we’ll send this paper to your inbox, with a link to the rest of the library. We publish a new one every few weeks and you’ll get those too, one click to stop them.
Your email is used to send you the papers and nothing else. We don’t share it.
How many jobs are waiting on nothing?
Run the third exercise in section 9 and bring the list. Eight minutes with our AI will price what the chasing is costing across your portfolio, and say plainly which part of the loop is worth closing first.