Anomaly Detection is an in-house tool by Tatvic that identifies certain user behavior or actions or a set of actions by users which do not conform to an expected pattern(s) in a dataset. Expected patterns can be generated from historical data sets or idealistic data sets that you can configure as well – we’re big on customizations at Tatvic!
This blog is the second blog on this tool where I am going to focus on various use cases and how Anomaly Detection Tool can help achieve these goals. We will see how Google Analytics Data and Anomaly Detection collaborate beautifully for metrics with:
- Only data
- Dimension filters
- Dimension values
- Dimension values with dimension filter
To get a detailed sense of Anomaly Detection as a tool, I invite you to read the first blog on the tool where we have elaborated on what is Anomaly Detection, the importance of Anomaly Detection, methods and Case Study.
Anomaly Detection Use Cases:
For every use case, Anomaly Detection Tool extracts real time data from Google Analytics and performs anomaly check depending upon reporting option(Intraday / Daily). All these use cases can be used with or without segments. Let’s discuss each of these use case in detail and see how they can fit into your business.
Metric only data
Every business has some of the key performance metrics which impacts business significantly. For instance Revenue, Average order value, E-commerce conversion Rate are key indicators for an e-commerce store whereas Pageviews, Users, Sessions can be more important to Lead Gen site. Based on your business KPIs, keep your eyes open and be vigilant about how these metrics are doing on a regular basis.
Metric with dimension filters
Sometimes in business, we also want to know how a metric is performing with regards to specific dimension values. For instance, I want to be notified if my Revenue from City A and/or Browser B has increased or decreased significantly. Anomaly Detection digs deeper into your dimensions and gives you an idea of how a set of dimensions affect your critical business metrics.
Metric with dimension values
With this tool, you can set a filter for some of the values of dimensions but say you want to set anomaly on top “N” values of dimensions. For instance, you know that you have 10 various Source / Medium from which your website received healthy traffic. With this tool, you can track the spike or drop from any of these Source / Medium real time.
Metric with dimension values with dimension filter
This is when you need Metric with dimension values and filters combined. Let’s extend the examples from our previous uses cases and let’s say I want to set an anomaly on top 10 Source / Medium from City A and/or Browser B. Guess what? You can do this with the help of Anomaly Detection Tool.
About Anomaly Alerts
Anomaly alerts can be reported either on intraday basis or daily basis. In Intraday reporting anomaly detection is performed several times a day while on daily basis anomaly detection is performed on previous day’s data and reported to value found out to be Anomalous. Anomaly Detection alerts also provide the time since the anomalous trend in data is observed.
Recently one of our clients received an alert for a drop in traffic from a specific browser. Anomaly Detection Tool reported the alert to the client within 3 hours of the drop. The client did an inspection on it and turned out that there was some javascript error on the page which led to the drop in traffic. Client team fixed the issue and traffic from that specific browser became normal on the same day.
Does the Anomaly Detection Tool sound like something that you would like to know more about? Are there any anomaly issues that you face recurrently? Tell us how you track your spikes and drops in traffic in the comment section below.
To request a demo of the tool, I invite you to check out the tool and leave a demo inquiry with our team.
