Chat with your data using Conversational Analytics

Key Takeaways

[Music] Hello. I'm here today to take you through how you can have a conversation with your data using natural language using Looker's new conversational analytics feature. There are a few different ways that we can start a conversation with your data. Firstly, we can click create and conversation at the top left hand corner. Secondly, on the menu option, you'll notice there's a new conversations tab. And then lastly, if you're within an explorer, we can actually go and start a conversation directly on top of that existing explorer. Can also use agents within the conversational analytics experience, but we're going to touch on that in a later video. So from here what I can do is I can start to ask questions of the data like what was daily events for the last 30 days and the LLM will leverage the context of your semantic model to answer that question. Firstly it might recompose the question that's been asked like we can see here where it's saying show daily events for the last 30 days. We then get the query result. So we can see the fields that have been selected. We can then also open this in the looker explorer if we want to add this to our dashboard or we can go and look at the insights tab at the bottom for some interesting insights into what the data is actually showing. Now what we can also do is ask follow-on questions like show me this as a line chart to change the type of configuration that we're looking at in the conversation. Or what we can do is ask a follow-on question like break this down by product or country to allow for an additional understanding of what the data is actually showing us. So from here you can now see show daily events for the last 30 days broken down by country. And now it's also going to give me a line graph of that nice trend over time. The conversation analytics experience also allows end users to ask questions of the data that previously might have taken multiple questions to get to the bottom of. For example, what I could say is, "Show me monthly events for the last 12 months for the country with the lowest bounce rate. So what you'll notice here is that firstly I have to identify what country has the lowest bounce rate. Secondly, once it's identified that the UK is that country, it's going to inject it into the follow-on question to show what event count was like on a monthly basis for the previous 12 months. So, we can start to see that trend over time. These are just some examples of the questions that you're going to be able to ask leveraging Looker's new conversational analytics feature. [Music]