To get the most insight out of your conversation data, building a customised AI model from your historical data means we can start tracking the really granular issues specific to you.
This does take some set up, and depending on the complexity of your data and quantity of tags required, it can take from 2-4 weeks. Once we view your data, we will let you know a more concrete timeline.
Your model gets built in stages:
1-2 weeks: Rough + ready AI tags. We like to get your model up and running asap. We always aim to provide you with a model at this time and it will include the AI tags that cover a large percentage of your data, and the accuracy will still have room for improvement.
2-3 weeks: Granular, accurate AI tags. We continue to improve your model by adding more data; this is where your model gets more and more refined, picking up on a variety of different ways of saying the same thing. AI tags get more specific and more accurate. Most customers are ready to use the AI tags within this time period.
3-4 weeks: Tricky data? Your model may need more time to learn and recognise patterns. This is usually the case for voice data or complaints data, where individual conversations can be long and go into other, more irrelevant details.
In this time, it's really helpful to know a few things:
Any AI tags you feel are missing (we try to catch them all but understanding any specific AI tags we can add is great!)
Any AI tags you think should be more specific, or where multiple tags can be merged
The best way to group your AI tags into 'Themes'