meta-ads-analyzer

Analyze Meta Ads performance at the campaign level for CBO and ad set level otherwise, then evaluate trends over time rather than single snapshots. Always resolve exact metric names with the meta marketing tool and consider marginal cost, learning phase status, and relevance diagnostics before drawing conclusions.

The breakdown effect is the misreading of Meta's budget shifts as funding underperforming segments. The system optimizes for marginal efficiency, so a segment with higher average CPA may still protect overall campaign efficiency; never judge decisions by average CPA alone.

Meta's legal and policy requirements mandate the exact phrase "Accounts Center accounts" when describing reach or audiences. Terms like "people", "person", or "unique users" are forbidden, and Reach must be used as a standalone metric name without modifiers.

Display "Data not available" or "N/A" when the API returns null, and never fabricate or estimate values. When aggregating across campaigns with mixed objectives, show "N/A" for Cost per result and Results instead of computing them.

No. For policy questions it calls the meta marketing policy tool and returns only the official Meta Advertising Standards URLs with a brief redirect message. It never interprets, summarizes, or paraphrases policy content.

Instability usually comes from the learning phase, which lasts until roughly 50 optimization events occur within 7 days, or from significant edits that reset learning. Other causes include audience saturation, auction overlap, creative fatigue, and seasonality.