Your cost per lead looks fine. Your cost per sale does not. And right now, somewhere in a Monday meeting, that gap is about to get blamed on the media — and a campaign that was doing its job is about to get cancelled.
Before you cut that budget, I want to understand one thing: what happened to the lead after it arrived? Because cost per lead and cost per sale measure two different machines. One measures your media. The other measures your media plus everything your store did — or didn’t do — after the shopper raised a hand. When the first number is healthy and the second one isn’t, the media is usually the wrong suspect.
This post is for dealer principals, general managers, and marketing directors at franchised new-car stores who are staring at that exact gap. Let’s walk it from first principles.
The lead arrived. Then what?
Here’s the funnel, stated plainly, with no vendor gloss on it:
- Lead arrives. A form fill, a call, a chat, a click-to-text off a Vehicle Detail Page (VDP).
- Lead gets logged in the Customer Relationship Management (CRM) system. Or it doesn’t.
- Lead gets worked. Someone — a salesperson or the Business Development Center (BDC) — owns it.
- Lead gets a response. A real one, from a human, in a timeframe the shopper still cares about.
- Appointment set.
- Show. They walk in the door.
- Sold VIN. Deal caps, gross gets booked, the Dealer Management System (DMS) records it.
Your ad budget bought step one. Every step after that belongs to the store. And the leak almost never sits at step one — it sits at steps two through four, in the unglamorous middle where a lead exists but nobody has answered it yet. That’s the part nobody puts on the marketing dashboard, which is exactly why the marketing dashboard takes the blame.
Cost per lead evaluates the media. Cost per sale evaluates the whole chain. Treating a bad cost per sale as proof of bad media is like blaming the water company for a cracked pipe inside your own wall.
What does the leakage actually look like in the data?
One vendor has published numbers on this, and they’re worth naming precisely. The Foureyes 2026 Automotive Dealer Benchmarks Report — drawn from 1.4 billion visits across 22,900+ dealer sites — reports the following across its own install base:
- 42.7% of qualified leads mishandled.
- 15.2% never logged in the CRM.
- 62.8% getting no salesperson response within 24 hours of returning to the site.
- 11.7% of sales leads buying.
Two honest caveats. First, this is one vendor’s benchmark across the dealers running its software, not an industry census — your stores may run better or worse. Second, the numbers still describe a shape, and the shape is the point: a large share of the leads the money bought never got the handling a bought lead deserves.
Read those four figures as a sequence. Roughly one lead in every six or seven never enters the system of record at all. Nearly two-thirds of shoppers who come back to the site — the warmest possible signal short of walking in — hear nothing from a salesperson within a full day. And well under half of qualified leads are handled the way a qualified lead should be. Against that backdrop, the fact that 11.7% of sales leads buy is less a media statistic and more a survival statistic.
Why does an unlogged lead make the media look worse than it is?
Because your cost per sale is a fraction, and an unlogged lead corrupts the fraction.
Walk the arithmetic. If 15.2% of leads never reach the CRM, those leads still cost money — the spend is fully in the numerator. But any deal those shoppers eventually produce can’t be matched back to the campaign, because the lead was never in the system to match. Some of them buy anyway. They come in on a Saturday, get skated to a fresh up, and the deal writes as a walk-in. The sale is real. The gross is real. The attribution is gone.
So the campaign’s reported cost per sale is computed on sold VINs it can prove, divided into spend it fully absorbed — a denominator missing part of its numerator’s output. The math doesn’t say “your store dropped these leads.” The math says “this campaign is expensive.” And that is the exact arithmetic that gets a working campaign cancelled: the money did its job, the handling didn’t, and the report card only had the media’s name on it.
If your GA4 numbers and your CRM numbers also disagree with each other — a separate problem, and a common one — that’s covered in why GA4, your CRM, and your vendor reports never agree. This post isn’t about which report to trust. It’s about what the store did with the lead regardless of which report you’re holding.
How fast is fast — and is 24 hours actually the bar?
No. Twenty-four hours is a floor for measuring failure, not a target for hitting success.
The Foureyes benchmark uses 24 hours because it’s a clean, defensible line for counting a miss: 62.8% of leads getting no salesperson response within 24 hours of returning to the site is a failure count, full stop. But don’t let the measurement window become the operating standard. A shopper who submitted a lead on a specific vehicle is comparison-shopping in real time. They likely submitted the same form at two other stores in your market. The store that answers while the shopper is still sitting there — still on the VDP, still deciding — is the store that sets the appointment. The store that answers tomorrow is answering a customer who mentally bought elsewhere last night.
