AI Anomaly Detection for Google Ads —
Know Before It Costs You

What is anomaly detection — and why does it matter?

Anomaly detection is the automated process of finding data points that don't fit an account's normal pattern. Every Google Ads account has a rhythm: weekday traffic differs from weekends, CPCs drift with auction pressure, conversions ebb and flow with seasonality. An anomaly is a day that breaks that rhythm — spend that doubles with nothing to show for it, a conversion tag that silently stops firing, an impression share that falls off a cliff because a competitor raised bids overnight.

The reason it matters is simple: the cost of a problem in paid media compounds by the hour. A broken conversion action doesn't just lose tracking — it starves Smart Bidding of signals, and the algorithm starts optimizing toward the wrong outcome. A disapproved ad in your top campaign doesn't just pause one asset — it reroutes budget into weaker ads. The longer an issue runs undetected, the more expensive it becomes and the longer the recovery takes.

Most advertisers find out about these problems one of three ways:

• Manually — someone happens to look at the right report on the right day. This doesn't scale past a handful of accounts.
• Retroactively — the month-end report shows the damage after the budget is already spent.
• From the client — the worst way. "Why did our leads stop last Tuesday?" is a conversation no agency wants to have.

Statistical outliers, filtered by AI, grounded in your change log

Pure statistics catch too much. Pure human review catches too little, too late. AdCortex layers the two: statistical models do the tireless nightly scanning that no human can, and an AI judgment layer applies the context a statistician can't — so what reaches your inbox is only the identifiers that actually deserve attention.

Why human judgment still matters — and how we encode it

A z-score can tell you Tuesday's spend was unusual. It cannot tell you that Tuesday was Black Friday, that you launched a new campaign Monday night, or that the "conversion collapse" is actually a known tag migration. That contextual reasoning has always been the human's job — which is exactly why pure statistical alerting tools drown users in false alarms until they stop reading them.

AdCortex encodes that judgment into the pipeline. The AI layer sees the statistical evidence and the account's change log side by side, the same way you would, and asks: is this surprising given what we know changed? Anomalies explained by deliberate changes get suppressed or downgraded; anomalies with no explanation get escalated with the context attached. You stay in the loop for the decisions that matter — the system just makes sure the only things reaching you are worth your attention.

Account averages hide problems. Segments expose them.

The most dangerous anomalies are the ones invisible at the account level. If one campaign's conversion rate collapses while another surges, the account-wide average barely moves — the dashboard looks green while a real problem burns underneath. This is Simpson's paradox in action, and it's why detection that only watches account totals misses the failures that matter most.

AdCortex runs detection at every level of the account hierarchy, with baselines and thresholds tuned per level — campaign-level data is naturally noisier than account-level data, so campaign tests use wider bands to stay precise. When an alert fires, it names the exact identifier that triggered it, so you're never left diffing reports to figure out where the problem lives.

Every segment draws from the same nightly data pipeline that powers AdCortex reporting, so an alert is never based on stale or sampled data — it's the same finalized numbers you'd see in the report the next morning.

Quantifying outliers: z-scores, IQR, t-tests, and significance

"That looks high" is not a detection strategy. Every anomaly AdCortex surfaces is backed by explicit statistical evidence — here's what's actually running under the hood, and why each test earns its place.

Z-scores: distance from normal, in units of noise

The rolling z-score asks the most fundamental outlier question: how far is today's value from its recent average, measured in standard deviations?

A z-score of 1 is unremarkable — roughly a third of ordinary days land there. A z-score of 2 puts today in the most extreme ~5% of expected outcomes; a z-score of 3 puts it in the most extreme ~0.3%. AdCortex flags account-level metrics beyond |z| ≥ 2.0 and campaign-level metrics beyond |z| ≥ 3.0 — campaigns get the stricter bar because their daily data is inherently noisier, and the whole point is to only fire when the signal clearly exceeds the noise.

IQR fences: outliers without assuming a bell curve

Z-scores implicitly assume data is roughly normally distributed — but ad metrics often aren't. Spend can be skewed by a few heavy days; conversion counts on small budgets are lumpy. The interquartile range (IQR) test makes no distribution assumptions at all. It looks at the middle 50% of historical values and flags anything far outside that band:

AdCortex uses k = 1.5 at the account level (the classic Tukey fence) and k = 2.0 at the campaign level. Because the IQR is built from medians and quartiles, a single freak day in the history can't distort the baseline the way it distorts a mean and standard deviation — making this test robust exactly where z-scores are fragile.

T-tests: is the difference bigger than chance?

Single-day tests answer "was today weird?" — but many real problems show up as a sustained shift: conversion rate down 15% for five straight days, none of them individually extreme. The two-sample t-test family answers the deeper question: is the difference between a recent window and its baseline window larger than natural day-to-day variance would produce by chance?

The t-statistic scales the observed difference by the pooled uncertainty of both windows — accounting for how noisy each period was and how many days of evidence you have. A large t-value (and correspondingly small p-value) means the shift is unlikely to be random fluctuation. This is the same machinery behind A/B test significance, applied to time: your last week versus your normal week.

Week-over-week and trend breaks: respecting seasonality

Comparing Monday to Sunday tells you about weekends, not anomalies. Week-over-week tests compare each day to the same weekday in prior weeks, which removes the strongest seasonal pattern in ads data before asking whether the change (40%+ at account level, 60%+ at campaign level) is abnormal. Trend-break analysis complements it by fitting the recent trajectory of a metric and flagging when today's value departs sharply from where the trend said it should be — catching gradual-then-sudden failures that point-in-time tests miss.

Statistical significance ≠ practical significance

Here's the part most tools get wrong: a statistically significant change is not automatically an important one. A campaign spending $3/day can triple its CPC with total damage of a few dollars — statistically extreme, practically irrelevant. That's why detection in AdCortex requires convergence:

• Multiple independent tests must agree. A z-score spike alone is a coincidence candidate; a z-score spike that also breaches the IQR fence and the week-over-week threshold is a pattern.
• Magnitude gates filter trivia. Peer-outlier tests only evaluate when there's meaningful spend at stake, so micro-budget noise never pages you.
• The AI layer applies the practical test. Even a statistically airtight anomaly gets weighed for business impact and checked against the change log before it earns a place in your digest.