Every marketing report a contractor receives ends in the same place: leads, and what each one cost. The agency’s dashboard says it. The Google Ads account says it. The lead marketplace invoice says it. And the owner, who has a truck payment and a payroll, reads that number and has no idea whether it is good.
That is because a lead is not something you can deposit. A lead is a phone number and a problem. Between it and money sit at least three steps, and each of them can fail for reasons that have nothing to do with the ad.
Why cost per lead misleads
Cost per lead measures the top of the funnel and stops. It cannot see:
- Whether the lead was answered. A call that rang out at lunch counts as a lead on the report.
- Whether it was the right work. A lead for a job you do not do, in a city you do not serve, counts the same as a lead for a repipe down the road.
- Whether it booked. A price shopper who wanted a number over the phone is one lead. So is the homeowner who booked a Tuesday install.
- What the job was worth. A drain clearing and a sewer line replacement are both one lead.
So cost per lead rewards whatever produces the most calls, not the most work. A channel that sends a flood of cheap, wrong-trade calls will beat one that sends a handful of expensive, right ones, right up until you look at the bank account.
It also hides problems inside the building. If half the leads from a campaign reach voicemail, cost per lead does not change at all. Cost per booked job doubles. That difference is the reason to track the second number: it is the only one that notices when the office, rather than the ad, is the problem. We wrote about telling those two apart in your ad spend is not the problem.
The funnel, one step at a time
Cost per booked job is not a new metric so much as the old one carried down the page. Each step is a count you probably already have somewhere:
| Step | Where the count lives |
|---|---|
| Spend | The ad platform, the marketplace invoice |
| Leads | Call tracking, form submissions, platform lead reports |
| Answered leads | The phone system’s call log |
| Qualified leads | Whoever answered: right trade, right area, real job |
| Booked jobs | The field-management or scheduling system |
| Revenue | Invoices, by job |
Divide spend by each count and you get a cost at every stage. The one that matters most is spend divided by booked jobs. The one that tells you where to look is the step where the count drops fastest.
A worked example
The numbers here are invented to show the shape of it. Put your own in the same places.
Say an electrical contractor spends the same amount each month on two channels, and call it $3,000 apiece for round arithmetic.
| Channel A | Channel B | |
|---|---|---|
| Spend | $3,000 | $3,000 |
| Leads | 100 | 40 |
| Cost per lead | $30 | $75 |
| Answered | 70 | 36 |
| Qualified | 30 | 30 |
| Booked | 12 | 18 |
| Cost per booked job | $250 | about $167 |
On cost per lead, Channel A looks two and a half times better. On cost per booked job, Channel B is ahead, and that is before anyone asks what the jobs were worth. If Channel B’s bookings skew toward panel upgrades and Channel A’s toward “can you come look at an outlet”, the gap widens again.
Nothing in that table required new software. It required someone to put five counts next to each other for the same month.
Where the join usually breaks
In practice the chain breaks at one of two links.
The first is lead to answered. Ad platforms generally report a call when it connects to your number or runs past a set length, not when a person speaks. If your phone system does not keep a log you can export, or nobody looks at it, you cannot tell answered from missed. That is fixable with the phone system you have, most of the time. Call tracking without lying to yourself goes through the pitfalls.
The second is booked to source. The field software has a “lead source” dropdown, and whoever books the job picks whatever is at the top of it, or “Google”, which covers four different channels. The fix is dull: fewer, clearer options, and the source filled in from the tracking number rather than from memory wherever the software allows it.
You do not need every job attributed perfectly. You need enough of them attributed consistently that the comparison between channels holds up month to month.
Using the number
Once you have cost per booked job by channel, three decisions get easier.
- Where to put the next dollar. It goes to the channel with room to grow and the lowest cost per booked job, weighted by ticket size.
- When to fix the office instead of the ads. If a channel has a healthy cost per qualified lead and a poor cost per booked job, the leak is in answering, quoting or follow-up, not in the targeting.
- What to ask an agency. “What did it cost per booked job?” is a fair question. If the answer is “we don’t have access to that”, then between you, you need to fix that before anyone judges the campaign.
Keep the number humble. A month is a small sample for a small shop, seasons move it, and a single large job can make a channel look brilliant. Look at it over a quarter, and look at the trend more than the value.
If you want to run your own figures through the same arithmetic, the ad ROI calculator takes spend, leads, booking rate and average job value and turns them into cost per booked job.