Cost per sale is total campaign cost divided by units the campaign actually caused — and both halves are where stores go wrong. The cost side has to include the handling cost, not just the media invoice. The units side has to exclude the sales you would have made anyway. This page works one 10,000-piece direct mail campaign end to end, with every input shown, and arrives at a defensible number: $541 per sale on a conservative reading, $364 if the campaign is measured the way most vendors report it. The gap between those two figures is the whole subject.
Why Cost Per Lead Is the Wrong Number
Cost per lead is popular because it is easy and because it flatters whoever bought the leads. It is also actively misleading, for a structural reason: it can always be improved by buying worse leads. A source at $12 a lead that closes at 2% costs $600 a sale. A source at $40 a lead that closes at 12% costs $333 a sale. The expensive source is the cheap one.
Cost per sale is the number that cannot be gamed this way, because it has the conversion baked in. It is harder to calculate honestly, which is the only reason it is less popular.
The Campaign
A conquest and equity mailing, 10,000 pieces, dropped into a franchise store’s primary market. Numbers are illustrative but every rate is inside the bands we publish in our benchmark table, so the arithmetic is reproducible against your own figures.
The cost side
| Input | Basis | Cost |
|---|---|---|
| List | 10,000 records, targeted selection | $1,200 |
| Print and production | 10,000 pieces | $3,900 |
| Postage | 10,000 pieces, presorted | $3,600 |
| Creative and setup | One-time | $800 |
| Call tracking | Numbers plus recording, one month | $150 |
| Media subtotal | $9,650 | |
| BDC handling | 300 responses × ~22 min fully loaded | $2,350 |
| Total campaign cost | $12,000 |
The handling line is the one almost always missing. Three hundred responses do not answer themselves. Someone takes the call, works the follow-up, confirms the appointment and re-works the no-shows, and those hours are as much a cost of the campaign as the postage. Leave it out and every campaign looks 20% cheaper than it is.
The response side
At a 3% response rate, 10,000 pieces produce 300 responses. Applying the middle of our bands: 65% reach a conversation (195 contacts), 50% of those set an appointment (98 appointments), 75% show (73 shows), and 45% of shows buy.
That is 33 units attributable to the campaign window.
The Two Answers
$12,000 over 33 units is $364 per sale. That is the number a vendor report will show you, and it is not dishonest so much as incomplete.
The problem is the last step. Some of those 33 buyers were already coming. They had a lease maturing, or they had been in twice already, and the mailer arrived in the same fortnight and got the credit. If you hold out a control group — a matched sample of the same list that receives nothing — and that control produces sales at a rate equivalent to 6 of your 33, then the campaign caused 27 units.
$12,000 over 27 units is $444 per sale.
And if you charge the campaign with its share of fixed marketing overhead — say $2,600 of the month’s retained and platform costs allocated across active campaigns — the fully loaded figure is $14,600 over 27, or $541 per sale.
Three defensible numbers from one campaign: $364 reported, $444 incremental, $541 fully loaded. None is wrong. What is wrong is quoting the first while implying the third.
What To Compare It Against
A cost per sale means nothing on its own. It has to sit against two other figures.
The first is front-end gross plus the service and parts contribution you can reasonably expect from that customer. If average front gross is $2,100 and you assign even a conservative lifetime service contribution, $541 is comfortable. At $1,200 front gross on a used-heavy mix, it is a different conversation.
The second is your other channels, calculated the same way. This is the part stores skip. Comparing direct mail at a fully loaded $541 against digital at a media-only $300 is not a comparison at all — it is two different formulas. Load both or load neither.
When a High Cost Per Sale Is Still the Right Buy
Two cases worth naming, because a rigid cost-per-sale rule will kill campaigns it should not.
The first is acquisition. A buy-back or equity campaign that sources trades is doing two jobs, and charging its full cost against retail units understates it. If 27 sales also brought in 19 acquired vehicles you would otherwise have bought at auction, the campaign has an inventory value that never appears in a cost-per-sale line.
The second is the slow month. The marginal value of a unit is not constant. In a month where the store is chasing a volume objective, or where fixed costs are already committed and the floor is quiet, a $541 sale that gets you over a threshold can be worth more than a $364 sale in a month you were going to hit anyway.
A Practical Checklist
- Write down every cost line, including creative, tracking and the BDC hours the response consumed.
- Count responses from tracked numbers and coded landing pages, not from memory.
- Hold out a control group from the same list. Without one, incrementality is guesswork.
- Calculate all three figures — reported, incremental, fully loaded — and label them.
- Compare to other channels on the same basis, or do not compare.
- Note any acquisition value separately rather than burying it.
Frequently Asked Questions
What is a good cost per sale for dealership direct mail?
It depends entirely on your gross, and anyone quoting a universal number is guessing. The useful test is the ratio: fully loaded cost per sale against front-end gross plus expected fixed-ops contribution. If the campaign costs a quarter of what the unit earns, it is working.
How big should the control group be?
Large enough that a couple of sales either way does not swing the result — typically a few thousand records on a campaign of this size. Below that the holdout produces noise you will then treat as signal.
Should the BDC cost really be charged to the campaign?
If you would not have incurred the hours without the campaign, yes. A phone room sitting idle is a fixed cost; a phone room working 300 extra responses is a campaign cost. Most stores are somewhere between, and apportioning the marginal hours is close enough.
What if we cannot separate campaign sales from walk-ins?
Then you are measuring correlation and should say so. Tracked numbers, unique offer codes and a match-back against the mailed list are the three mechanisms that turn a guess into a number. Attribution covers how they fit together.
Summary
Cost per sale is total campaign cost over units the campaign caused. Include the handling cost or the figure is fiction; hold out a control group or the units are inflated. Worked end to end, our 10,000-piece example produces $364 reported, $444 incremental and $541 fully loaded — all three defensible, and the difference between them is the difference between a vendor report and a number you can take to the owner. Compare it to gross, compare it to other channels on the same basis, and allow for the campaigns that are buying inventory as well as selling it.
