Build a Meta Ads Agent That Manages $100K+/Month

Most people skip this and wonder why their agent underperforms. The quality of your knowledge base is the single biggest factor in how well the agent operates.

• Business Profile — Positioning, target demographics, pricing tiers, seasonal trends, LTV by segment
• Product Catalog — SKU-level inventory, best sellers, margin leaders, new launches, bundle opportunities
• Brand Guidelines — Voice, visual rules, messaging frameworks, what NOT to say
• Historical Performance — Last 90–180 days of campaign data, creative benchmarks, CPA by channel

You can't optimize what you don't understand. Before the agent touches anything, it audits everything. This is where hidden money is found.

• Campaign Structure — Logical organization, redundancy, naming conventions
• Creative Performance — Best formats, fatigue cycles, hooks that drive clicks
• Audience Segmentation — Overlap issues, exclusion opportunities, targeting gaps
• Spend Efficiency — Where budget is going, what's getting results, what's wasted
• Creative Fatigue Signals — Which ads are dying and when CTR/CVR drop

The audit tells you what's broken. The rules tell your agent how to fix it automatically. Think of these as your standing orders.

• ROAS > target + 20% → increase budget 15%/day (max 50%/week)
• ROAS < target − 30% for 3 days → reduce budget 25%
• 60% budget to proven winners / 30% stable / 10% tests
• Rotate creatives every 4–7 days based on fatigue signals
• Kill creatives with <1% CTR after 1,000 impressions
• Exclude purchasers from prospecting campaigns
• Pause ad sets with frequency >3.5 in 7 days

Data access is the engine. Start read-only — earn write access once you trust what you've built.

Your agent doesn't just optimize — it tells you what's happening and why. No more logging into dashboards.

• Daily — Spend, revenue, ROAS vs. targets, top/bottom performers, pacing alerts
• Weekly — Structure health, creative trends, audience analysis, scaling opportunities
• Monthly — Month-over-month comparison, compound gains, quarterly planning recommendations

Every agent starts here. It analyzes, recommends — you execute. This phase builds trust and catches mistakes before they cost money.

Once proven, grant write access. But the safety rules below are non-negotiable — don't skip them.

• Hard spending cap — Set in Meta Ads Manager. Agent can never exceed it. Ever.
• Budget change limits — Max 20% increase per campaign per day. Max 50% per week.
• Approval gates — New campaigns, budget increases over $500/day, core audience changes = human sign-off required
• Kill switch — Auto-pause everything if ROAS drops 40%+ in a single day

This is the biggest competitive advantage. Your agent tests creatives faster than any human team — finding winners before your competition even knows they exist.

AI agents are powerful, not magic. The compounding effect is what separates them from everything else.

• Days 1–30: Learning phase — conservative recommendations, 5–15% efficiency gains
• Days 30–90: Compound intelligence — rules refined, creative cadence accelerates, 15–30% gains
• Days 90+: Full autonomy — 90% of decisions autonomous, 25–50% efficiency gains

Work through these in order. Don't skip steps. The sequence matters.

• Build the knowledge base — Brand docs, product catalog, historical data
• Run the full account audit — Campaign structure, creative performance, spend efficiency
• Define optimization rules — Budget, creative rotation, targeting, bid strategies
• Set up API access (read-only first) — Meta, Shopify, creative library
• Configure reporting — Daily, weekly, monthly cadence
• Run read-only for 2–4 weeks — Refine rules, build trust
• Set hard safety rules — Spending caps, budget limits, approval gates, kill switch
• Grant write access — Start small, monitor closely
• Enable creative automation — Batch uploads, A/B testing pipeline
• Monitor and refine — Daily first week, weekly after that