E-commerce Intents

E-commerce intents are topics that the AI already learned to recognize and so you can use them to detect what the customer message is about.

These intents are already available in multiple languages and the AI is continuously learning from many of our e-commerce customers.

New intents are added every month and if you need a specific intent not covered here then you can either use the keywords actions or suggest to the DG product team as one of the new intents.

Category vs Intent

Intents represent specific e-commerce topics such as Missing Item(s), Promocode Not Working, Where is My Order, Return Status, etc. Many of these intents are related and therefore grouped into a single category. For example the following intents are part of the Order Status category: Missing Item(s), Where is My Order, Order Confirmation Not Received.

When the AI makes a prediction it does it in 2 steps:

1. Detect the Category
2. Detect a specific intent from one of the intents available in the detected category

As part of detecting a category or a specific intent the AI is looking at how confident it is about its prediction. If it is not confident enough it will avoid making a wrong prediction. In most cases the AI is either confident about both the category and specific intent or not confident about either, however sometimes the AI would be confident about the category but not confident enough about any of the specific intents. In that case the name of the predicted intent will match the name of the category rather than any specific intents.

Tags

Each of the intents is also associated with a tag. When the AI makes a prediction it will return the name of the category, the name of the intent, the associated tag and some other information such as confidence, message cleaning steps, etc. You can use the tag output to tag the tickets/case with those intents so you can then use it analytics inside your CRM.