The short answer: the cheapest lead is often the most expensive car. Cost-per-lead tells you what you paid to make a phone ring; cost-per-sale tells you what you paid to put a deal in the finance office — and the two rankings are almost never the same. Once you divide spend by the units a channel actually sold (and look at the gross behind them), the "expensive" source that closes often beats the "cheap" one that doesn't. Most dealers can't run that number today, and this is why.
Cost per lead is a vanity metric
Every channel looks great on a cost-per-lead report. That's the whole problem — cost-per-lead rewards volume, and volume is the easiest thing in the world to buy. A source can flood you with cheap, tire-kicking, half-abandoned form fills and post a beautiful CPL while selling almost nothing.
The math that matters is one step further down:
If you budget off cost-per-lead, you will systematically defund your best channel and pour money into your worst one — and the report will thank you for it the whole way down.
The only formula that matters
Cost per sale is not complicated. It's:
Two words carry all the weight: by channel, and actually sold. Total marketing spend divided by total units is a fine board number, but it hides everything — it can't tell you where to move the next dollar. You need it split by source, and you need the denominator to be sold units that source can be honestly credited with, not leads, not "influenced," not clicks. Then you look at the gross behind those units, because cost per sale on its own is only half the picture.
What it actually costs to sell a car
Real, in-circulation benchmarks — useful as guardrails, not gospel. Your store's numbers depend on your market, your mix and your brand, so treat these as "is my number sane?" checks, not targets to hit.
| Benchmark | Rough figure | What it tells you |
|---|---|---|
| Common "healthy" target, total marketing per unit | $250–$350 / vehicle | The number a lot of dealers think they run at |
| Actual average ad spend per new vehicle sold | ~$739 / vehicle | The number a lot of dealers actually run at |
| Self-generated Google Search, cost per sale | ~$380–$765 / sale | Owned demand, on the cheaper end when it closes |
| Third-party marketplace leads, cost per sale | ~$8,000–$12,000 / sale | The eye-opener — and exactly why you measure per channel |
These figures are in-circulation industry benchmarks, not a single authoritative study — treat them as sanity checks against your own numbers, not targets.
That last row is not a typo, and it's not an argument to cancel the marketplaces. It's an argument to measure them — some stores' third-party spend earns its keep on units they'd never have seen, and some is lighting money on fire, and a cost-per-lead report can't tell you which store you are. If sorting that out is your actual question, there's a full guide on measuring third-party ROI — this piece is the same math pointed at every line, not just that one.
Cheap leads, expensive cars
Here's the trap laid out as a table, because it's the whole point of the guide:
| Channel A (cheap leads) | Channel B (pricey leads) | |
|---|---|---|
| Cost per lead | $50 | $100 |
| Close rate | 5% | 20% |
| Leads to make one sale | 20 | 5 |
| Cost per sale | $1,000 | $500 |
| Avg. gross per deal | $1,200 | $2,600 |
| Net contribution / deal | $200 | $2,100 |
Channel A wins the cost-per-lead beauty contest and loses on every metric that pays the bills. If your reporting stops at the first row, Channel A gets next quarter's budget. If it reaches the last row, you'd move every dollar you could into Channel B. Same spend, opposite decisions — and the difference is entirely in how far down the funnel you're willing to measure.
Why you can't compute this today
Most stores can't produce an honest cost-per-sale-by-channel number, and it's not a discipline problem. It's a plumbing problem. Three things have to sit in the same place, and in the average dealership they live in three systems that never talk:
- Spend lives in ad platforms and a stack of invoices. What you paid, by channel, by month.
- Source lives in the CRM — a single field, typed by a busy rep, wrong a large share of the time.
- Sold + gross lives in the DMS, in a completely separate world from either of the above.
To get cost per sale by channel, you have to join all three: this delivered unit → this lead → this source → this spend → this gross. Miss the join and you're back to dividing total spend by total units and calling it a strategy. This is exactly why the number is rare — and it's the same joined-up attribution the attribution guide is about. Cost per sale is what that attribution is for. And it inherits that join's weakest link: if the source on the lead is wrong in the first place, every cost-per-sale number below is a real spend figure divided by mis-assigned units.
Units lie; gross tells the truth
One more trap, because it's the one that survives even after you fix the plumbing. Two channels can post identical cost per sale — say $600 — and be nothing alike. One sells $2,800-gross deals; the other sells $900-gross deals with a chargeback waiting in three of them. Same cost per sale, wildly different businesses.
That's why the honest version of this metric always carries gross alongside it. Cost per sale tells you what the deal cost to acquire; gross tells you what the deal was worth. You need both in the same view, by channel, or you'll optimise toward cheap deals that don't make money — which is just the cost-per-lead mistake wearing a nicer suit.
The number moves when you fix the funnel, not just the invoice
Last point, and it's a hopeful one. Cost per sale has two levers, not one. You can cut spend — or you can raise the close rate on the spend you already have. Channel B in the table didn't win by being cheaper; it won by closing four times as often. A faster first response, a cleaner handoff, a lead that doesn't sit unanswered overnight — all of that lowers cost per sale without touching the ad budget, because the denominator (sold units) goes up while spend stays flat. The cheapest way to lower cost per sale is often to stop leaking the leads you already paid for.
What to do first
You don't need a new agency. In order:
- Stop budgeting off cost-per-lead. It's the number that flatters your worst channel. Demote it to a diagnostic and promote cost per sale.
- Join sold units back to lead source — even by hand. A monthly reconcile of every delivery to its lead record beats a year of CPL reports.
- Put gross next to cost per sale, by channel. The rankings will change. That change is the finding.
- Look at close rate before you cut spend. Half your "expensive" channels are only expensive because the funnel leaks after the lead lands.
- Then automate the join. Once spend, source and sold-gross live in one place, cost per sale stops being a quarterly spreadsheet exercise and becomes a number you can watch on the board every morning.
None of this is glamorous. But it's the difference between knowing what a car costs you to sell and guessing — and right now, for most stores, it's a guess dressed up as a report.
Questions dealers ask
- How do you calculate cost per sale at a dealership?
- Divide a channel's total ad spend by the number of vehicles that channel actually sold in the same period — not the leads it generated, the units it sold. So $10,000 spent on a source that produced 10 delivered deals is a $1,000 cost per sale. Do it per channel, not just store-wide, and look at the gross behind those deals at the same time, because two channels with the same cost per sale can make very different money. The hard part isn't the arithmetic — it's honestly crediting each sold unit to the source that earned it, which requires joining your ad spend, your CRM source field and your DMS sold-and-gross data together.
- Why is cost per lead a misleading metric for car dealers?
- Because it rewards volume, and cheap volume is the easiest thing to buy. A channel can post a great cost-per-lead by flooding you with low-intent form fills that almost never close. A $50 lead that closes at 5% costs $1,000 per sale; a $100 lead that closes at 20% costs $500 per sale — so the cheaper lead is twice as expensive per car. Budget off cost per lead and you'll defund your best-closing channel and overfund your worst. Cost per sale is the number that survives contact with the finance office.
- How much should a dealership spend to sell a car?
- Benchmarks in circulation put a healthy target around $250–$350 in total marketing per unit, while the actual average ad spend per new vehicle sold runs closer to ~$739 — so many stores spend roughly double what they think. But the useful number isn't the store-wide average; it's cost per sale by channel, which ranges from a few hundred dollars for self-generated search that closes to several thousand for third-party marketplace leads. Treat published figures as sanity checks, not targets — your real number depends on your market, mix and how well you close the leads you already pay for.
