Ask a marketing company how things are going and you will get cost per lead. There is a reason for that, and it is not a conspiracy. It is the last number in the chain they fully control.
It is also the easiest number in this industry to improve without helping you at all. Widen the targeting. Soften the form. Drop the qualifying questions. Run a cheaper offer. Cost per lead falls, the report looks better, and your calendar does not change.
A lead is a row in a spreadsheet. An estimate is a person standing in your customer's kitchen. Those are different things, and only one of them has ever paid for a van.
The chain, in order
Every dollar you spend walks the same path. Naming each step is what makes it possible to say where the money is actually going.
- Spend, which buys clicks or impressions.
- Leads, which are people who raised a hand.
- Contacted, meaning somebody had a real conversation with them.
- Booked, meaning a date and time in a calendar.
- Attended, meaning the estimate actually happened.
- Quoted, meaning a written number was delivered.
- Sold.
Cost per lead stops at step two. Cost per booked estimate reaches step four, and cost per attended estimate reaches step five. That extra reach is the whole argument, because steps three, four and five are where most contractors are actually losing the money and nobody is reporting on them.
The definition, and the version to actually use
Cost per booked estimate is the money in, divided by the estimates that got booked. Cost per attended estimate divides by the ones that happened.
Use attended. A booking is an intention. An attended estimate is an event. And be explicit about whether the fee is inside the number, because that is where most arguments between contractors and vendors actually come from. Track both. Label both.
The arithmetic, worked three ways
All three of these are illustrations. The numbers are invented, round, and chosen because they divide cleanly. They are not our results and they are not a projection for your market. Put your own numbers in the same slots.
Illustration one. A month of ad spend, handled reasonably.
| Step | Rate | Result |
|---|---|---|
| Ad spend | $3,000 | |
| Cost per lead | $60 | 50 leads |
| Booked, of all leads | 40% | 20 booked |
| Showed up | 75% | 15 attended |
| Closed | 30% | 4.5 jobs |
| Cost per booked estimate | $150 | |
| Cost per attended estimate | $200 | |
| Cost per sold job | $667 |
Illustration two. Same money. The vendor halves the cost per lead, which looks like a win on every report. The desk cannot keep up with double the volume, so calls go out slower and fewer attempts get made.
- $3,000 spend, $30 a lead, so 100 leads.
- Booking rate falls to 20%, so still 20 booked.
- Show rate falls to 60%, so 12 attended.
- Cost per attended estimate: $250. Up from $200.
The cost per lead halved. The cost of the thing that actually pays you went up by a quarter. That is not a trick question, it is the most common month in this industry, and every report in it looked like an improvement.
Illustration three. Same ads as illustration one, nothing touched in the ad account, but the desk that calls and books gets a script, a cadence and a confirmation sequence.
- $3,000 spend, $60 a lead, so 50 leads.
- Booking rate 50%, so 25 booked.
- Show rate 85%, so 21 attended.
- Cost per attended estimate: $143. Down nearly 30% from illustration one.
Nobody changed a headline, a bid or an audience. The whole gain came from steps three, four and five, which is exactly the part most marketing companies do not touch and do not report on. The mechanics of getting there are in the appointment setting playbook.
What you are allowed to pay, worked from your margin
Most contractors judge cost per lead against a number they heard somewhere. That is backwards. The only number that matters is the one your own margin permits.
An illustration, again with invented round numbers.
- Average job sells for $6,000.
- Gross margin is 35%, so the job leaves $2,100 of gross profit.
- You decide you will spend a quarter of that gross profit to buy the job. That is $525 of allowable cost per sold job. This is a decision, not a rule. Pick your own share.
- You close 30% of attended estimates, so one job takes 3.3 of them. Allowable cost per attended estimate: about $159.
- You attend 30% of the leads you are handed, so one attended estimate takes 3.3 leads. Allowable cost per lead: about $48.
