Insights Demo

0:00

S… Speaker 1 (Insights Demo)

Hey, team. My name is Nick, and I run Partner Solution Engineering here at Cresta, and I'm going to do a quick walkthrough of the Insights product, which is part of our Cresta platform.

0:09

S… Speaker 1 (Insights Demo)

Insights is primarily targeted at company leadership and is designed to take the unstructured data within the contact center and structure it in a way that helps them make better business decisions that drive outcomes and answer critical business questions.

0:23

S… Speaker 1 (Insights Demo)

So, for example, why are customers calling? How can we improve customer experience? What's driving high handle time? What's driving volume?

0:29

S… Speaker 1 (Insights Demo)

These are all critical questions that are hidden in the context of interactions, and we're going to use Insights to help solve them.

0:37

S… Speaker 1 (Insights Demo)

So, typically, the first step in any Insights journey starts with what we call topic discovery.

0:42

S… Speaker 1 (Insights Demo)

So, topic discovery is how we start to do what the name would imply, which is discover topics that we want to focus on for a given outcome.

0:49

S… Speaker 1 (Insights Demo)

So, typically, the first thing that we do when we engage with a customer is we train a bespoke call reasons model specific to that customer and their use case.

0:58

S… Speaker 1 (Insights Demo)

So, this helps us understand why customers are calling as kind of a foundational element on which to start building our data and our outcomes and insights.

1:07

S… Speaker 1 (Insights Demo)

So, what you're seeing on the screen is the representation of that visually with the bubble size representing the volume of the topic and the topics being broken into both primary topics.

1:18

S… Speaker 1 (Insights Demo)

So, that's the larger bubbles and secondary topics, which are the smaller bubbles contained within.

1:24

S… Speaker 1 (Insights Demo)

Now, this process is a little bit different than your kind of legacy call reasons methods.

1:31

S… Speaker 1 (Insights Demo)

So, first, we don't rely on any sort of metadata, so that can be agent dispositioning, that can be IVR.

1:37

S… Speaker 1 (Insights Demo)

We do not use that.

1:39

S… Speaker 1 (Insights Demo)

This helps kind of account for those things that are not necessarily the best way to understand why customers are calling.

1:46

S… Speaker 1 (Insights Demo)

And then we also don't require any sort of manual input from the user.

1:49

S… Speaker 1 (Insights Demo)

So, a lot of other analytics platforms require a taxonomy, manual builds, keywords.

1:55

S… Speaker 1 (Insights Demo)

We cut right through that.

1:56

S… Speaker 1 (Insights Demo)

This comes out automatically after we train the model.

1:58

S… Speaker 1 (Insights Demo)

So, for all intents and purposes, you plug us in.

2:01

S… Speaker 1 (Insights Demo)

This comes out as a result, and it's bespoke to every customer.

2:04

S… Speaker 1 (Insights Demo)

We do not use – we do not share models, and we do not have industry models that are applied broadly.

2:11

S… Speaker 1 (Insights Demo)

It is very specific to each customer and their specific use case so that we can learn how customers talk.

2:17

S… Speaker 1 (Insights Demo)

So, this is the first kind of step, and you can easily see here's the largest call topics that are driving volumes on unauthorized credit transactions here in the middle.

2:25

S… Speaker 1 (Insights Demo)

You can also see fee-related account closures here on the right.

2:28

S… Speaker 1 (Insights Demo)

So, these are our two top volume drivers.

2:31

S… Speaker 1 (Insights Demo)

If I click on any one of these, it'll pull up a more detailed view that contains information such as trending.

2:37

S… Speaker 1 (Insights Demo)

It'll have sentiment breakdowns, which we're going to talk about in a moment, example keywords,

2:41

S… Speaker 1 (Insights Demo)

and also you can find example interactions with some filtering capability if you wanted to do some research and deep dive into this topic.

2:49

S… Speaker 1 (Insights Demo)

So, this is kind of the first layer of us exploring potential topics.

2:54

S… Speaker 1 (Insights Demo)

We're going to bubble back out and reset our view, and we're going to start to layer on specific outcomes.

2:59

S… Speaker 1 (Insights Demo)

So, one of the things that makes Crestwood different is we focus specifically on outcomes.

3:04

S… Speaker 1 (Insights Demo)

So, while understanding call reasons is a good kind of first step, it doesn't help solve the problem.

3:09

S… Speaker 1 (Insights Demo)

So, in order to do that, we're going to layer over some of our automated outcomes.

3:14

S… Speaker 1 (Insights Demo)

And, again, these are predictive, automated, based off of our small language models, and these come out of the box.

3:20

S… Speaker 1 (Insights Demo)

So, if I wanted to layer something over like sentiment to help drive customer experience,

3:24

S… Speaker 1 (Insights Demo)

I am quickly going to be able to find which topics drive negative sentiment as indicated by the red bubbles.

3:30

S… Speaker 1 (Insights Demo)

So, you can see here, here's a giant primary cluster around fees, as we would expect,

3:35

S… Speaker 1 (Insights Demo)

that is driving negative experiences, as well as some kind of more nuanced hidden within larger topics.

3:41

S… Speaker 1 (Insights Demo)

So, I can repeat this process for any of five predictive outcomes that we have.

3:46

S… Speaker 1 (Insights Demo)

So, sentiment, resolution rate, and CSAT score are predicted out of the box.

3:50

S… Speaker 1 (Insights Demo)

Handle time is based off of the call itself, so the metadata within the call.

3:54

S… Speaker 1 (Insights Demo)

And we also have one that's not shown here, which is around successful conversions, and that's only for sales use cases.

4:00

S… Speaker 1 (Insights Demo)

So, I can quickly navigate through these outcomes to best understand what outcome am I trying to drive

4:06

S… Speaker 1 (Insights Demo)

and combine that with the call topics to understand where to look.

4:11

S… Speaker 1 (Insights Demo)

So, we're already starting to develop some hypothesis around where to focus for different objectives.

4:16

S… Speaker 1 (Insights Demo)

We can start to drill in and start to explore even deeper with what we call automation discovery.

4:22

S… Speaker 1 (Insights Demo)

So, if I click on an individual bubble, this is going to run what's called automation discovery,

4:29

S… Speaker 1 (Insights Demo)

which essentially takes whatever topic that I selected, and it's going to break out that particular topic into its desired call flow.

4:38

S… Speaker 1 (Insights Demo)

And that's based off of the human interactions that are handling those types of calls,

4:43

S… Speaker 1 (Insights Demo)

and help us better understand how that particular call process goes,

4:48

S… Speaker 1 (Insights Demo)

and just as importantly, understand how to action that particular set of data.

4:53

S… Speaker 1 (Insights Demo)

So, I've just run an example call flow here for our account activity inquiry flow.

4:59

S… Speaker 1 (Insights Demo)

and just actually run aange set of data courses via the school Perché omnipotence.

4:59

S… Speaker 1 (Insights Demo)

So, it's a lot of interesting interesting things that are when I can do this for almost 100 years ago,

4:59

S… Speaker 1 (Insights Demo)

and this helps...

5:00

S… