Google Ads Performance Analyzer

Say this to activate the skill

---
name: google-ads-analyzer
description: Analyzes Google Ads exports (CSV/XLSX), builds pivot views, and delivers prioritized optimization recommendations. Activate for campaign analyses, monthly reports, or budget reviews.
---

This skill turns raw Google Ads data into a structured report with recommendations backed by the numbers.

## When this skill activates

- A Google Ads CSV or XLSX export is provided for analysis.
- A monthly or quarterly report for management or a client is requested.
- An optimization plan by campaign, ad group, or keyword is needed.

## Workflow

1. Inspect the input: identify columns (campaign, impressions, clicks, cost, conversions, conv. value), confirm date range and currency. Remove total rows such as "Total" before any aggregation.
2. Clean the values: parse thousands separators, currency symbols, and percent formats into raw numbers. Always compute CTR, CPC, CPA, conversion rate, and ROAS yourself from raw values, never reuse preformatted columns.
3. Build pivots per level: campaign, then ad group, then keyword or asset. Show cost share and conversion share per level and check which few campaigns absorb most of the budget.
4. Flag outliers: spend with zero conversions, CPA well above the account median, high CTR with weak conversion rate (landing page suspect), lost impression share (separate budget from rank).
5. If a search terms report is available, list irrelevant queries that carry cost as negative keyword candidates.
6. If a prior period exists, compute the delta per KPI; mark findings based on low conversion counts as uncertain.
7. Formulate recommendations: each action with a number from the data, the expected lever, and the concrete next step, sorted by budget impact.
8. Assemble the report; export as XLSX with pivot sheets on request.

## Output format

- Executive summary in 3 to 5 sentences, stating period and currency.
- KPI table per campaign: cost, conversions, CPA, ROAS, CTR, CPC.
- Top 3 and bottom 3 list with a short reason per entry.
- Prioritized action list: action, data evidence, lever, effort.

## Quality rules

- Every recommendation cites a concrete number from the data; generic tips without evidence are forbidden.
- Total and summary rows are removed before aggregation.
- Report ROAS only when conversion values exist; otherwise use CPA as the lead metric.
- Findings based on low conversion counts are labeled as uncertain, not overinterpreted.
- Missing columns or data gaps are named explicitly, never replaced by estimates.
- Period, currency, and data source appear in the report header.

Share this skill