GA4 Custom Insights

GA4’s Custom Insights takes an interesting and useful feature in GA4, Analytics Insights, and makes it even more useful and actionable. We use it to monitor our clients’ GA4 metrics for anomalies and issues. This capability has quickly become a standout feature, enabling us to proactively notify clients about significant changes.

In this article, learn what Custom Insights are, how to set them up, and see some examples of great Custom Insights that will keep you on top of website traffic changes.

What Are GA4 Custom Insights?

GA4 custom insights allow you to set email alerts for specific metric changes, acting as an early warning system for potential issues with your website, tracking, or marketing campaigns. For instance, you can receive an email if your conversions drop by more than 25%.

Setting Up a GA4 Custom Insight: A Step-by-Step Guide

Here’s how to quickly set up an alert for a significant drop in page views:

1. Navigate to ‘Home’ in your GA4 property.
2. Scroll to the bottom and click ‘View all insights’, or search for ‘insights’.
3. Click ‘Create’ in the upper-right corner, then ‘Create new’ under ‘Start from scratch’.
4. Configure your alert:
   1. Evaluation frequency: Daily (default)
   2. Segment: All Users (default)
   3. Metric: Views
   4. Condition: % decrease more than 25%
   5. Comparison period: Same day last week
   6. Choose insight name: this can be whatever you want, but if you are creating multiple insights, it’s helpful to include the property name, metric and condition, since this will be the subject line that appears in your inbox.
5. By default, alerts are sent to your logged-in email. Update this if necessary.

That’s it – your first alert!

Tips for Effective Alerts

• Adjust Frequency: Consider weekly alerts if daily checks result in too many false positives.
• Be judicious and don’t create too many or overly-sensitive alerts. Alerts are only useful if you pay attention to them. If you are getting daily notifications and most of them are false alarms, you’ll start ignoring them.
• One of the options for an insight condition is ‘Has anomaly’. Selecting this option will show spikes as well as drops (and benefits from some fancy math to compensate for noise in the data), but it can take a while to collect enough data to start working. Google says daily alerts take 90 days to gather sufficient training data, and weekly alerts take 32 weeks! I’ve found it usually takes less than that, but I usually set up %-decrease alerts in tandem with anomaly alerts so I’m covered while the anomaly alerts are learning.

Example Custom Insights

You will find yourself frustrated by some of the things alerts can’t do, for example you can’t monitor specific key events (conversions), but there are a lot of things you can do. Below are some ideas to get you thinking.

Monitoring specific traffic sources

You can use the Segment option to monitor specific sources, mediums and campaigns. You can’t monitor Sessions, but you can monitor Active users, which is a pretty good proxy. For example, you can monitor Google Ads by adding a Dimension for First user source = google and First user medium = cpc , or monitor organic traffic with First user medium = organic.

Monitoring for bots

In my experience, bot traffic tends to be entirely composed of First visits, so you could just monitor for anomalies in this metric, or set a %-increase alert. But you will get false positives if there are legitimate surges in traffic to your property. I find that a lot of bots come from outside the target countries of a business or organization. So, for example, if my client is US-based, I’ll set up an alert for First visits and segment where Country does not exactly match the United States.

Monitoring for ecommerce tracking failure

Hopefully this never happens to you, but it’s happened to me and it sucks. Most of the time, you will be monitoring for % drops or anomalies, but you can also monitor for zero values. This is especially useful if you work on a property that gets a relatively small number of transactions or key events (conversions) per day. To set this up, monitor the metric Transactions or Gross purchase revenue, and set the condition to ‘Is less than or equal to’ zero.

Those are just a few ideas. If you get creative with segments, there’s a lot you can do!

Implementing Alert-Driven Workflows

We’ve set up Custom Insights for all clients using our Managed Analytics service, so that we are aware of concerning changes and can investigate and take appropriate action. Email alerts initiate a workflow that streamlines decision-making and task delegation. This workflow helps us enhance both the speed and accuracy of our responses, saving costs and improving clients’ digital marketing performance.