The short answer: you can't judge a lead source by the number of leads it sends or what it charges per lead. You judge it by tracing its leads all the way to delivered gross, then dividing that by what you paid. Every listing site can be worth it or worthless for a given store — the only way to know yours is to measure it on your own numbers. This guide is deliberately neutral: no provider is the villain, and the good ones welcome being measured.
Why lead counts and CPL mislead
Cost-per-lead is the metric everyone quotes because it's the one everyone can see. It also hides the three things that decide profit:
- Close rate varies wildly by source. One site's shoppers convert at double another's. Two sources with the same CPL can have completely different cost-per-sale.
- Gross varies by source. Some sources bring price-shoppers, some bring buyers who value the right car. Volume at thin gross can lose to fewer, healthier deals.
- Duplicates inflate everyone. The same shopper often submits on the website and a listing site. If both get "credit," your best sources look worse and your overlap looks like performance.
The metric that actually matters
Stop comparing cost-per-lead. Compare cost-per-sale and return on gross per source:
| Metric | How to read it |
|---|---|
| Leads | Volume only — the start of the question, never the answer |
| Cost per lead (CPL) | What the vendor charges — useful only next to close rate |
| Close rate by source | What share of that source's leads become delivered deals |
| Cost per sale | Total spend ÷ delivered deals from that source — the real cost |
| Return on gross | Total delivered gross from that source ÷ total spend — the real verdict |
A worked example
The abstraction is what makes this easy to nod at and never do, so here it is with numbers. Two sources, each costing the same $3,000 a month:
| Source A | Source B | |
|---|---|---|
| Monthly spend | $3,000 | $3,000 |
| Leads | 100 | 40 |
| Cost per lead | $30 | $75 |
| Close rate | 4% | 15% |
| Delivered units | 4 | 6 |
| Cost per sale | $750 | $500 |
| Return on gross (at $3,000 avg) | 4.0× | 6.0× |
On cost per lead, Source A looks twice as good and is the one that survives the budget meeting. On cost per sale and return on gross — the two numbers that actually describe what you got for the money — Source B wins decisively, on 60% fewer leads. That gap is the entire argument for measuring past CPL, and it's the same trap covered in cost per sale by channel: the cheapest lead is regularly the most expensive car.
How to actually trace it
The reason most stores never do this is that the answer lives in two systems that don't talk: the lead is in your CRM, the sale is in your DMS, and nothing automatically connects them. The work is joining them:
- Stamp every lead with its true source the moment it arrives — not the form provider, the actual origin (CarGurus, AutoTrader, website, Facebook).
- Collapse duplicates so one shopper is one record, and the sale is credited to the source that truly created them.
- Match each delivered deal back to its lead by matching the buyer across CRM and DMS — so every sale carries a source.
- Divide real gross by real spend per source, over a window long enough to be fair to slower-closing sources.
Do that and the ranking often surprises people: the source with the loudest dashboard isn't always the one paying the bills, and a quiet source can turn out to be the most profitable in the store.
Comparing the two, honestly
"CarGurus vs AutoTrader" is the search everyone runs, and the honest answer is that there isn't a general one. Which marketplace performs better depends on your market, your franchise or independent mix, your price positioning and the inventory you actually stock. Two rooftops in the same group, forty minutes apart, routinely rank them in opposite orders. Any article that declares a winner is describing someone else's store.
North of the border the two names change — for most Canadian dealers the comparison is AutoTrader.ca versus Kijiji Autos, with CarGurus Canada in the mix — but the method below is identical. That Canadian marketplace landscape is its own guide: dealership software built for Canadian dealers.
What travels is the method, not the verdict. To make a comparison mean anything, hold five things constant across both:
- The same window — a full quarter, not a month, so a slow-closing source isn't punished for its cycle length.
- The same definition of a close — delivered units, consistently, not booked on one side and delivered on the other.
- The same gross basis — front plus back, or front only, but the same one for both.
- Duplicates collapsed across both — a shopper who inquired on both platforms in the same week is one customer, and crediting the sale twice flatters whichever one you happen to check first.
- Response time held equal — if one source's leads get answered in four minutes and the other's sit until morning, you're measuring your own floor process, not the platform. Marketplace leads are the shortest clock in the store because they're shared with competing dealers the moment they land.
Questions to ask before you cut or renew
- What is this source's cost per sale, not per lead?
- What gross do its deals actually deliver, on average?
- How many of its "leads" were duplicates of shoppers I already had?
- Is it slow-closing? Am I measuring it over a long enough window?
- If I cut it, do those buyers show up from another source — or disappear?
Questions dealers ask
- How do you calculate ROI on CarGurus or AutoTrader leads?
- Two divisions, both per source and both over the same window. Cost per sale is total spend with that source divided by the delivered units it actually produced - not leads, units. Return on gross is the total delivered gross from those units divided by the same spend. A worked example: two sources each costing 3,000 dollars a month. The first delivers 100 leads at a 30 dollar cost per lead and closes 4 percent, so 4 units at 750 dollars per sale. The second delivers 40 leads at 75 dollars each - two and a half times the cost per lead - but closes 15 percent, so 6 units at 500 dollars per sale. On cost per lead the first looks twice as good; on cost per sale and return on gross the second wins decisively. Use a window long enough to be fair to sources that close slowly, and count delivered units rather than booked ones.
- Is CarGurus better than AutoTrader for dealers?
- There is no universal answer, and any published comparison that gives you one is describing someone else's store. Which marketplace performs better depends on your market, your franchise or independent mix, your price positioning and your inventory - the same two platforms routinely swap places between two rooftops in the same group. The only comparison that means anything is the one you run yourself: same measurement window, the same definition of a close, the same gross basis, duplicates collapsed across both so a shopper who inquired on both isn't counted twice, and cost per sale plus return on gross computed for each. Run that for a full quarter before you cut or renew either one.
- Is CarGurus or AutoTrader worth it for a dealership?
- It depends entirely on your store, your inventory and your follow-up — which is why you have to measure it yourself rather than trust anyone's dashboard. The only honest answer comes from tracing each source's leads to actual delivered deals and gross, then comparing that against what you pay. A site that looks expensive on cost-per-lead can be your best source on cost-per-sale, and vice versa.
- Why can't I just trust the provider's own lead dashboard?
- Provider dashboards count the leads and activity they generate, which is real but incomplete. They can't see which of those leads bought, at what gross, or which ones were duplicates you were already working from another source. ROI lives in your DMS and CRM, not in the vendor's reporting.
- How do I attribute a sale back to the right lead source?
- Match the buyer of each delivered deal back to the lead that started them, by matching on the customer across your CRM and DMS. Once every sale carries the source that created it, you can divide real gross by real spend per source and rank them honestly. The hard part is the matching, which is why it usually needs to be automated.
