For years, franchise marketing conversations were dominated by targeting: audience segments, lookalikes, interest stacks, geo-radius tuning, the endless tinkering that used to separate a sharp media buyer from an average one. That conversation has largely moved on. Meta’s Advantage+ and Google’s Performance Max now handle most of those decisions automatically, processing signal combinations no human campaign builder could track manually. Ask anyone running franchise accounts on these platforms in 2026 how much manual audience control they still touch day to day, and the honest answer is: not much, and that’s a good thing.
Don’t get us wrong, we use AI throughout our own process. It’s a great tool to generate and test creative variants faster than any manual workflow could, to catch fatigue patterns across hundreds of location-level accounts before they show up as a revenue problem, and to process the same signal-matching that used to eat a media buyer’s whole week. The platforms’ automated targeting is part of that same shift, and we lean into it rather than fight it. The tools are genuinely good at what they do.
What we don’t do is let automation run unsupervised. AI is very good at optimizing toward whatever it’s told to optimize toward. It’s not good at deciding what that should be, or at knowing your category well enough to catch it when the numbers look fine but the leads don’t. That’s still a human job, and it’s a bigger one than it used to be because the tools have taken over so much of what used to fill the day.
Where the human still has to be behind the wheel
- Someone has to decide what’s worth testing. AI can generate ad copy variants on demand, at a volume no team could produce manually. It can’t tell you which variant is actually worth putting in front of an audience for your category, or that a location three miles from an established competitor needs a different angle than one with no competition nearby. That judgment call happens before the algorithm ever sees the creative.
- The algorithm can only optimize among what it’s given. If creative variety is thin, automated targeting doesn’t have room to find a real winner. It just optimizes among mediocre options and performance quietly plateaus. Reading that correctly (a creative gap, not a targeting ceiling) takes a person who knows to look there.
- Automated systems chase signals, not intent. Advantage+ and Performance Max optimize toward whatever they can measure — clicks, form starts, add-to-carts — because that’s the data they have. Weak creative can generate plenty of that kind of signal without producing a single qualified lead. The platform isn’t wrong; it’s doing exactly what it’s told. Someone still has to check that what it’s been told to chase actually matches what the funnel needs.
- “We use AI” isn’t the same as “we have a strategy.” Plenty of franchise systems have handed targeting to automation and quietly counted that as their AI strategy, full stop. The tools are doing their part. The gap shows up in creative testing discipline and account oversight — the parts that still need a person paying attention, not just a platform running.
- Testing has to be read in context, by someone who knows the category. Structured tests, real sample sizes, results pushed back down across the whole system — that’s a discipline, not a setting you turn on. AI can execute the test. It can’t tell you what the result means for a franchise system with forty locations at forty different stages of maturity.
Where this leaves you
The brands pulling ahead right now aren’t the ones avoiding AI, and they’re not the ones handing it the whole account and walking away either. They’re the ones who know exactly which decisions to give to automation — targeting, execution, scale — and which ones still need a person who understands the category watching closely: what to test, how to read it, and when the numbers are lying to you.
If you want a second opinion on where that line sits in your own account, we do free strategy calls for exactly this. No pitch deck, just a look at what’s actually happening — what the automation is handling well, and where it needs a human decision it isn’t getting.
