Maintenance KPIs: what to measure and how to count it

Most property managers know maintenance is slow before they can say where. A tenant says nobody called back, a property owner asks why a leak took two weeks, and the answer is buried in texts and call logs. A few numbers, counted the same way every month, show which stage is slow and who owns it.

This guide defines six measures, says where each one starts and stops, and what to change when it slips. It sets no targets for you, because the right target depends on your properties, your vendors and your state’s repair rules. Pick your own, write them down and compare each month against the last. The vendor scorecard template turns three of these into a quarterly sheet per vendor.

Count from timestamps, not memory

Every measure below is the time between two events, or a count of work orders. That only works if each event has a time on it. Before you measure anything, make sure your records show:

• When the tenant first reported the problem, by any channel
• When someone first replied to the tenant
• When a vendor or technician said yes to the job
• When the visit happened
• When the tenant confirmed the fix, or the photo of the finished work arrived
• When the work order was closed

If a step happens by phone or text outside your software, it needs a note with a time, or the number will flatter you. Use the median, the middle value, rather than the average: one job that waited three weeks for a part shouldn’t hide that most jobs moved in two days.

1. First response time

What it measures: how long a tenant waits to hear that a person or system has their request. Starts: the tenant’s first message or call. Stops: the first reply that says what happens next. An automatic “we got your message” with no next step doesn’t count.

Count it separately for business hours and for nights and weekends. A team that replies in minutes at noon and at 9:00 AM ET the next day for a 10:00 PM ET text has two different problems. If the night number is slow, the fix is usually a written after-hours policy and an on-call rotation, not more effort during the day.

2. Time to assign

What it measures: how long a request waits before someone has agreed to do the work. Starts: the first report. Stops: a vendor or in-house technician says yes to the job. Offers that go unanswered don’t stop the clock.

This is where most delays hide. A job offer that is missing a photo, access notes or a spend limit gets a question back instead of a yes. A job sent to one vendor with no backup waits for as long as that vendor is busy. When time to assign slips, look at why vendors don’t answer and at your follow-up schedule before you look for new vendors.

3. Time to visit

What it measures: how long from the yes to someone on site. Starts: the vendor or technician accepts. Stops: the visit happens. Rescheduled visits keep the clock running.

Split it by job level. An emergency and a dripping faucet belong in different rows, and your emergency definitions decide which row a job is in. Check the emergency and urgent rows against your state’s rules. Repair timelines by state has the states that set a clock.

4. Time to close

What it measures: the whole wait, from the tenant’s report to a confirmed fix, often called resolution time. The glossary entry on response vs resolution time sets the two side by side. Starts: the first report. Stops: the tenant confirms the fix, or a photo or note shows the finished work, and the work order is closed.

Close only on confirmation. A work order closed on the vendor’s word looks fast here and shows up later as a repeat repair. The post on confirming a repair before you close has the one text to send the tenant.

Open work orders by age

Time to close only counts jobs that finished. Once a week, also list the work orders still open, sorted by how long they have been open. The oldest few are usually waiting on a part, a quote or a property owner’s approval. Each one needs a named next step and a date.

5. Repeat repairs

What it measures: how often the same problem comes back. Counts: work orders for the same unit and the same problem within a window you choose, such as 90 days, as a share of all closed work orders.

Count a failed repair that goes back to the same vendor as a warranty callback, not a new job. That keeps the vendor’s scorecard honest and keeps you from paying twice. The post on stopping repeat repairs covers how to spot one and when to call the vendor back.

6. First-time fix

What it measures: how many jobs were done in one visit, the first-time fix rate. Counts: closed work orders with a single visit, as a share of all closed work orders. Track it per vendor and per trade.

A low number often points upstream. Jobs offered with no photo or no model number send the vendor out to look before they can fix. Better triage and a complete tenant maintenance request form raise it more often than a vendor change does.

The six measures on one page

A monthly review that takes an hour

1. Export last month’s work orders with their timestamps from your software.
2. Work out the median for each time measure, split by job level and by business hours or after hours.
3. Count repeat repairs and first-time fixes, per vendor.
4. Compare with the month before. Look for the measure that moved, not the one that is highest.
5. Pick one change for the next month, give it an owner and write it down.
6. Fill in the vendor scorecards at the end of each quarter from the same export.

Share the numbers with property owners in plain words. “Most repairs were confirmed fixed within four days, and two came back” is easier to act on than a chart. The owner posts cover approval limits and what owners want to see.

What not to measure

• Numbers you can’t count from the record, such as how hard a vendor tries
• Averages that mix emergencies with routine jobs
• Ticket counts on their own, which grow with doors and say little about speed
• Anything you won’t act on. A number nobody changes anything for is noise

Where Now Ours fits

Ours answers tenants at any hour in your company’s name, offers jobs to your technicians first and then to your vendors, and books the visit. Every message and booking is recorded on the case with who it went to and when, and “done” means the tenant confirmed the fix or the vendor or technician marked it done with a photo or note. You can export your cases any time as CSV or JSON, so the timestamps this guide asks for are already there.