Mazda builds applied AI once, and runs every campaign on it with Recommend
Most commercial work is disposable. A campaign is planned, built, run and archived; the next one starts from nothing. The organisation gets the result, but keeps almost none of the machinery that produced it.
Mazda took a different route. The infrastructure built for the 6e launch was designed to be kept — and it has since carried a CX-60 offer and a CX-30 offer. Three campaigns, two genuinely different kinds of brief, one system the client owns.
Build once, decide many times
A launch and a cash back offer are not the same problem. One introduces a car nobody knows. The other has to move a car people already recognise, on a deadline, without competing on price alone.
Normally each gets its own build. Here both ran on the same foundation: applied AI holding the models and their features against what buyers are actually thinking about, and the pages and content produced from that. The first brief pays for the infrastructure. Every one after it starts from something that already exists.
A feature only matters once you can picture using it
Automotive marketing defaults to specification: range, torque, charging time, screen size. All true, and none of it tells a buyer what the car would be like in their life.
The system holds each model's features against the situations people are actually in — how they use a car, and when. The output is a benefit someone can picture themselves using rather than a number they have to interpret.
In a category built on spec sheets, that is the substantive editorial decision, and applied AI is what makes it possible to make it at scale.
The client keeps the machinery
The infrastructure is Mazda's. It does not get rebuilt for the next brief and it does not leave when a campaign ends.
That is the difference between buying work and gaining a capability. The organisation is not commissioning the same foundation again every time the objective changes — it is standing new work up on something it already owns.
Every activation informs the next
Three campaigns through one system means three sets of evidence about which situations resonate, which framings carry, and which parts of the infrastructure earn their place.
The fourth brief does not start from a blank page. It starts from what the first three taught, which is why each one has been quicker to stand up than the one before.
What's next
The infrastructure keeps running, and the next brief starts from where the last one finished.
That is what applied AI looks like inside a commercial team — not a tool bought, but a capability gained: it reads what buyers are thinking about, decides how a model belongs in that, produces the work, and leaves something behind that the next campaign builds on.