Sonar detects anomalies by learning what 'normal' looks like across scoped sets of interactions. For example, if there is more contacts than usual about a specific booking code.

Until now, you could filter anomalies by model on the Anomalies screen , but you couldn't see or control the settings driving them.

A new Models screen under Sonar > Models changes that. It shows every anomaly model configured for your account, including the tags it monitors, its sensitivity level, alert thresholds, and recent activity metrics , so you can see what's being watched and how each model is set up. From the same screen, you can also create new models and edit existing ones, all within the EdgeTier platform.

What's new

• A new Models section has been added to Sonar. From there, you can view all the anomaly models currently running in your account; including the tags they monitor, their sensitivity level, their alert thresholds, and some metrics on their recent activity.
• A new Create model button launches a guided, three-step setup flow: Basic details (name, interaction types, languages, and tags to monitor), Sensitivity (pick a level that sets the alert threshold and minimum interaction count, alongside a 14-day interaction volume chart that flags a warning if daily volume looks too low for reliable detection), and Notifications (choose who gets alerted)
• Existing models can be updated the same way, via a new Edit option in each model's menu, pre-filled with its current settings.

Where to find it

1. Open Sonar in the left-hand navigation.
2. Click Models, a new tab that appears alongside Anomalies.
3. The Models page lists all anomaly models currently running in your account.
4. Click Create model to set up a new one, or use the ⋯ menu on an existing model and select Edit.

How to create an anomaly model

You will need a role with the Manage Anomaly Models permission to create or edit models.

1. Go to Sonar > Models and click New model.

1. Basic details: Give the model a name, then choose the interaction type, languages, and tags you want it to monitor. Sonar then monitors relevant interactions for any ‘out of the ordinary’ content, such as spikes in contacts about a topic like cancellations.

   As you set the filters for the models, Sonar shows the average daily interaction volume for that combination. If it's too low for reliable detection, a warning banner appears (hover the info icon next to it to see the threshold it's checking against). Click Next.