Analytics Market Chart

Google Analytics Intelligence Reports

Analytics Intelligence for Google Analytics represents a step away from passive reports toward meaningful analysis.

Spotting undercurrents and long-tail trends has never been very easy with Google Analytics. You usually had to know what you were looking for. And if a negative trend was developing, it could sometimes take days or weeks before you spotted it.

Google recently announced a new feature that goes a long way toward solving this and should change the way you use Google Analytics. Analytics Intelligence automatically scans your reports and alerts you to any unusual trends in your traffic patterns.

How Analytics Intelligence Works

Google uses proprietary algorithms to make predictions about what your traffic should look like. Every night it scans 11 dimensions and 18 metrics for unexpected activity. If any of these dimensions or metrics fall outside of what Analytics Intelligence expected, it triggers an alert.

It looks for unexpected trends in daily, weekly and monthly traffic. In other words, if the traffic for a day was unusual, it triggers an alert. If traffic for an entire week was unusual, it triggers a different alert.

Analytics Intelligence also shows you just how unexpected an event was. If you have more visits than expected, it will tell you how many visits it expected and how different the reality was from the expectation.

Reading Analytics Intelligence Reports

Analytics Intelligence allows you to focus on just the most unusual events or to dig into even the most minute aberrations by allowing you to set an alert sensitivity. When the sensitivity is set high, it will show every alert, including those that may not be significant. The lower the sensitivity, the more unexpected an event needs to be to get displayed.

Click on a date in the bar graph to see the alerts for that day, week or month. Each alert is triggered independently, and no correlation between alerts is implied. Although, it's often obvious if there is some correlation between two or more alerts.

The alert shows the expected value and how different the actual value was as a percentage. If it was a positive change, the percentage is green. If the change was a bad one, the percentage is displayed in red. It also displays the dimension affected and what percentage of overall traffic this aberration represents. At the far right, is a bar graph that represents how relevant the data is. The more filled in it is, the greater deviation from the norm it represents.

Analytics Intelligence Best Practices

Once you understand how to read the reports, it's time to put them into effect. I have long advocated knowing what questions you want answered before beginning an analysis. Analytics Intelligence, however, is another great starting point, because it asks questions for you. It invites you to figure out why your visits from organic searches were so much higher that day, or why your goal conversions dipped on another day.

Start reading the reports with the report sensitivity set as low as possible. In other words, start with the most dramatic changes and move to the less significant alerts. Rather than immediately pursuing an answer for each alert, make a list of questions to answer. Some of the alerts are probably related.

There is a link for most alerts to create an advanced segment out of that alert. This is usually the next step. Create an advanced segment to single out the unusual traffic for any related alerts, and then start to dig into the issue.

Custom Alerts

You can also create custom alerts. You have the option of having them emailed to you. Custom alerts will show up as blue bars on the Analytics Intelligence graph. This is best used for keeping an eye on site-specific issues or things that you expect will go wrong. Just like standard alerts, these will only process once a day. They cannot be used to spot trends immediately.