How we count
One number, one definition. The home screen, the owner report, the data export and your ChatGPT / Claude assistant all use the rules below, so they never disagree with each other. Where a rule differs by market, it says so.
The SLA clock
- Starts when a person applies triage to a request (accepts or edits the AI's category and urgency). A request that nobody has triaged has no deadline yet — it shows as awaiting triage instead.
- Window by urgency: Emergency 2 hours · Urgent 24 hours · Normal 3 days · Low 7 days.
- Pauses on the weekend for Normal and Low: Friday–Saturday in Qatar, on Doha time; Saturday–Sunday in Australia, on Sydney time for every Australian agency. Emergency and Urgent run through weekends. Public holidays are not paused.
- Stops when the request becomes a work order, or when a PM closes it without one (fixed by the tenant, not the landlord's job, no access, a duplicate). The SLA measures triage to dispatch decision; the repair itself is tracked on the work order.
- Breached = the deadline passed while the request was still open. At risk = the deadline is within the next 24 hours.
"Completed" and "fixed" are two different things
- Work done — the contractor (or your team) marked the work done. If there is a tenant to ask, the job now waits for their answer.
- Completed — the job is closed: the tenant answered fixed, your team closed it, or the confirmation period ran out with no answer (recorded as no reply, never as confirmed). When there is nobody to ask (a compliance job, or a request with no tenant contact), the contractor's "done" closes it. The completion date is that closing moment; the contractor's time is kept separately as work done.
- Fixed — the tenant answered 1 to "did it get fixed?" (WhatsApp), clicked Yes (email or the tracking page), or your team recorded the answer. Not fixed on a job waiting for the answer sends it back to in progress, with a fresh link for the contractor. On a job that is already completed it is a flag for you to handle; the status doesn't change.
- The contractor scorecard counts not fixed only among jobs where the tenant answered, so a contractor whose tenants never reply is neither punished nor rewarded.
Repeat failure
A job is a repeat failure when it is open on an asset (the specific AC unit, chiller, lift, heater) that already had a job finished within the last 30 days, counted from when that work was marked done. Requests and jobs must be linked to the asset for this to work; jobs without an asset never count.
Maintenance spend
- Spend = the final cost recorded on completed work orders. Cost ceilings, estimates and open jobs are not spend.
- Amounts are in the agency currency (QAR or AUD), before any tax you add in your accounting system.
- Grouped by kind (maintenance or compliance), property or month, using the job's completion date.
Contractor scorecard
Over the last 365 days per contractor: finished jobs (work marked done, whether or not the tenant has answered yet); tenant answers; not fixed count and rate (over answers); median days from the work order being created to the work being marked done; open jobs; last finished job; tenant rating: the average of the 1–5 ratings tenants gave in the same 365 days, to one decimal, shown from 3 ratings (one per job: a later rating replaces the earlier one, and it counts for the contractor who was on the job when it was rated). The contractor is measured to when they finished, not to when the tenant got round to answering.
Unit turns and vacancy days
A turn opens when a move-out is recorded and ends when the next lease starts, or when you close it.
- Ready-by date: your date, else the day before the next lease starts, else 14 days after the move-out.
- Days empty: whole days nobody lived there, from the day after the move-out to the day before the next tenant moves in. Moved out on 30 November and in on 15 December is 14 days. A tenant moving in the day the last one leaves is 0. On the property's own date.
- Vacancy days: those days across the portfolio over the last 365 days, turns going on included. Only from move-outs recorded in PMFriend: a unit that was already empty when you started isn't counted.
- Average days empty: per turn that ended in the same 365 days.
- Open jobs: the property's jobs a tradie hasn't finished, plus requests nobody has made a job of. A job only waiting for the outgoing tenant's answer is finished.
- Rent lost: days empty at the outgoing lease's rent per day (a year of rent over 365 days). An estimate from the lease, not from payments: PMFriend doesn't see those.
- Jobs on the turn: the final cost of jobs finished since the move-out was recorded (a finished turn stops counting when it ended), the budget of those still going, and how many have no price yet.
