Analytics Guide: Stop Losing Traffic and Not Knowing Why!

Contributor, June 12, 2018

How to Quickly Detect Sudden Traffic Drops on Your Website in Google Analytics

One of the most significant issues facing many online businesses is detecting an unexpected drop or a spike in their online business activity.

There are many ways to categorize anomalies.  The following are typical examples of data anomalies:

  1. Credit card fraud where average transactions exceed the expected spending habits of an individual.
  2. Spam from referral traffic that can cause a significant spike in traffic stats.
  3. Unusual activity relating to excessive login attempts into a host server.

The biggest issue in anomaly detection is to identify the real anomalous deviations.  Anomaly detection is a machine learning technology that tries to predict anomalous incidents. To evaluate the accuracy of any anomaly tool, one needs to conduct many measurements and assess how they stack against the real results.

What is Anomaly Detection?

Analytics Intelligence Anomaly Detection is a statistical technique to identify “outliers” in time-series data for a given dimension value or metric.

  • Intelligence selects a period of historical data to train its forecasting model. For detection of daily anomalies, the training period is 90 days. For detection of weekly anomalies, the training period is 32 weeks.
  • Intelligence applies a Bayesian state space-time series model to the historical data to forecast the value of the most recent observed data point in the time series.
  • Intelligence flags the data point as an anomaly using a statistical significance test with p-value thresholds based on the amount of data in the reporting view.”

Google is looking at historical data to create upper and lower value to compare for anomalies.  The Bayesian state space-time series model is a specific model that is being used by Google for detecting data anomalies, there are many other models out there.  Testing with p-value thresholds is to ensure that the detections are statistically valid.

An Example of Anomaly Detection in Google Analytics

Google provides these alerts automatically, so one needs to keep an eye for these alerts.


Being able to detect these anomalies could help an online business to spot irregular activities right away and fix it fast before it creates a significant issue for the business.

If you could benefit from professional buy-side or sell-side consulting or assistance with your Google Analytics account, we can help you! Feel free to contact us here.