Meta Ads for a local restaurant: the KAIVOR method

Why Meta Ads still works for restaurants in 2026

When Meta Ads “stops working”, it’s usually a targeting or a creative problem. Done right, it’s still the best ROAS for local acquisition.

The account structure that scales

For a local restaurant, keep the structure simple: over-segmentation kills the algorithm’s learning.

Campaign 1 — TRAFFIC / SALES (budget at campaign level, Advantage+ or CBO type)
├── Broad geo-targeted audience (5 km) — 1 ad set
│   ├── Creative A: hero dish
│   ├── Creative B: kitchen POV
│   └── Creative C: tested offer (lunch menu)
└── Lookalike 1% of loyal customers (if base > 1000 contacts)
    └── same 3 creatives

Campaign 2 — RETARGETING (separate, smaller budget)
└── Site visitors 30d + page engagement 30d
    └── Creative D: strong offer + urgency ("this weekend only")

The rules that matter more than the structure:

• 3 creatives minimum per ad set. The algorithm needs material to optimise.
• Don’t touch anything during the first 7 days (learning phase).
• Don’t raise the budget by more than +20%/day: beyond that, learning resets to zero.
• Broad audience > hyper-targeting. Locally, the geographic radius already does 90% of the targeting.

The optimal geo-targeting

• 5 km radius around the restaurant (10 km in rural areas)
• Exclusion: other towns
• Broad audience (no hyper-targeting)
• Lookalike of loyal customers, 1%

The 3 creative angles that convert

1. The hero dish: 1 signature dish, close-up, ASMR sound
2. The chef’s story: 30 sec storytelling with the chef’s face + kitchen
3. The tested offer: “€12 lunch menu” with an immediate call-to-action

A realistic budget

€300/month minimum. Below that, it’s impossible to test properly (1 creative = €50/day minimum).

Target ROAS for restaurants

5x is realistic for a fast-food / pizzeria with a €15-25 ticket. 8x+ for a premium restaurant with a €40+ ticket.

Free ads audit → 15-min diagnostic.