- Usually takes: the median days from a job being raised to the work being done, over the last 365 days, rounded up. A tradie's own figure needs 3 finished jobs; for a job nobody has yet, your agency's figure needs 5. With fewer, it isn't shown.
- Likely ready: the latest of the open jobs' booked visits and usual finish dates. An estimate, and it doesn't add time for cleaning that nobody has booked.
- Needed by, on the tradie's page: two days before the ready-by date, not before the move-out; the ready-by date for the cleaning job.
Time in each status (not on screen)
Every job records when it enters each stage: Draft → Awaiting owner approval → Dispatched → Scheduled → In progress → Waiting for the tenant's answer → Completed. Time per stage is a running total (a job that returns to a stage adds to its earlier time). Medians are taken over jobs that have left the stage. Available through the AI assistant and the data export, never on screen — see Where do jobs get stuck?.
Maintenance cycle time (not on screen)
The measure behind the pilot guarantee.
- Starts when the request reaches PMFriend: the tenant's WhatsApp message, e-mail or form, or the moment your team logs a phone call.
- Ends when the work is done: the contractor, or your team for them, marks the job done. A job the tenant sends back as not fixed counts from when it is done again. When a request needs several jobs, the last one ends it. The tenant's answer isn't waited for; the time to the close (the tenant's fixed, your close, or the end of the confirmation period) is reported next to it.
- Counts requests whose jobs are all finished, at least one of them completed. Requests closed without a job, duplicates, planned maintenance (raised ahead of its due date) and compliance jobs have no cycle time. Requests still open are counted separately and join once their work is done.
- Median = the middle request. The slowest 10% = the 90th percentile: one request in ten took at least that long. Calendar days, weekends included, with one decimal.
- By month means the month the work was done, in your market's time (Sydney for Australia, Doha for Qatar). Also by urgency and by category.
- The pilot counts requests that arrived from its first day and compares the median with the starting point an admin recorded from your own job records; the target is 30% lower.
- Available through the AI assistant, the data export
(
cycle_times.csv,request_cycle_times.csv) and the pilot report an admin e-mails to themselves. Never on screen.
Attention Inbox thresholds
| Item | Rule |
|---|---|
| Awaiting triage | request untriaged for 1 day or more (urgent after 3) |
| SLA at risk / breached | deadline within 24 h / already passed |
| Stuck work order | open job (dispatched, scheduled, in progress) for more than 14 days |
| Stale draft | draft never dispatched for more than 7 days |
| Tenant says it's not fixed | a not fixed answer nobody has marked handled |
| Repeat failure | see above (30 days) |
| Compliance overdue | the obligation's due date has passed (one item per obligation; not for an archived property) |
| Aqarat licence expired | the agency licence date has passed (Qatar) |
| Intake gone quiet | e-mail or WhatsApp intake is set up but nothing has arrived for 7 days |
| Condition record disputed | a tenant disputed an issued move-in, move-out or bond-out record and nobody has marked it handled |
| Rent behind | rent is 7 days or more behind the paid-to date recorded on the lease (urgent at 14; not for an archived property) |
| Unit turn at risk | the ready-by date has passed and it isn't marked ready (urgent); 3 days or fewer left with something still open; a visit booked after the date; likely ready after the date at the tradies' usual speed; or the cleaner booked before the work is done. Not for an archived property |
Contractor insurance and trade licences are not Attention items. Each insurance policy, and each licence the register check knows, gets one notification in the bell and on the Notifications page, 30 days before it expires (insurance: urgent if it had already expired when PMFriend first saw it). Record the renewed policy's date, or let the weekly register check find the renewed licence, and the new date gets its own notification 30 days before it runs out. Expired insurance ranks the contractor lower in the suggestions; it doesn't block them.
Owner report
The owner digest covers the last 14 days: the requests reported in that time, the jobs completed in it with their final cost, and compliance that is overdue or coming up. Nothing in it is estimated.