#The three models, honestly described
In-house means an employee — usually a marketing coordinator or manager — who sits in the building. They know the inventory, they know the GM, they can walk to the service drive and ask a question. They are also usually the only marketing person in the building, which means they have no one to check their thinking against.
Agency means an outside firm on retainer. They have seen more stores than you have, they have specialists, and they can execute at a volume one person cannot. They also have other clients, their own incentives, and a strong structural preference for work that is easy to keep billing for.
The third model has a few names — fractional, virtual, embedded. A senior marketing person working with you part-time, usually a day or two a week. They do not replace execution. They decide what gets executed and hold the vendors to it.
#What each one is actually good at
In-house is best when the work is continuous, local and relationship-heavy. Store events, community sponsorships, coordinating with sales managers, getting photos taken properly. Things that require being there.
Agencies are best at specialised, high-volume execution. Paid media management, creative production, technical SEO work that needs tooling and repetition. Work with a clear scope and a measurable output.
The fractional model is best at the thing neither of the others does well: deciding. Which channels get money, which vendor is underperforming, whether the website platform is the constraint, what to stop doing. That is a judgement job, and judgement scales badly by the hour.
#How each one fails
In-house fails through isolation. A capable coordinator with no senior marketing counterpart drifts toward whatever the most confident vendor recommends, because they have no basis to push back. Over a couple of years the marketing programme becomes a collection of things salespeople sold to the store.
Agencies fail through scope. The retainer covers what the retainer covers. When the real constraint is outside it — the website platform, the CRM, how leads are handled after they arrive — the agency will keep optimising inside its box and reporting improvements while the actual number does not move.
The fractional model fails when it is hired as cheap execution. If you bring in a senior person and then ask them to build campaigns and post to social, you are paying director rates for coordinator work and getting neither.
#The question that actually decides it
Not 'what can we afford' — the models overlap more on cost than people expect once you count an employee properly. The question is: who currently makes the call when two vendors disagree?
If the honest answer is the GM, and the GM has no marketing background, you have a decision problem and hiring more execution will not touch it. If the answer is a capable internal person who is already making good calls, you need execution capacity, not more strategy.
Most dealerships are in the first situation and buy for the second. That is the single most common structural mistake we see.
#What a hybrid actually looks like
The arrangement that tends to work: one internal person who owns the day-to-day and knows the store, specialist agencies for the work that genuinely needs specialists, and one senior person accountable for whether the whole thing is working.
The important part is that the senior person is not also a vendor. If the same firm decides the strategy and gets paid to execute it, you have removed the only independent check in the system. That is not an accusation of bad faith — it is just that nobody reliably recommends less of their own service.
Write down who owns which decision before you hire anyone. Most of the dysfunction we are asked to fix is not a skill problem. It is that nobody agreed who decides.
#What the first ninety days should look like
Whichever model you choose, the first ninety days decide whether it works. Week one is not a campaign — it is an inventory of what already exists: every vendor, every contract, every login, every recurring charge. Most stores cannot produce that list on request, which is itself the finding.
Then a scope conversation with each vendor. Not a performance review — a clarification of what each one believes they are responsible for. The gaps and overlaps that surface here are usually where the wasted spend lives, and they are invisible from any single vendor's report.
Only after that does anything change. The instinct to demonstrate value quickly by launching something is the most common way a good hire starts badly: new activity layered onto an unexamined system makes the system harder to read, not better.