Q: I keep hearing Meta is automating everything. Is that actually true, and should I be worried?
Mostly true, and the worry is valid if you go in blind.
Meta has been rolling out what they’re calling “full funnel automation” and it’s accelerating fast. By late 2026, the default experience for most advertisers will be Advantage+ campaigns where Meta controls your creative combinations, your audience, your placements, and your bidding strategy. You’ll set a budget and a goal, and Meta’s AI will handle the rest.
For large e-commerce brands with broad audiences and lots of purchase data to feed the algorithm, this works well. For a local solicitor, a Donegal skincare clinic, or a Dublin-based financial advisor trying to reach people in a 5km radius? It’s a different conversation.
Q: What’s the problem with AI targeting for hyperlocal businesses?
Meta’s AI is optimised for scale. It wants to find you conversions across the widest possible audience because that’s where it has the most data to work with. When you tell it “only Dublin city, only people aged 35 to 55 who own a home,” you’re boxing it into a pool so small it gets uncomfortable.
The algorithm doesn’t panic, it just starts bending your rules. It shows your ad to people just outside Dublin. It pushes spend toward placements and demographics you didn’t intend. It interprets “reach more people likely to convert” and decides Cork is close enough.
For a national brand, that’s a rounding error. For a business where every client needs to physically walk through a door in Rathmines, that’s wasted budget.
Q: Can I still override the AI targeting?
Yes, but your options are narrowing and you need to know where to push back.
The main levers you still control are:
Location targeting with “people who live in this location.” This sounds obvious but it matters. There’s a difference between “people in or recently in this location” (which picks up tourists and commuters) and “people who live here.” For local service businesses, always use “lives in.” Advantage+ campaigns default to broad. You have to go into manual sales campaigns to lock this down properly.
Audience controls in Advantage+ Shopping. Meta has introduced “audience controls” within Advantage+ campaigns. You can set a minimum age, exclude certain locations, and lock in a specific country. You can’t do granular interest targeting but you can at least stop the AI from serving your ads to 18-year-olds in Galway.
Manual campaigns with CBO. If the AI is genuinely causing problems, you can still run manual campaigns. You lose some of the AI-assisted delivery benefits, but you regain control over who sees your ads. For very small geographic targets, this is often the better call in 2026, particularly for lead gen rather than e-commerce.
Q: What about AI-generated creative? Can Meta now just make my ads for me?
It can, and it will try to if you let it.
Meta’s Advantage+ creative features can now generate image variations, rewrite your copy, translate your ads, and mix and match your assets into combinations you never approved. Some of this is useful. A lot of it is not.
The issue for local businesses is brand voice and local relevance. An AI-generated version of your ad might be technically fine but feel generic. It won’t say “serving Donegal families for 15 years” unless you put that in the copy to begin with. It won’t know that your clients respond to specific local landmarks or cultural references.
The fix is to be deliberate about your creative inputs. If you’re feeding Meta three images and two copy variations, the AI works with what you give it. Give it better inputs and you get better outputs. If you want full control, turn off Advantage+ creative enhancements in the ad setup. They’re opt-out, not opt-in, so you have to actively switch them off.
Q: What about AI-managed bidding? Should I let Meta control that too?
For most local businesses, yes, with caveats.
Manual bidding (target cost, bid cap) used to be the go-to for keeping costs predictable. The problem now is that Meta’s AI-managed bidding has genuinely gotten better, and fighting it manually often leads to under-delivery, especially in small audiences.
The smarter approach is to use Highest Volume bidding (Meta’s default) but set a clear campaign budget and monitor your cost per result daily in the early days of a campaign. If cost per lead is running higher than your acceptable threshold, you can layer in a cost per result goal rather than a hard bid cap. This nudges the algorithm rather than caging it.
The one time I’d use a bid cap for a local business is if you’re in a very niche vertical with a high-value lead, like financial services or legal, and you can’t afford runaway costs. In that case, a bid cap protects you even if it limits reach.
Q: So what’s the actual strategy for local businesses in 2026?
Work with the AI where it helps you, override it where it doesn’t.
Specifically:
Use Advantage+ Shopping campaigns if you’re selling products online and you have purchase data. Let the AI do its thing there.
Use manual campaigns with locked location targeting for service businesses that need local precision. Dublin 2 to Dublin 8 only? Do that manually.
Feed the AI excellent creative inputs rather than fighting its creative tools. The AI is as good as what you put in.
Turn off the Advantage+ creative enhancements you don’t want. They’re on by default. Check every campaign.
Use conversion events higher up the funnel if you don’t have enough purchase or lead data to train the algorithm properly. “Add to cart” or “lead form open” gives the AI more signal than waiting for full conversions in a small audience.
And most importantly, don’t let Meta’s push toward automation convince you that campaign structure, creative quality, and offer clarity no longer matter. They matter more now, because the AI amplifies whatever you put in front of it. Put in something generic and it will scale generic. Put in something that genuinely speaks to your local audience and it will find more people who respond to that.
The businesses that get the best results from Meta’s AI in 2026 are the ones who treat the AI as a distribution engine and take full responsibility for the quality of what goes into it.
