<aside> ✨

Explore is the core of the WatchTower platform, enabling you to label and understand your data.

</aside>

What Are Tags?

https://www.loom.com/share/b2e2949a8c154b5c90cfa4f0ea75c8e4

Within EdgeTier, “tags” are used to identify language of interest within any interaction that happens between customers and agents, bots, or surveys. Tags can be applied to any text: agent speech, customer chat messages, survey responses, bot messages.

Tags are applied on a sentence-by-sentence level to interactions at near real time speed within EdgeTier, and can be used, for example, to:

• Reduce reliance on manual agent labelling for interactions, decreasing handling time and improving reporting accuracy
• Identifying troubling agent behaviour through agent language use across 100% of interactions
• Tracking specific topics of interest over time after an event, for instance, identifying all customers referring to a product feature after release.
• Being alerted, in real-time, to particularly time-sensitive customer interactions, such as abusive customers, regulatory issues, repetitional risk, or vulnerable customers.

There are three types of tags within the EdgeTier platform:

• Phrase Tags: These are tags you create yourself within the system. You name the tag, and create triggers to help the system learn when to apply tags to interactions. Phrase tags allow you to label what matters to you and be alerted to particular issues. More below!
• AI Tags: An LLM approach to automatically applying labels to your interactions. AI Tags fall within the contact reason category (for now)! AI Tags allow you to visualise and report on what and why your customers contact you the most. More here.
• External Tags: Tags we ingest from your CRM or similar platform.

How Are Phrase Tags Applied?

EdgeTier uses an intelligent language-matching system to create tags based on trigger sentences, or simply “triggers”. Triggers are example phrases that indicate when tags should be applied to interactions in EdgeTier.

Triggers are matched to sentences within each interaction, and can be matched based on the meaning of each sentence. Triggers are language agnostic and very quick to create.

For example, the trigger phrase “I would like to cancel my account” will be a good trigger to match sentences like:

• I want to cancel my account