MEASUREMENT / CARBIDE INSIGHTS

The dealership marketing KPIs

SHORT ANSWER

Dealership Marketing KPIs

Dealership marketing totals never reconcile because every vendor counts a lead in its own system and several of them count the same person. The way out is to separate activity metrics, which describe effort, from leading indicators, which move before sales do. Impressions, sessions, posts and even total leads are activity. Non-branded organic sessions, service appointments booked online, phone calls over thirty seconds and unique connected shoppers are leading indicators. Splitting branded from non-branded is the single most useful change most stores can make to a report, because branded traffic flatters every channel that touches it. The number that matters most is the one your general manager would act on, and a usable monthly report is short enough to be read and ends with what to do next instead of with a list of things that happened.

Updated September 1, 2026

#Why the totals never reconcile

The same shopper visits from a marketplace, clicks a paid ad, searches your name, and submits a form. Four systems each record a touch and each claims the customer, because each is measuring its own contribution in isolation.

Sum the vendor reports and you get more leads than humans. This is not usually dishonesty: it is that last-click attribution in four separate tools produces four separate last clicks.

The practical consequence: you cannot evaluate channels by comparing vendor reports to each other. They are not measuring the same thing, and none of them can see the others.

#Activity metrics versus leading indicators

Impressions, sessions, follower counts and rankings describe activity. They can all rise while the store sells nothing more, which is what makes them comfortable to report.

Leading indicators are narrower and less flattering: non-brand organic sessions to high-intent pages, branded search volume, form completions from pages that convert, service appointment bookings, and cost per acquired customer by channel.

The distinguishing question: if this number doubled, would the store sell more? For impressions the plain answer is often no. For non-brand inquiries on model and service pages it is usually yes.

#Splitting branded from non-branded

Branded search reflects demand created elsewhere: by the manufacturer, by your reputation, by tier-one advertising. Non-branded reflects demand you are newly capturing.

Reported together, a flat program with rising brand awareness looks like marketing success. Reported separately, you can see which lever moved.

This applies to both organic and paid. A paid account whose growth is entirely branded is largely buying traffic that would have arrived anyway, and that is worth knowing before renewing the budget.

#The number that matters most

Cost per acquired customer, by channel, over a window long enough to include the actual sales cycle. Not cost per lead, which rewards channels producing large volumes of cheap, unqualified inquiries.

This requires connecting marketing spend to the CRM, which is where most stores stop because the data is messy. It is worth pushing through: the ranking of your channels by cost per lead and by cost per acquired customer is frequently different, and sometimes reversed.

Be realistic about the window. Vehicle purchase cycles are long. Judging a channel on a 30-day window measures how quickly it produces inquiries rather than how well it produces customers.

#What a usable monthly report contains

One page. Non-brand organic sessions and inquiries, branded search volume, paid spend and non-brand paid conversions, service bookings, total acquired customers, and cost per acquired customer by channel. Ask directly.

Plus a change log: what was released, changed or broken this month, by whom. Without it you cannot separate market movement from things you did, and a year later nobody remembers why April dipped.

What it should not contain: rankings for hand-picked terms, impressions without context, and any metric where the vendor being evaluated is also the sole source of the number.

#Running the monthly review so it changes something

One hour, one page, same agenda. Whoever owns marketing presents the numbers; vendors do not present their own performance in that meeting. A vendor summarizing its own results is not reporting, it is advocating, and the distinction is worth protecting.

Start with the change log instead of the results. What did we do, release or break last month? Reading results without that context produces confident explanations for random variation, which is how programs end up chasing noise.

End with a decision. Every monthly review should produce at least one thing to stop, start or change, with a named owner. A review that concludes everything is fine and nothing changes is a status update, and status updates do not need an hour.

TABLE

Activity, leading indicator, or outcome

Vendors report from the left column because it always goes up. Hold them to the middle one.

Metric	Class	What it tells you	Safe to judge a vendor on?
Impressions and reach	Activity	That something ran	No
Sessions, total	Activity	Mostly branded and returning visitors	No
Total leads	Activity	How many forms exist, plus duplicates	No
Non-branded organic sessions	Leading	Whether new people are finding you	Yes
Calls over 30 seconds	Leading	Real conversations, not misdials	Yes
Service appointments booked online	Leading	Fixed ops demand you captured	Yes
Sold units and RO count	Outcome	The result, lagging and multi-causal	As context, not as attribution

Metrics only settle arguments when everyone agrees what is being counted, and that argument runs through most of the dealership marketing insights library.

Questions.

Which single metric should we watch?

Cost per acquired customer by channel, if you can get it. It resists the distortions that affect cost per lead and it is the number that maps to the decision you are making. Where the next dollar should go.

How do we handle double counting between vendors?

Accept that vendor reports cannot be summed, and reconcile against a single source that sees the actual outcome, usually the CRM. The goal is not perfect attribution, it is a consistent basis for comparing channels.

Are rankings worth reporting at all?

As a diagnostic, occasionally. As a headline metric, no: the term list is chosen by whoever makes the report, which makes it easy to flatter. Search Console clicks segmented by intent are harder to game and more useful.

What is the most misleading metric dealerships get reported?

Click-through rate on branded search, presented as if it measured marketing performance. It mostly measures how well known the store already is, which advertising did not create and cannot easily move. It answers a different question than the one being asked.

How often should marketing reporting happen?

Monthly for the numbers that move slowly (SEO, content, AI-search visibility), weekly for anything spending real budget in real time (paid search, paid social). Reporting more often than a metric can meaningfully change just adds noise.

Should service department marketing be measured the same way as vehicle sales marketing?

No. Service demand is steadier and less price-sensitive, so a service campaign's success looks like booked appointments and retained customers over time, not the immediate cost-per-lead comparison that makes sense for a vehicle sale.

What is the difference between an activity metric and a leading indicator?

Activity metrics describe effort. They can rise while the store sells nothing more, such as impressions, sessions, follower counts and rankings. A leading indicator moves before sales do, such as non-branded organic sessions to high-intent pages, calls over thirty seconds and service appointments booked online. The test is whether the store would sell more if the number doubled.

Why should branded and non-branded search be reported separately?

Branded search is borrowed demand. It reflects demand created elsewhere, by the manufacturer, your reputation or tier-one advertising. Non-branded reflects demand you are newly capturing. Reported together, a flat program with rising brand awareness looks like marketing success, and reported separately you can see which lever moved.

What belongs on a one-page monthly marketing report?

Non-brand organic sessions and inquiries, branded search volume, paid spend and non-brand paid conversions, service bookings, total acquired customers, and cost per acquired customer by channel. Add a change log of what was released, changed or broken that month, and leave out hand-picked rankings and any number where the vendor being judged is the only source.

How long a window should we use before judging a channel's cost per acquired customer?

A long one. Use a window long enough to include the real sales cycle, because vehicle purchase cycles are long. A 30-day window measures how quickly a channel produces inquiries, not how well it produces customers, and it can reverse the ranking you would get from a longer view.

Should vendors present their own results in the monthly marketing review?

No. Whoever owns marketing at the store should present the numbers. A vendor summarizing its own results is advocating, not reporting, and the review should start with the change log and end with at least one thing to stop, start or change, with a named owner.

Which metrics are safe to judge a marketing vendor on?

Non-branded organic sessions, calls over thirty seconds and service appointments booked online, because they lead sales and are hard to inflate. Impressions, total sessions and total leads are not safe, since they describe activity or include duplicates, and sold units are context instead of attribution because they lag and have many causes.

PRIMARY SOURCES