ads-auditlisted

# Reviewing campaigns ## First: is there enough data? Check before any assessment. From `ads/references/benchmarks.md`: | Metric | Minimum | |---|---| | CTR | 1,000 impressions | | CPC | 100 clicks | | Conversion rate | 100 clicks and 10 conversions | | CPA | 30 conversions | | Trend | 14 days | | Comparison of two groups | 100 clicks each | If fewer than four of seven are supported, **output no overall figure**. Say this instead: ``` Overall assessment: NOT POSSIBLE (2/7 metrics supported) Supported: impressions (3,400), clicks (47) Not supported: CTR, CPC, CVR, CPA, ROAS Reason: 47 clicks is below the threshold of 100. Recommendation: keep running for another 10 days. ``` This is the most important part of this skill. A number built on thin data looks like poor performance when data is merely missing — and that is how working campaigns get switched off. ## Fetching data ``` get_insights(scope="account", time_granularity="daily", since=..., until=...) get_insights(scope="campaign", entity_id=..., time_granularity="daily") get_insights(scope="ad_group", entity_id=..., time_granularity="none") ``` For breakdowns: `segment="device"` or `segment="country"`. Exactly one — the API allows no more, and none in combination with `hourly`. Available metrics: `impressions`, `clicks`, `spend`, `ctr`, `cpc`, `cpm`, `conversions`, `cpa`, `post_click_cvr`, `order_created_roas`. For ROAS use **only** `order_created_roas` — `roas` and `attributed_sales_*` are deprecated.