Maintenance Performance Metrics for Rental Portfolios

Metrics versus KPIs: a distinction that saves time

The two terms are often used interchangeably, but there is a useful difference. Engineering consultancy GEI Consultants, drawing on the terminology of the Society for Maintenance & Reliability Professionals, describes metrics as quantitative values used to track specific processes, and a KPI as a high-level indicator of what matters most. In practice, you might record a dozen metrics and promote only three or four to KPIs that your executive team reviews monthly.

A second filter comes from Fiix, a maintenance software company, which cautions against collecting maximum data: if a number does not inform a specific decision or change an outcome, it is not a KPI. Apply that test to everything below. If nobody would act on a rising cost per unit or a slow vendor, stop reporting it.

A peer-reviewed review of maintenance performance measurement adds a caution: maintenance is complex enough that numbers need qualitative judgment next to them. A property manager who knows that a particular boiler is on its last season will read the data better than a spreadsheet will.

The nine metrics to track

The formulas below are standard ways of calculating each measure. Adapt the definitions to your own workflow, then keep them fixed so trends are comparable month to month.

1. First response time

What it measures: the time between a tenant reporting an issue and a meaningful reply, meaning an acknowledgement with a next step, not an automated receipt.

How to calculate it: timestamp of first substantive reply minus timestamp of the request. Report the median and the 90th percentile, because averages hide the overnight and weekend requests tenants remember.

How to read it: this is the metric tenants feel most. It also drives volume, because tenants who hear nothing call again, and each repeat contact adds work for your team. Track after-hours response separately from business hours.

2. Average resolution time

What it measures: how long a request stays open from report to confirmed completion.

How to calculate it: completion date minus request date, averaged across closed requests. Split it by category (plumbing, heating, appliances, locks) and by priority, since a blended number mixes a dripping tap with a failed furnace.

How to read it: look at the tail, not just the mean. A handful of requests that sit for weeks waiting on a part or a tenant reply can say more about your process than the average does. Break the total into stages (waiting on tenant, waiting on vendor, waiting on approval) to find where time actually goes.

3. Maintenance cost per unit

What it measures: total maintenance spend divided by the number of units, usually monthly or annually.

How to calculate it: repairs, vendor invoices and in-house labour for the period, divided by occupied units. Keep routine repairs and capital replacements in separate lines, otherwise one new roof makes a good year look terrible.

How to read it: compare a building against its own history and against similar buildings in your portfolio, not against a generic figure. Age, heating system type and unit mix all change what is normal. If you want a quick view of what your own numbers imply, our rental maintenance cost calculator is a starting point.

4. AI troubleshooting success rate

What it measures: among requests where automated troubleshooting was attempted, the share where the tenant confirmed the problem was fixed.

How to calculate it: requests closed after guided troubleshooting, divided by requests where troubleshooting was offered. Count a request only when the tenant confirms the outcome or no repeat request follows within a set window, such as seven days.

How to read it: the denominator matters. A very high rate can mean the system is only attempting easy fixes, which is fine, or that tenants are closing conversations to get rid of them, which is not. Pair it with the repeat-request rate for the same unit and issue. Typical candidates for guided steps are a tripped breaker, a reset button on a garburator or a thermostat setting, which is why troubleshooting by text first works for some issues and never for others, such as gas smells or active leaks.

5. Percentage of requests resolved without vendor dispatch

What it measures: the share of all requests closed without sending a tradesperson.

How to calculate it: requests closed with no vendor visit, divided by total requests in the period.

How to read it: this is the metric that connects most directly to spend, because every avoided visit saves a service fee and the coordination time around it. It is not a target to maximize blindly. Pushing it too high can mean real faults are being deferred. Watch it alongside the reopen rate, meaning requests that return within 30 days for the same problem. If both the no-dispatch share and reopen rate rise, the troubleshooting is closing things that are not fixed.

6. Vendor response time

What it measures: how quickly a vendor accepts a job or proposes a time after being contacted.

How to calculate it: vendor acceptance (or first booking proposal) timestamp minus dispatch timestamp, by vendor and by trade.

How to read it: emergencies and routine work should be measured separately. A plumber who answers urgent calls within the hour but takes three days on routine quotes may still be your best emergency resource. The goal is to know which vendor to call first for which job, not to rank everyone on one number.

