skill-118-click-fraud-invalid-traffic-monitoring

Skill 118: Click Fraud and Invalid Traffic Monitoring

Purpose

Monitor for suspicious click patterns indicating click fraud or invalid traffic. Early detection prevents budget waste and protects Smart Bidding from learning on fraudulent data.

Data Requirements

Data Source: Custom GAQL Required

GAQL Query - Click Analysis:

SELECT
  campaign.name,
  segments.date,
  metrics.clicks,
  metrics.impressions,
  metrics.cost_micros,
  metrics.conversions,
  metrics.invalid_clicks,
  metrics.invalid_click_rate
FROM campaign
WHERE segments.date DURING LAST_14_DAYS
  AND campaign.status = 'ENABLED'

GAQL Query - Geographic Analysis:

SELECT
  campaign.name,
  geographic_view.country_criterion_id,
  metrics.clicks,
  metrics.impressions,
  metrics.cost_micros,
  metrics.conversions
FROM geographic_view
WHERE segments.date DURING LAST_30_DAYS

Run via /google-ads:get-custom with query name fraud_detection.

Analysis Steps

1. Check Invalid Click Metrics: Review Google's invalid click data and refunds
2. Identify CTR Anomalies: Flag campaigns/periods with CTR >5-10%
3. Analyze Click Patterns: Look for click bursts, unusual geographic concentration
4. Cross-Reference Conversions: High clicks with zero conversions = suspicious

Thresholds

Condition	Severity
Invalid click rate >5%	Critical
CTR >10% on Search, >2% on Display	Warning
Click burst (>2 std dev above average)	Warning
High clicks from non-target geographies	Warning

Output

Short (default):

## Click Fraud Analysis
**Account:** [name] | **Period:** Last 14 days | **Invalid Click Rate:** [%]

### Summary
- **Total Clicks:** [count]
- **Invalid Clicks:** [count] ([%])
- **Google Refunds:** $[amount]

### Suspicious Activity
- **[campaign]**: CTR [%] - Abnormally high
- **[date] [hour]**: [count] clicks in 1 hour - Possible click burst
- **Clicks from [location]**: [count] - Not in target geography

### Recommendations
1. Request manual review from Google for [campaign]
2. Add geographic exclusions for [locations]
3. Consider third-party click fraud protection