Every service business owner suspects that some of what they spend produces work and some produces nothing — and has no reliable way to tell them apart, so the budget gets set by whichever month felt busy.
In this article
- Half your marketing works. Nobody knows which half.
- The question that costs nothing
- Making the middle of the chain visible
- The four numbers that decide the budget
- A year, in numbers
- Reputation is a channel, and it can be measured
- How to set next month's number
- Why cutting too fast is the expensive mistake
- Counting the second job
- Five mistakes that waste the budget
- The numbers to track
- Frequently asked questions
Half your marketing works. Nobody knows which half.
Every service business owner has the same suspicion: some of what they spend on marketing produces work, and some of it produces nothing, and there is no reliable way to tell them apart. So the budget gets set by feeling. The channel that felt busy last month gets more; the one that felt quiet gets cut — sometimes exactly when it was starting to work.
The reason this persists is not laziness. It is that the standard measurements are borrowed from businesses that sell online, where the click and the purchase happen in the same session. In a service business the customer sees an ad on Tuesday, calls on Thursday, has the work done the following week and pays on completion. Nothing connects those events unless someone deliberately connects them.
Spend → calls → booked jobs → completed revenue. Most companies measure the first and the last and guess at the middle. The two numbers in the middle are where every real answer lives — because a channel that produces cheap calls that never book is more expensive than a channel that produces expensive calls that always do.
The question that costs nothing
Before any software, there is one habit worth more than all of it: asking every caller how they found you, and writing the answer down in the same field, every time.
Three rules make the answer usable:
- Ask it as part of helping, not as a survey. "Just so I know where to send you — how did you come across us?"
- Record it in a fixed list, not free text. Ten options that everyone uses beats a hundred variations of the same three.
- Accept that it is imperfect. People say "Google" when they mean a review site, and "a friend" when they saw the truck. Directionally it is still the most valuable data most service companies could have and do not.
This one field, filled consistently for ninety days, answers more questions than most analytics dashboards.
Keep reading — free
The rest of this article is worth the 20 seconds
Tell us where to send it and the full piece unlocks right here, along with everything else on this blog.
Rather just talk? Message us on WhatsApp.
Making the middle of the chain visible
Three mechanisms turn guesswork into measurement, in increasing order of effort:
Distinct phone numbers by channel. A different tracking number on the truck, on the yard sign, in the ad and on the website makes call volume attributable without asking anyone anything. It is the single most effective step available to a service business.
A booking field in the job record. Every job carries the source it came from, all the way through to the invoice. That is what allows revenue — not calls — to be attributed.
Consistent naming. The same source labels in the phone system, the scheduling software and the invoice. Most measurement failures are not missing data; they are three systems calling the same thing three different names.
The four numbers that decide the budget
- Cost per call, by channel. Spend divided by calls generated. Useful, and by itself misleading.
- Booking rate, by channel. The share of those calls that became scheduled jobs. This is where channels separate violently — the same money can buy callers who are ready and callers who are shopping.
- Cost per booked job, by channel. Spend divided by jobs, which is the number that should set the budget.
- Average ticket, by channel. The forgotten one. A channel producing $180 jobs and a channel producing $900 jobs cannot be compared on cost per job alone.
Multiply the last two and you get the only conclusion worth acting on: revenue per dollar spent, by channel. Everything else is an input to that.
A year, in numbers
A service company spending $68,000 a year across four channels, average ticket $430 overall.
- Search ads: $31,000 → 610 calls ($51/call) → 64% booked → 390 jobs → $79/job, avg ticket $465 → $5.85 revenue per dollar
- Local service listings: $14,000 → 340 calls ($41/call) → 71% booked → 241 jobs → $58/job, avg ticket $390 → $6.72 per dollar
- Direct mail: $16,000 → 195 calls ($82/call) → 58% booked → 113 jobs → $142/job, avg ticket $610 → $4.31 per dollar
- Truck, signs and referrals: $7,000 → 280 calls ($25/call) → 81% booked → 227 jobs → $31/job, avg ticket $395 → $12.81 per dollar
Two conclusions that were invisible before the measurement. Direct mail looked expensive on cost per call and is worse than it looks on cost per job — but it brings the highest average ticket, so it is not the failure it appeared to be, it is a channel for a specific kind of work. And the cheapest revenue in the company, by a wide margin, comes from the truck, the signs and the referral system — the things that get the smallest budget and the least attention.
