Most dashboards measure the wrong things
Ask a service business owner what they track and you will usually hear revenue and job count. Ask what decision either number changed last month and the conversation stops.
That is the test. A metric earns its place on a dashboard only if a specific value of it triggers a specific action. Revenue was up eleven percent is not actionable. Answered-call rate fell to 71 percent on Saturdays is actionable — you know what to fix and roughly what it is worth.
As of August 2026, the twelve numbers below are the ones that consistently pass that test in field service businesses. Each entry gives the calculation, a realistic range, and — most importantly — what you do when the number is bad.
The intake numbers
These four govern how much of the demand that exists actually reaches your calendar. They are the cheapest to fix and the most commonly ignored.
1. Answered-call rate
Calculation: answered calls ÷ total inbound calls.
Healthy range: 95 percent or better. Below 85 percent, this is almost certainly your largest single loss.
Count abandoned calls in the denominator. A caller who hung up after six rings was demand you failed to capture, and excluding them is the most common way this metric gets flattered into meaninglessness. The call abandon rate and ring time guide covers how ring duration drives abandonment.
When it is bad: the fix is coverage, not effort. Nobody answers more calls by trying harder while under a sink. Either somebody is on the phone during production hours or something is. The after-hours playbook covers the evening and weekend portion, which in most trades is 20 to 35 percent of total call volume.
2. Booking rate
Calculation: jobs booked ÷ answered calls, excluding wrong numbers, vendors, and existing-customer service questions.
Healthy range: 40 to 60 percent for most trades. Emergency-heavy trades run higher; high-ticket quoted trades run much lower.
When it is bad: the problem is almost always pricing or availability, in that order. Either the number you quote loses to a competitor, or you cannot offer a slot soon enough. Listen to ten recorded calls that did not book and you will know which within an hour. If quoting is inconsistent between whoever answers, that inconsistency is itself the problem — see quoting from a price list.
3. Speed to lead
Calculation: median minutes from a form submission or missed call to a real human or system response.
Healthy range: under five minutes. Under one is better.
This is one of the most reliably predictive numbers in service businesses, and the speed-to-lead guide covers why the decay curve is so steep. The practical shape: a lead responded to in two minutes and the same lead responded to in two hours are not the same lead.
When it is bad: automate the first touch. A missed-call text-back is the cheapest version; answering the call in the first place is the better one, and the comparison between the two covers when each is enough.
4. Cost per booked job, by channel
Calculation: channel spend ÷ jobs booked from that channel.
Healthy range: depends entirely on average ticket. The rule of thumb worth using is that it should be under 10 to 15 percent of the revenue that job produces.
Note carefully: booked job, not lead. The two diverge sharply, especially in trades with a long sales cycle, and reporting only cost per lead is how contractors keep funding channels that generate cheap leads that never buy. The cost per lead guide covers the distinction, and call tracking and attribution covers the plumbing needed to calculate it at all.
The production numbers
These four govern how much revenue you generate from the demand you captured.
5. Technician utilization
Calculation: billable hours ÷ total paid hours.
Healthy range: 60 to 75 percent.
This is the most misunderstood metric in field service. Owners see 65 percent and conclude the crew is idle a third of the day. In reality the non-billable third is travel, material pickup, callbacks, and the gap where a customer was not home. Pushing toward 90 percent generally means overbooking, which produces late arrivals, rushed work, and callbacks — all of which cost more than the utilization gained.
When it is bad: low utilization is nearly always a routing or scheduling problem. Jobs scattered across the service area, poorly estimated durations, or a dispatcher optimizing for fairness rather than geography. The dispatch guide covers the mechanics; the service radius and travel fee guide covers the pricing side of the same problem.
6. Average ticket
Calculation: total revenue ÷ completed jobs, tracked separately by job type.
The aggregate number is nearly useless. Tracked by job type, it tells you which work is worth pursuing and which is filling the schedule at a loss.
