Google Ads anomaly detection involves comparing current metric values against a rolling historical baseline (typically 7-14 days) and flagging deviations that exceed predefined thresholds. A CPC increase above +25% vs. the 7-day average is a yellow alert; above +40% is a red alert requiring immediate investigation. Conversion rate drops are the most urgent anomaly type: a decline of more than 20% warrants same-day review of landing pages and conversion tracking. AI-powered monitoring can automate this detection across all metrics simultaneously.
Not every metric fluctuation in Google Ads is an anomaly. This cheat sheet defines normal variation ranges, yellow alert thresholds, and red alert thresholds for 8 key metrics — so you can distinguish statistical noise from real problems that require action.
✅ Key Takeaways
- Define thresholds before anomalies happen — reactive monitoring is always late
- Percentage deviations should be measured against a rolling 7-day or 14-day baseline
- CPC can naturally vary ±15% daily — only investigate spikes above ±25%
- Conversion rate anomalies are the most urgent: even a 20% drop warrants immediate review
- Always cross-reference multiple metrics before declaring a true anomaly
How to Interpret These Thresholds
Yellow alert thresholds indicate the metric has moved outside normal variation and should be monitored. Red alert thresholds indicate an anomaly requiring immediate investigation and likely corrective action. All percentages are measured as deviation from the 7-day rolling average baseline.
📊 Reference Table
| Metric | Normal Range | 🟡 Yellow Alert | 🔴 Red Alert | First Action |
|---|---|---|---|---|
| CPC | ±15% vs 7d avg | >+25% | >+40% or >−30% | Check auction insights, bid changes |
| CTR | ±10% vs 7d avg | <−20% | <−35% | Review ad relevance, QS changes |
| Conversion Rate | ±15% vs 7d avg | <−20% | <−35% | Check landing page, tracking |
| ROAS | ±20% vs 7d avg | <−25% | <−40% | Check CVR + CPC combination |
| Impression Share | ±10% absolute | <−15% abs | <−25% abs | Check budget, bid strategy |
| Cost/Conversion | ±20% vs 7d avg | >+30% | >+50% | Check CVR + CPC together |
| Search Imp. Share Lost (Budget) | >10% | >20% | >35% | Review budget allocation |
| Clicks | ±20% vs 7d avg | <−30% | <−50% | Check status, Quality Score, bids |
Frequently Asked Questions
What counts as a Google Ads anomaly?
An anomaly is a statistically significant deviation in a key metric from its historical baseline. In practice: any metric moving more than 2 standard deviations from its 7-14 day rolling average is worth investigating.
How should I set up anomaly alerts in Google Ads?
Use Google Ads custom rules (under Tools → Automated Rules) to trigger email alerts when metrics exceed your defined thresholds. Set rules at both campaign and account level, with daily or 3-day lookback windows.
Why did my CPC spike suddenly?
Common causes: increased competition in the auction, Quality Score drop (raising minimum CPC), broad match expansion into more competitive queries, or budget increase causing Google to enter higher-cost auctions.
When should I pause a campaign due to anomalies?
Only pause if you have clear evidence of a structural problem (tracking broken, landing page down) or if spend is accelerating with zero conversions over 3x your normal CPA. Pausing due to normal variance destroys learning.
Can AI detect Google Ads anomalies automatically?
Yes. AI-powered tools like AdPredictor analyze your account daily, compare each metric against its rolling baseline, and flag anomalies with context — distinguishing noise from real problems that require action.