Not "cut direct mail." Doubled the smallest line. Moving $6,000 from direct mail into vehicle graphics, yard signs and a structured referral request produced 190 additional jobs the following year — because the channel returning $12.81 per dollar had been the one nobody was funding.
Reputation is a channel, and it can be measured
Reviews are treated as something that either happens or does not, which is why they rarely appear in a marketing budget. They behave like a channel: they have a cost, a volume and a conversion rate, and they respond to effort.
What makes them measurable:
- Requests sent — how many customers were actually asked, as a share of completed jobs. Most companies discover this number is under 20% and assumed it was most.
- Review rate — reviews received divided by requests sent, which says whether the ask is working.
- Recency — how many arrived in the last ninety days, which matters more to a prospect than the lifetime total.
- Calls attributed to listings, which is where reviews convert.
The mechanics that move those numbers are unglamorous: ask at the moment the work is finished and the customer is satisfied, send a direct link rather than instructions, ask from the technician rather than the office, and follow up once. A company that lifts its request rate from 20% to 70% of completed jobs usually sees listing calls climb within a quarter — at a cost of essentially nothing, which is why it beats every paid channel on the revenue-per-dollar comparison.
How to set next month's number
Once the four metrics exist, budgeting stops being a debate. A workable method:
- Rank channels by revenue per dollar spent, not by size or by habit.
- Fund capacity, not preference. The best channel gets more money until it stops absorbing it — many local channels have a ceiling, and pushing past it wastes spend.
- Keep one experiment running, sized so that losing it entirely would not hurt, and give it the full ninety days.
- Set the total against booked capacity. Generating more calls than the crew can serve produces poor service and bad reviews, which costs more than the wasted spend.
That last point is the one most often missed. Marketing that outruns the schedule is not growth; it is a reputation problem being purchased at full price.
Why cutting too fast is the expensive mistake
Service marketing has a lag that online businesses do not. A yard sign works for a year. A review posted in March produces calls in September. A customer who saw the truck three times before calling cannot say which time mattered.
Two practical consequences:
- Give a channel at least ninety days before judging it, and longer for anything reputation-based.
- Do not change three things at once. If the budget, the message and the landing page all change in the same month, the result teaches nothing.
The corollary is that seasonality has to be removed from the comparison. Comparing July to November in a trade with a summer peak measures the calendar, not the channel.
Counting the second job
Cost per job compares channels on the first transaction, which understates the ones that bring customers who stay.
The simple version, which is enough: for each channel, look at the customers acquired twelve months ago and add up everything they have spent since. A channel with a higher cost per job that produces customers who return twice is cheaper than a channel with a low cost per job that produces one-time work.
In most service businesses, referral and reputation channels win this comparison decisively — which is the second reason they deserve more budget than they get.
Five mistakes that waste the budget
- Measuring leads instead of booked jobs. It rewards the channels that generate the most noise.
- Ignoring average ticket by channel. It makes the channel that brings the biggest jobs look like the worst performer.
- Judging a channel in thirty days. The lag guarantees a wrong conclusion.
- Letting the source field go unfilled. Ninety days of blanks makes the entire year unmeasurable.
- Cutting the cheapest channel because it has no invoice. Referrals and vehicle graphics have costs that are easy to overlook and returns that are hard to beat.
The numbers to track
- Cost per booked job, by channel, reviewed monthly.
- Average ticket, by channel, reviewed alongside it.
- Revenue per dollar spent, by channel — the number that sets next month's budget.
- Percentage of jobs with a source recorded, which should be above 90% before any conclusion is drawn.
- Total marketing spend as a share of revenue, tracked over time rather than judged against a benchmark.