7. Vendor completion time and first-visit fix rate

What it measures: the time from vendor acceptance to finished work, and whether the first visit solved the problem.

How to calculate it: completion time is the finish timestamp minus acceptance. First-visit fix rate is jobs closed on the first visit divided by all dispatched jobs.

How to read it: repeat visits cost twice, in invoices and in tenant patience. A vendor with fast arrival but a low first-visit fix rate is often slower in practice than one who takes a day longer and finishes the job. Add invoice amount against the category average to spot pricing drift.

8. Tenant satisfaction

What it measures: how tenants rate the maintenance experience after a request closes.

How to calculate it: a one-question rating sent after completion, with the response rate reported next to the score. A score from 5 percent of tenants is not the same as one from 40 percent.

How to read it: this is the quality check that speed metrics lack. The City of Toronto’s housing agency was told this directly: an audit of Toronto Community Housing found that its 39 property-management KPIs focused mostly on timeliness, such as routine work orders completed within five business days, and recommended adding measures of work quality and tenant satisfaction. A work order closed on time is not necessarily a problem solved.

9. Staff intervention rate

What it measures: the share of requests that need a person on your team to step in, whether to approve spend, resolve a conflict, chase a vendor or answer a tenant.

How to calculate it: requests with at least one staff action, divided by total requests. For more precision, record staff minutes per request.

How to read it: this is the scalability metric. If your portfolio grows 30 percent and staff touches per request stay flat, your team’s workload grows with the portfolio. If the rate falls, the same team can carry more doors. It is the clearest link between maintenance coordination and headcount.

The nine at a glance

What “good” looks like: reading benchmarks carefully

Tenants and managers often ask for a benchmark. Canada has few universal ones, and the ones that exist come from specific regulatory or social-housing contexts.

• Emergencies: under Toronto’s RentSafeTO programme, landlords must respond to urgent service requests within 24 hours, including loss of heat, water, hydro or gas, and breaches of building security. That is a municipal rule for Toronto, not a national one. Check your own municipality and province, whether that is Nova Scotia, New Brunswick, Newfoundland and Labrador or Prince Edward Island, for the applicable standard.
• Routine work: Ottawa Community Housing reports responding to 99.48 percent of regular maintenance requests within five business days since the start of 2022. It is a useful illustration of how a large landlord states a service level, but a social-housing provider with its own staff and processes is not a direct comparison for a private portfolio.
• Regulatory reporting: Toronto also publishes annual RentSafeTO performance information on evaluations, service requests and enforcement, a reminder that maintenance performance is something tenants and regulators can see.

The practical conclusion: your first benchmark is your own last quarter. Set a baseline, set a service standard for each priority class that meets your legal duties, and improve against yourself. Building type matters too. As SiteClara notes, needs differ by asset type, so a 1960s walk-up and a 2015 mid-rise should not share one set of expectations.

From metrics to NOI

Metrics only pay for themselves when they change a decision. Here is how each of the nine connects to the operating statement.

Lower operating costs. CMHC’s operations manual for multi-unit residential buildings says that making a few daily tasks more efficient could cut annual operating costs by 25 percent or more. That is a general statement about operations, not a promise for any one portfolio, but it points at the right place to look: repetitive daily work. Intake, triage, scheduling and follow-up are repetitive by nature. Measuring the staff minutes they consume is the first step to shrinking them.

Fewer avoidable visits. Every request resolved without dispatch saves a service call. Even a small improvement in that share compounds across thousands of units, so the metric deserves a monthly review.

Better vendor decisions. Response time, completion time and first-visit fix rate turn vendor selection from habit into evidence. You can reward reliable trades with more work and have a documented conversation with the ones who slip.

Fewer surprises. The ratio of planned to reactive work is a leading indicator. As SiteClara explains, a rising reactive share often means preventive maintenance is being skipped or equipment is ageing. If your HVAC work is mostly reactive, scheduled checks that are specific and measurable help. These measurable rooftop unit checks show how a preventive routine can be turned into numbers a manager can track.

Scaling without proportional headcount. Staff intervention rate and minutes per request tell you how much of the coordinator’s day is spent on routine back-and-forth. If that falls while volume rises, you add doors without adding coordinators.

