The faster you can find the conversation topics that your virtual agent isn’t trained for, the more successful you will be in improving the VA’s ability to handle customer inquiries.
Let’s be honest, getting your virtual agent into production was a lift. You invested a ton of time getting it launched, and now six to twelve months in, there is an underlying expectation that it will always know the exact right things to say. But, that’s not always the case. Customers phrase questions in ways the bot doesn’t recognize, they ask about new products, or there’s updated messaging that the bot hasn’t yet been trained for.
One of the hardest jobs as a chatbot owner is finding and making sense of customer conversations that didn’t go as planned, and then figuring out whether it’s a matter of improving the NLU, or creating new intents to handle the range of inquiries.