Every owner has a theory about where their work comes from. The theory is built out of memorable events — the big job that came from the truck wrap, the disaster that came from a lead marketplace — and memorable events are a famously bad sample. The jobs that quietly arrive from the same channel every month are the ones nobody remembers, which is precisely why they get underfunded.
You do not need attribution software to fix this. You need one question asked consistently, one field filled in, and a willingness to look at revenue rather than lead count.
The one question, and where it goes wrong
"How did you hear about us?" is the whole methodology. It fails in practice for three specific reasons, all fixable:
- It is asked at the wrong time. Asked at the end of a booking call, it gets a distracted "the internet". Asked early, conversationally — right after the name — it gets a real answer.
- The answers are free text. "Google", "google search", "internet", and "found you online" become four different sources in your records, and the reporting is worthless. Use a short fixed list of six to eight options plus "other".
- It is optional. Any field that can be skipped will be skipped during the busy weeks, which are the weeks whose data you most need. Make it required to save the record.
Count revenue, not leads
This is the single change that most often reverses an owner's conclusion. A channel producing 40 leads a month at a 12% close and a $600 average ticket generates about $2,900. A channel producing 8 leads at a 60% close and a $3,200 ticket generates over $15,000. On a lead-count report, the first channel looks five times better. It is not.
| Channel A | Channel B | |
|---|---|---|
| Leads per month | 40 | 8 |
| Close rate | 12% | 60% |
| Average ticket | $600 | $3,200 |
| Revenue | $2,880 | $15,360 |
| Monthly spend | $1,200 | $400 |
| Return | 2.4× | 38× |
Lead-count reporting is the default in most tools because leads are easy to count. It systematically flatters cheap, high-volume, low-intent channels — which is exactly the kind of channel that is easiest to sell you more of.
Four numbers per channel, once a quarter
- 01Leads
The raw count. Useful only as an input to the other three, never as the headline.
- 02Close rate
What share became jobs. This is where lead marketplaces and high-intent referrals separate most violently — often by a factor of five.
- 03Revenue
Total booked value, not average. One channel that occasionally produces a $22,000 job is a different asset from one that reliably produces $700 ones, and the average hides that.
- 04Cost per job
Spend divided by jobs won. Not cost per lead — cost per lead is the metric that makes bad channels look affordable.
Four numbers, six channels, once a quarter. It takes an hour if the source field has been filled in, and it is the single highest-return hour in a service business's marketing year.
What the numbers usually reveal
Three findings turn up so often they are almost predictions:
- Referral and repeat work is bigger than assumed and gets no budget at all. It routinely accounts for 40–60% of revenue while receiving none of the marketing spend and no systematic encouragement. A referral engine is cheaper to build than any ad channel and nobody builds it because it does not have a monthly invoice reminding them it exists.
- One paid channel is quietly unprofitable. Usually the one bought most recently, and usually defended on lead volume. Cost per job, rather than cost per lead, is what exposes it.
- The best channel is under-served. The highest-close, highest-ticket source almost always turns out to be capacity-limited rather than demand-limited — meaning more money in it would work, and nobody has tried because it is not the channel anyone talks about.
TARGETS
The part that makes it durable
All of this depends on the source field being filled in on every lead, forever, including in July. That is not a discipline problem to be solved with reminders; it is a design problem. The field must be on the screen where the lead is created, required, and a short pick list — three properties that together take about ten minutes to configure and determine whether you have data in a year.
When it is right, you stop having the annual argument about whether the advertising is working, because it is a number instead of an opinion. And you find out that the answer differs by channel, by season, and by job type — which is the sort of thing a business can actually act on.
Cost per lead is the number that makes a bad channel look affordable. Cost per job is the one that tells the truth.
Common questions
How do you track where leads come from without analytics software?
Ask "how did you hear about us?" early in every booking conversation, record it in a required field with a short fixed pick list rather than free text, and review revenue by source quarterly. That combination gives most small service businesses better attribution than a tracking platform they never configure properly.
Why is lead count a misleading marketing metric?
Because it flatters cheap, high-volume, low-intent channels. A source producing 8 leads at a 60% close and a $3,200 ticket is worth five times one producing 40 leads at 12% and $600 — but only revenue and cost per job make that visible.
What should you measure per marketing channel?
Leads, close rate, total booked revenue, and cost per job won. Cost per lead is the number to avoid leading with, since it is the one that makes an unprofitable channel look affordable.
What do most service businesses find when they track lead sources properly?
That referral and repeat work is a much larger share of revenue than assumed and receives no budget, that one recently-added paid channel is unprofitable when measured per job, and that their best channel is capacity-limited rather than demand-limited.