Built for Scale: Conversational Analytics

By October 31, 2019 Pypestream Digital Labs

What types of KPIs do enterprises set when deploying conversational AI for customer service?

Claudio: Enterprises set two types of goals. So, it’s breaking down really in the experience and then more like the bottom line cost reduction for the businesses. So, when the experience side it really comes down to NPS, are we delivering quality engagement, quality service to our end users or our consumers. On the other side, it’s really how much ROI are we getting from the conversational solutions that we’re deploying.

What are conversational heatmaps? How are they used?

Jack: Our heat maps are, actually, a very powerful tool for our customers to, not only visually understand what’s going on in there automation tool that we built for them, but also to quickly, at a glance, identify key areas of opportunity. So, I like to describe a heat map as you’re putting on a set of glasses and you have a different, different set of lenses depending on the view that you want. If you want to see abandonments, if you want to see the most popular traversed paths, where as people are search exiting – this is what our heat map can do for you.

Really being able to drill down into these key components when you have complex solutions ranging from thousand, two thousand nodes is really where the heat map will shine.

Elana: Conversational heat maps are also used to see drop-off and where users are struggling so that we can provide insight back to the customer and optimize for further use cases.

Claudio: So, it really gives this visibility as far as what are the areas of opportunity where we can continue to optimize, continue to make the solutions better, and what are also the most efficient routes that we can continue to capitalize on, and what’s working well as far as the, the end result on the KPIs that businesses are trying to influence.

What is an Engagement Dashboard? How is it used?

Claudio: The engagement dashboard really provides this high-level view of the top level metrics that a company might be interested in. So, this can be anywhere from how many automations that we have on a particular use case, how many claims did our customers file, you know, in the last seven days, last 30 days? So, it gives that flexibility of knowing what is happening within the solution – are we actually having an impact with the automations that we were trying to accomplish, as well as providing some high-level visibility into overall volume and traffic as it flows throughout the solution.

Jack: You can drill down into specific subsets of dates, see different kind of out-of-the-box metrics, oftentimes ones that clients are familiar with, such as total aggregate chat volume, chat volume by day broken out by agent, but, in addition to that, we do augment it further to, kind of, bring in what the messaging paradigm has in terms of contextual awareness of where people are going. So, as part of KPIs we might be looking at key milestones that people have achieved, how many people are dropping off at different points, all of that, we may visualize, in the engagement analytics dashboard.

How do Pypestream customers use raw data?

Elana: Pypestream customers can use raw data any way they want to. We will give them a dashboard of all the things that we’ve previously talked about in their metrics, but once they have access to their data it’s free for them to do whatever types of analysis they want to.

Claudio: What we found is that customers like linking it to their own analytics tools and also marrying it to their own first party data set. So, for example, knowing the performance of a particular call center in the context of the data that we give them, gives them a visibility into performance of a solution and be able to look at both side to side, really the, the call center performance as well as the conversational solution performance and being able to really calculate what is the, the cost saving ROI of the entire interaction and engagement with, with Pypestream as a whole.

Why is conversational data important to an enterprise?

Claudio: First of all, data is insight and data is, its power, data runs the world. So, really being able to have insight into how things are working empowers our customers to really make the changes needed to further optimize towards their goals.

Elana: Conversational data, I think, is the most important to enterprise evolution because it tells the enterprise exactly what users want.

How is conversational data more actionable than traditional website clicks?

Elana: Conversational data is the most actionable data that you can gather because website clicks show you what the user is doing when they have the options, conversational data is letting your user tell you what they want.

Claudio: Conversational data is really valuable for businesses because it provides a lot of visibility into the intent and into the, the actual needs of users as they interact with the solution. Whereas, historically speaking, looking at just click data just provides a visibility into actions but not necessarily intents and really conversational requests, for example. So, really providing visibility into what users actually want from their businesses and how they want to be serviced, it’s incredibly insightful.

Analytics is still a very open space to be exploited and, and we’re definitely pioneering in the solutions that we’re building within the analytics front.

Is Pypestream data shared with third parties ever?

Ray:  So, one of the differentiators for Pypestream is that our basic mantra is that a customer’s data is the customer’s data. We never share that data with any other third party. We do perform some analytics on that data, but that, those analytics are solely for the benefit of the customer that owns that data.

 

 

Watch more Built for Scale here.

Leave a Reply