When it is bad: either your prices are stale or your technicians are not presenting options. Both are fixable in a week — one by updating the price list, one by changing what gets offered on site.
7. First-time fix rate
Calculation: jobs completed on the first visit ÷ total jobs.
Healthy range: 80 to 90 percent for most trades.
Every point below that is a second drive, a second hour, and a customer whose confidence dropped. The two usual causes are parts availability and incomplete information at intake — the technician arrived without knowing what they were walking into.
When it is bad: look at intake first. If the person or system booking the job is not capturing model, symptom, and access details, the technician is guessing what to load on the truck.
8. Rework and callback rate
Calculation: jobs requiring a return visit for the same issue ÷ total jobs.
Healthy range: under 5 percent.
Callbacks are the quietest profit leak in field service because they rarely get recorded as anything. They show up as a slow week rather than a cost. Tracking them by technician and by job type surfaces both training gaps and job types you should stop selling. The job costing guide covers attributing that cost properly.
The money numbers
These four govern how much of the revenue you actually keep.
9. Days to cash
Calculation: median days from job completion to payment received.
Healthy range: under 3 days for residential, under 30 for commercial.
Residential service businesses that mail invoices routinely sit at 20-plus days for work the customer would have paid for on the spot. That is a self-inflicted financing problem. The payment links guide covers collecting at sign-off.
10. Gross margin by job type
Calculation: (revenue − direct labor − materials) ÷ revenue, per job type.
The aggregate hides everything. A shop can run a healthy blended margin while one popular service loses money on every call, subsidized by the rest. Break it out and the answer is usually obvious and uncomfortable.
11. Revenue per technician per day
Calculation: total revenue ÷ (technicians × working days).
This is the single best measure of whether the operation is actually improving, because it combines utilization, average ticket, and first-time fix into one number. Track it monthly. A business that adds a truck should see this hold steady; if it falls, you added capacity without adding demand.
12. Repeat and reactivation rate
Calculation: jobs from customers who have used you before ÷ total jobs.
Healthy range: 25 to 45 percent for maintenance-adjacent trades, lower for one-time services.
This is the metric that tells you whether you are building a business or renting one from an ad platform. If nearly every job is a stranger, your customer list has no value and your marketing spend can never come down. Fixing it means capturing the customer record at intake and touching it deliberately afterward — reviews, reminders, and outbound follow-up on your own past customers.
The scoreboard
| Metric | Calculation | Healthy range | Review cadence |
|---|---|---|---|
| Answered-call rate | Answered ÷ total inbound | 95%+ | Weekly |
| Booking rate | Booked ÷ qualified answered | 40–60% | Weekly |
| Speed to lead | Median minutes to first response | Under 5 min | Weekly |
| Cost per booked job | Channel spend ÷ jobs booked | Under 10–15% of ticket | Monthly |
| Technician utilization | Billable ÷ paid hours | 60–75% | Weekly |
| Average ticket | Revenue ÷ jobs, by type | Trade-dependent | Monthly |
| First-time fix rate | First-visit completions ÷ jobs | 80–90% | Monthly |
| Rework rate | Return visits ÷ jobs | Under 5% | Monthly |
| Days to cash | Median completion to payment | Under 3 days residential | Monthly |
| Gross margin by job type | (Rev − labor − materials) ÷ rev | Trade-dependent | Monthly |
| Revenue per tech per day | Revenue ÷ (techs × days) | Trending up | Monthly |
| Repeat rate | Returning-customer jobs ÷ jobs | 25–45% | Quarterly |
How the twelve connect to each other
Treating these as twelve independent gauges is the most common mistake. They form a chain, and knowing the chain tells you where to look when a number moves.
Demand captured → work produced → cash kept. Answered-call rate and booking rate determine how many jobs enter the system. Utilization, average ticket, and first-time fix determine how much revenue those jobs generate. Days to cash, margin, and rework determine how much of it you keep.