In my view, the right internal standard is response within the same shift the lead arrived. Minutes, not hours. If a lead comes in at 11 a.m. and gets its first human touch at 4 p.m., you didn’t respond — you archived. The 24-hour figure should show up on your reporting exactly once: as the line below which a lead is declared mishandled. Everything above that line is where the actual competition happens.
And a note on where this is heading: as AI-assisted search starts shaping how shoppers find inventory, the leads that do arrive will be fewer and better-qualified. That’s an emerging shift, not today’s crisis — but it makes every mishandled lead more expensive over time, not less.
What should I measure instead of cost per lead?
Cost per lead isn’t wrong — it’s just insufficient. It grades the media and stops. To separate a media failure from a handling failure, you need one number per stage of the middle, and you need to know which system owns each number so nobody can argue about whose report is right.
- Logged rate. Of leads that arrived (from every source — forms, calls, chats), what percentage exists in the CRM? Owner: the CRM, audited against raw source counts — phone system logs, form submissions, chat transcripts. This is the first number to pull, because if it’s low, every number downstream is fiction.
- First-response time. Median minutes from lead arrival to first human response. Owner: the CRM. Not the auto-responder. A template email that fires in four seconds is not a response; it’s a receipt.
- Response-within-shift rate. Percentage of leads that got a human response before the shift that received them ended. Owner: the CRM, but the BDC or sales desk answers for it. This is your operating standard; 24 hours is merely your failure line.
- Appointment set rate. Of responded leads, how many became a scheduled appointment? Owner: the CRM.
- Show rate. Of set appointments, how many walked in? Owner: the CRM, reconciled against the showroom log.
- Sold. Of shows, how many became a sold VIN? Owner: the DMS. This is the only stage where the DMS gets a vote — and it should be reconciled back to the CRM monthly, or your funnel and your financial statement will quietly drift apart.
Six numbers. Even rough versions beat none, because they end the arguing. Once you can say “logged rate is fine, first-response time is the problem,” half the internal noise about the ad budget stops on its own. Chaos persists in these conversations because nobody can say which stage is broken, so everyone debates tactics — more spend here, cut spend there — when the answer is a stopwatch.
One boundary worth stating: none of this is an attribution-modeling exercise. Which channel deserves credit for a sold VIN is a real question with its own trade-offs, and it lives in the attribution pillar. The six numbers above don’t assign credit. They locate the leak.
So which is it — a media problem or a handling problem?
Run the one diagnostic that actually separates them: hold media constant and fix handling first.
Don’t touch the campaign. Same spend, same channels, same targeting, for a full measurement window — long enough to cover your typical lead-to-sale cycle. During that window, fix exactly one thing: the handling. Audit the logged rate and close the gaps where leads enter but never reach the CRM. Set the response-within-shift standard and staff the BDC or the floor to hit it. Track the six numbers weekly.
Then read the sold number.
If sold VINs move while media spend didn’t, you have your answer: it was never the media. The campaign was buying real shoppers the whole time, and the store was spending them. The cost per sale that looked ugly wasn’t measuring your advertising — it was measuring your advertising divided by your follow-up.
If sold VINs don’t move — logged rate is high, first-response is fast within the shift, appointments set and show at healthy rates, and the sold column still won’t budge — now you’ve earned the right to question the media. Now a cut is a decision, not a guess. Cut the campaign at that point and you’re cutting on evidence — and if the next question is where that budget should go instead, the budget pillar walks through building the number from your own unit goals rather than a rule of thumb.
The order matters for a reason beyond logic: fixing handling is inexpensive. It’s process, accountability, and a stopwatch. Cutting a working campaign is expensive twice — once when the leads stop, and again when you rebuild the pipeline six months later after the floor goes quiet and everyone suddenly remembers the ads were doing something. Fix the low-cost thing first. Then judge the expensive thing fairly.
Where this leaves the operator
For what it’s worth on background: I spent four years inside an 11-location RV dealership, La Mesa RV, and 25 years in marketing across industries — and the pattern this post describes is not automotive-specific. It’s industry agnostic. Money buys attention; process converts it; and when conversion fails, the money takes the blame because the money is the line item everyone can see. Franchised new-car stores just run this pattern at higher stakes, because every lead maps to a specific VIN and a specific gross.
The dealer groups I want to work with aren’t the ones with the biggest budgets. They’re the ones willing to look at the whole funnel holistically before making a cut — to take decision ownership over the six numbers instead of outsourcing judgment to whichever vendor report looks worst this month. If your cost per lead is fine and your cost per sale isn’t, you don’t have a verdict yet. You have a question. And the answer is sitting in your CRM, your phone logs, and your showroom traffic sheet, waiting for someone to compose those numbers into one honest picture before the wrong budget gets cut. That audit — first-principles, alongside the business, before the cancellation email goes out — is exactly the work I do.