- Twelve-month revenue per customer, by acquisition channel.
Measuring marketing in a service business does not require sophisticated tools. It requires one field filled every time, distinct phone numbers, and the discipline to compare channels on booked jobs and average ticket rather than on how busy the month felt. That alone usually reveals that the best-performing channel in the company is the one receiving the least money.
Find out which channel is actually paying for itself
Most service companies compare channels on cost per lead, which rewards whichever one generates the most noise. Send us a year of spend by channel with your job records and we will build the revenue-per-dollar comparison.
Frequently asked questions
Why is marketing so hard to measure in a service business?
Because the standard measurements are borrowed from businesses that sell online, where the click and the purchase happen in the same session. In a service business the customer sees an ad on Tuesday, calls on Thursday, has the work done the following week and pays on completion — and nothing connects those events unless someone deliberately connects them. The chain that matters is spend, then calls, then booked jobs, then completed revenue. Most companies measure the first and the last and guess at the middle, which is exactly where the answers live: a channel producing cheap calls that never book is more expensive than a channel producing expensive calls that always do. Making the middle visible requires distinct phone numbers by channel, a source field on every job record carried through to the invoice, and consistent naming across the phone system, the scheduling software and the invoice.
Which four numbers should set the marketing budget?
Cost per call by channel, which is useful and by itself misleading. Booking rate by channel, which is where channels separate violently, because the same money can buy callers who are ready and callers who are shopping. Cost per booked job by channel, which is spend divided by jobs rather than by leads. And average ticket by channel, the one most often forgotten — a channel producing $180 jobs and a channel producing $900 jobs cannot be compared on cost per job alone. Multiplying the last two gives the only conclusion worth acting on, which is revenue per dollar spent by channel. Everything else is an input to that. In practice, at least 90% of jobs need a recorded source before any of these conclusions can be trusted.
What does a real channel comparison look like?
In a company spending $68,000 across four channels with an overall average ticket of $430: search ads at $31,000 produced 610 calls at $51 each, 64% booked, 390 jobs at $79 per job with a $465 average ticket — $5.85 of revenue per dollar. Local service listings at $14,000 produced 340 calls at $41, 71% booked, 241 jobs at $58, ticket $390 — $6.72 per dollar. Direct mail at $16,000 produced 195 calls at $82, 58% booked, 113 jobs at $142, but the highest ticket at $610 — $4.31 per dollar. And truck graphics, signs and referrals at $7,000 produced 280 calls at $25, 81% booked, 227 jobs at $31, ticket $395 — $12.81 per dollar. The owner's move was not to cut direct mail but to double the smallest line, shifting $6,000 into vehicle graphics, yard signs and a structured referral request, which produced 190 additional jobs the following year.
How long should a channel run before judging it?
At least ninety days, and longer for anything reputation-based, because service marketing has a lag that online businesses do not. A yard sign works for a year. A review posted in March produces calls in September. A customer who saw the truck three times before calling cannot say which time mattered. Two practical consequences follow: give a channel the full ninety days before deciding, and do not change three things at once — if the budget, the message and the landing page all change in the same month, the result teaches nothing. Seasonality also has to be removed from the comparison, since comparing July to November in a trade with a summer peak measures the calendar rather than the channel. Cutting too fast is the expensive mistake, and it usually happens right when a channel is starting to work.
Can reviews be treated as a marketing channel?
Yes, and they should be, because they have a cost, a volume and a conversion rate and they respond to effort. Four numbers make them measurable: requests sent as a share of completed jobs, which most companies discover is under 20% when they assumed it was most; review rate, meaning reviews received divided by requests sent, which says whether the ask is working; recency, since how many arrived in the last ninety days matters more to a prospect than the lifetime total; and calls attributed to listings, which is where reviews convert. The mechanics that move those numbers are unglamorous — ask at the moment the work is finished and the customer is satisfied, send a direct link rather than instructions, ask from the technician rather than the office, and follow up once. Lifting the request rate from 20% to 70% usually raises listing calls within a quarter at essentially no cost.