That structure produces some non-obvious diagnostics:
Utilization up, revenue per tech flat. The crew is busier but the work is worth less. Usually means the schedule is filling with low-ticket jobs because nobody is prioritizing, or that drive time is eating the gain.
Booking rate down, answered-call rate up. Almost always a good sign being misread. Answering more calls means answering more junk calls — vendors, wrong numbers, tyre-kickers — which dilutes the ratio. Recalculate booking rate against qualified calls only before concluding anything is broken.
First-time fix down after adding a technician. New hire ramp, not a systemic failure. Track it per technician for the first ninety days and it resolves itself; track it only in aggregate and it looks like a company-wide decline.
Average ticket rising while repeat rate falls. Frequently means prices moved past what your existing base will pay. Worth catching early, because the repeat number lags by months.
Cost per booked job stable while cost per lead falls. The channel got cheaper at generating leads that do not convert. This is the pattern that makes cost-per-lead reporting actively misleading rather than merely incomplete.
The general principle: a metric that moves alone is usually noise; a metric that moves while its neighbours hold is usually real. Reading them as a chain is what turns a dashboard into a diagnostic.
Benchmarks are directional, not verdicts
The ranges in this guide come from common patterns in field service, and they are useful for spotting a number that is badly wrong. They are not a scorecard.
Trade structure moves them substantially. Emergency-heavy trades book at far higher rates than quoted trades, because the caller has a broken thing right now. High-ticket work with long sales cycles has structurally lower close rates and structurally longer days to cash, and neither is a failure. Wide rural territories will never hit the utilization of a dense urban route, no matter how good the dispatcher is.
The comparison that always means something is your own number last quarter. A booking rate of 38 percent is neither good nor bad in isolation; 38 percent after three months at 46 percent is a problem worth a morning of investigation.
Two habits make that comparison reliable. Freeze the definitions — decide once whether abandoned calls count in the denominator, whether travel counts as billable, what window makes a return visit a callback, and then never quietly change it, because a definition change looks exactly like performance change. And write the number down monthly, somewhere that persists. A dashboard that only shows the current state cannot tell you a trend, and trend is where almost all the useful information lives.
The numbers to stop reporting
Total calls received. Volume without answer rate or booking rate tells you nothing about performance. A month with more calls and worse conversion is a worse month.
Website traffic. Unless it ties to calls and bookings, it is a proxy for a proxy.
Jobs completed. Without margin, this rewards taking cheap work.
Social media followers. No service business has ever traced a booked job to a follower count.
Average review rating in isolation. Rating without recency and volume is misleading — a 4.9 from eleven reviews three years ago loses to a 4.6 from two hundred reviews this year. The reviews guide covers what actually moves the needle.
Getting the data without a data project
Every metric above spans at least two systems: calls, jobs, and invoices. That is why most service businesses do not track them — assembling the numbers by hand from a phone bill, a calendar, and an accounting export is a two-hour job nobody does twice.
The practical requirement is that calls, jobs, and money share one customer record. Once they do, these calculations are queries rather than projects. That is also what makes a natural-language operations layer useful: instead of building a report, you ask which jobs last week had no assigned technician, or which invoices are past thirty days, and get a sentence back. The Jarvis brain covers how that works against live operational data — which matters less because it is clever and more because it is the only version of reporting owners actually use.
If your data lives in four disconnected tools, the all-in-one versus point solutions comparison covers the consolidation tradeoff, and the migration guide covers doing it without losing a week.
Where to start
Do not instrument all twelve. Pick the two intake numbers — answered-call rate and booking rate — and measure them honestly for one month. In most service businesses those two explain more variance in revenue than the other ten combined, and both are fixable faster and cheaper than anything downstream.
Once intake is solid, add utilization and days to cash. Add the rest when you have a specific question they would answer.
Want to see what your current answered-call rate is actually costing? Talk to us, or look at /pricing to see where the intake layer starts.
