Automated Resolution

You've put in the work to get your bot up and running. And now that it's live, it's in every channel your customers are.

Your goal for your bot is to resolve the most inquiries with the least amount of effort for your team and for your customers. And now you're ready to start measuring how your bot is performing against that goal. Enter Automated Resolution.

Measuring Automated Resolutions will revolutionize how you quantify your bot's performance, helping you focus on what's most important to your customers.

Automated resolution or AR refers to a fully automated conversation between a customer and a business that's relevant safe, and accurate and does not involve a human. By relevant, we mean that Ada effectively understands the inquiry and provides directly related information or assistance.

By accurate, we mean Ada provides correct up-to-date information.

And by safe, we mean that Ada interacts with customers in a respectful manner and avoids engaging in topics that cause danger and harm. Finally, by no human involvement, we mean Ada handles the chat without any agent support.

But how does Ada actually measure your Automated Resolution rate?

Every hour, Ada takes a sample of engaged conversations that occurred twenty four hours ago. A large language model assesses each conversation transcript for accuracy, relevancy, and safety, and labels them as resolved or not resolved. Resolved conversations that did not go to a human agent are automatically marked as automated resolutions.

Let's look at AR in relation to a similar metric you've likely heard of containment.

Remember, containment tells you how many conversations didn't reach a human. But without context, it can only tell you so much. It can be harder to tell how many of those contained conversations were resolved and why. 

Did a customer end a conversation because they were satisfied? Or did they leave without the answer they needed? Containment would tell you these situations were successful because they both deflected from an agent.

AR, on the other hand, is about more than whether a conversation ended without agent involvement. It gives you more insight with the help of machine learning to tell you if the customer got the help they needed.

Because only contain conversations count as automated resolutions, your containment rate will always be higher than your AR rate.

While the industry standard metric of containment has come close to measuring success, it's only a piece of the puzzle. Automated Resolution rate is more holistic, and it's a better indicator of your bot's performance.

In the reports section of the Ada dashboard, you can monitor AR rate in the Automated Resolution rate report.

Your AR rate depends on your bot's content. Which is completely in your control. It's easy to see where it's working and where you can make improvements. AR and continuous improvement go hand in hand. To identify opportunities for improvement, just scroll down to see a record of resolved and unresolved conversations and review their transcripts.

Within a conversation, you can explore why it was labeled resolved or not resolved. You also have the ability to provide feedback if you agree or disagree with a resolution label. This helps Ada improve our classification model. It does not override the automatic classification your bot gave this conversation.

By measuring Automated Resolution, you'll have an improved understanding of which conversations were truly resolved and which were not, and have the insights to continuously improve on them. AR does the heavy lifting of decoding your bot's effectiveness so you can focus on what matters most, coaching your AI and continuously improving your CX with data backed insights.

The more Automated Resolutions your chatbot can provide, the more you'll empower your customers and your team.