Draft next month's campaigns from evidence

The problem this solves

Campaign planning meetings run on impressions of the data rather than the data. Last month's performance lives across two ad accounts; competitor movement lives in ad libraries nobody checked; and the plan gets sketched from whatever the loudest memory in the room insists happened. Then someone spends days translating the sketch into actual campaign structures, and by launch, the month it was based on is ancient.

The evidence for a genuinely better plan exists the whole time. Your own accounts say which audiences, angles, and placements earned their budget and which burned it. The ad libraries say where competitors are concentrating and what they abandoned. Reading both, drawing the conclusions, and building the structures is simply more hours than any one planner has between the end of one month and the start of the next - so evidence loses to instinct on a schedule.

How the mission runs

1. Read last month's evidence. The mission pulls performance from Google Ads and Meta Ads: which campaigns, audiences, and creatives earned their spend, which decayed, and where the marginal budget clearly worked hardest. Reads only - the accounts are untouched.
2. Read the competitive board. Ad Libraries captures what your competitors are running right now: the angles they are scaling, the offers they are pushing, and what quietly disappeared since last month. Their spend patterns become directional evidence for your plan.
3. Draft the campaign structure. From both evidence streams the agent drafts next month: campaign architecture, audience definitions, angle per audience, and budget split - with each choice annotated by the evidence behind it, so the plan reads as an argument rather than a decree.
4. Push everything paused. The draft campaigns are created in your accounts in a paused state - fully built, spending nothing. This is the platform's standing rule for ad writes: campaigns land paused or in approval queues, and money moves only after a human turns things on.
5. Hand over the approval brief. You get a summary of what was built and why: each campaign, its evidence, its budget, and what to check before activating. Approve the lot, activate selectively, or edit structures in the platform first - launch is entirely your call.

The prompt

This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.

What comes back

Next month's campaigns fully built in Google Ads and Meta Ads and sitting paused: structures, audiences, angles, and budgets all in place, each annotated with the performance and competitive evidence that shaped it, plus an approval brief that walks you through the reasoning. The planning cycle compresses from weeks of meetings to a review session, and nothing spends until you activate it.

Make it yours

• Constrain the budget: give the mission a hard monthly cap and let the draft argue the allocation across platforms within it.
• Plan a single platform: draft only the Meta month in full depth when that is where the interesting decisions live.
• Add a challenger slate: alongside the evidence-led plan, draft one deliberately contrarian campaign to keep the account learning.

Frequently asked questions

Go deeper

• Playbook: the paused-draft campaign workflow
• Budget allocation by evidence: splits that trace to data
• Approval queues that keep autonomy fast
• AI advertising operations: the complete guide