Now you have a number to hold a vendor to that came from your own books instead of from a blog. If your allowable cost per lead is $48 and your market delivers leads at $70, that is not a vendor problem. It is an arithmetic problem, and it has three honest answers: sell a bigger average job, close a higher share, or accept fewer jobs. To watch those inputs move against each other, the calculator on the homepage runs the same chain: run your numbers.
The one page scorecard
Ask for this weekly, in this shape, from whoever runs your advertising. If they cannot produce it, that is the answer to a different question you were about to ask.
- Ad spend, and fees, on separate lines.
- Leads delivered.
- Leads contacted, and the median time to first call.
- Estimates booked.
- Estimates attended.
- Jobs quoted, and total quoted value.
- Jobs sold, and total sold value.
- Cost per booked estimate, and cost per attended estimate.
- Cost per sold job.
Nine lines. It fits on half a page and it takes ten minutes to produce if the tracking was set up properly. The reason most vendors do not send it is not effort.
How to instrument this in a week
None of this needs new software. It needs five things set up once and then left alone.
- One pipeline, five stages, named exactly as below. New, Contacted, Booked, Attended, Sold. Not seven stages, not clever names. If two people can disagree about what a stage means, the report is worthless.
- Tag every source at entry. UTM parameters on every ad link and a separate tracking number per channel. If you cannot tell a Google lead from a Meta lead from a yard sign, you cannot tell which one to spend more on.
- Capture the time of the first outbound attempt automatically. Median time to first call is a report, not a memory, and it is the number that moves the whole chain.
- Send the Booked and Sold stages back to the ad platforms. Both Google and Meta accept offline conversions. Once the platform is learning from booked estimates rather than from form fills, it starts finding the people who answer their phone. This is the most under-used tool available to a contractor and it is free.
- One sheet, same nine lines, same day every week. The discipline matters more than the tool. A spreadsheet somebody fills in on Monday beats a dashboard nobody opens.
The definitions have to be written down first
Every one of these words means something slightly different to the person paying and the person being paid. Agree them in writing before the first invoice, not during the first argument.
| Word | What it has to mean |
|---|---|
| Lead | Real name, reachable number, inside your service area, for work you actually do. Nothing else is billable |
| Contacted | A two way conversation with a human. Not a voicemail |
| Booked | Date, time window, address confirmed, and the homeowner acknowledged it |
| Attended | Somebody from your company was at the property and did the measure |
| No show | Attended was attempted and nobody was home. Decide in advance whether it is billable |
| Sold | Signed, with the deposit taken |
Who owns which link
Half of the fights between a contractor and a marketing company are really arguments about which link broke. Settle that up front and most of the heat goes out of it.
| Link | Owner | What breaks it |
|---|---|---|
| Spend to leads | The advertiser | Weak creative, wrong offer, no negative keywords |
| Leads to contacted | Whoever calls | Slow first call, two attempts and stop |
| Contacted to booked | Whoever calls | Selling the floor instead of booking the measure |
| Booked to attended | The confirmation sequence | No reminders, and booking a week out |
| Attended to sold | You | Price, presentation, product range |
The last row is the only one that is entirely yours, and it is the subject of from estimate to signed contract.
The two things a vendor says when the number goes bad
"Your close rate is the problem." Sometimes true, and there is a clean test. Compare your close rate on their leads against your close rate on referrals and repeat customers over the same period. Same closer, same kitchen table, different source. If the gap is wide, the problem is upstream of the close, which is the argument in "I need to think about it" means you lost the estimate.
"Your market is expensive." Possible, and it is checkable rather than debatable. Ask what a click costs now versus three months ago, and what changed. "Expensive market" as a feeling is not an answer. As a number, it is a real one.
The short version
Cost per lead measures the vendor's convenience. Cost per booked estimate measures whether anything is happening. Ask for the second one, in writing, weekly, with the definitions agreed in advance and the fee shown on its own line.
A vendor who will not be measured past the first link in the chain has told you exactly what they think their job ends at. That is useful information, and it costs nothing to ask for. The rest of the questions worth asking are in seven questions to ask any agency and own your stack.