A worked example

Take a hypothetical 600-unit portfolio, with illustrative numbers rather than benchmarks. Suppose last quarter’s records show that 1,500 requests came in, that a coordinator touched nearly every one, and that the median first response outside office hours was the next morning. That points to three actions: set up after-hours intake, measure how many requests are simple enough to troubleshoot by text, and record staff minutes per request. Next quarter you compare. If the no-dispatch share rose, reopen rate stayed flat and staff touches per request fell, you have evidence of real efficiency. If reopen rate rose, you have evidence the opposite happened. Either way, you are deciding with data.

Why the data is hard to collect, and how SMS fixes it

Most portfolios receive requests through several channels: phone messages, emails, a tenant portal that some tenants never open, and texts to a manager’s personal phone. Each channel records a different timestamp, or none. Vendor updates arrive by phone call. Closing the loop with the tenant is manual. When you try to calculate resolution time you are stitching together fragments, and the numbers are only as good as someone’s memory.

Structured, single-channel intake changes this. When every request begins as a text conversation, the first message is the start timestamp, the reply is the response timestamp, the troubleshooting outcome is on record, and the dispatch, scheduling, follow-up and completion all sit in one thread tied to a unit and an issue category.

How GoLando handles it

GoLando is a Canadian-owned and operated AI maintenance coordination platform. Tenants text us to report an issue, with no app, no portal and no phone call. The AI handles the conversation, troubleshoots where it is safe to do so, identifies emergencies, and coordinates repairs with approved service providers when professional help is needed. Scheduling, follow-ups and job completion are part of the same workflow, so request activity and vendor performance are recorded as the work happens.

A central dashboard gives managers visibility into maintenance requests, vendor activity, property maintenance history, expenses and performance metrics. That is the raw material for the nine metrics above. It sits alongside your existing property management software, since GoLando integrates with Buildium and is designed to complement the workflows you already use rather than replace them. See how tenant requests by text work, and how automated vendor dispatch keeps your own trades or the optional GoLando Vendor Network moving.

What we have learned about maintenance metrics

We build maintenance coordination software, so here is our view of how portfolios get this wrong.

The most common mistake is measuring what the system makes easy to measure. Most tools can tell you how many work orders were closed. Fewer can tell you how many should never have become work orders. We think the most valuable question in a rental portfolio is not how fast you dispatched, but whether you needed to dispatch at all. That is why we put the no-dispatch share, the repeat-request rate and the staff intervention rate next to the traditional speed metrics.

The second mistake is treating tenant communication as separate from maintenance performance. A tenant who gets an answer in a minute at 11 pm and a clear next step is a different customer from one who leaves a voicemail. Response time is a service metric and a cost metric at once.

The third is trusting automation without checking it. Any AI troubleshooting rate should be read with the reopen rate and tenant satisfaction beside it. We would rather you see a lower, honest number than a flattering one.

Finally, metrics support judgment, they do not replace it. Property managers know their buildings. Good data lets them spend that knowledge on the buildings and vendors that need it, instead of on chasing texts.

— Ryan

Putting a KPI programme in place

1. Choose four or five KPIs first. A reasonable starter set is first response time, resolution time, cost per unit, no-dispatch share and tenant satisfaction. Add the rest when the first set is stable.
2. Write definitions down. Decide what starts and stops each clock, what counts as an emergency, and which costs count as maintenance.
3. Capture a baseline. Run one quarter without trying to improve, so you know the starting point.
4. Segment. Report by building, issue category and priority. Portfolio-wide averages hide the one building dragging everything down.
5. Review monthly, act quarterly. Monthly reviews catch problems. Quarterly reviews are for changing vendors, budgets and processes.
6. Pair speed with quality. Never show a timeliness metric without a quality or satisfaction measure beside it.

See your maintenance metrics in one place

If you are coordinating maintenance across hundreds or thousands of units, GoLando gives your team a text-based front door for tenants and a dashboard for the numbers. Plans are tiered by portfolio size: Starter is $39 per month, Pro is $69, Pro+ is $149, Mogul is $249, Scale is $449, and Enterprise starts from $999 per month, all in CAD, with annual billing available. Review the plans and pricing, or see how maintenance automation for property managers fits an existing workflow at golando.ca.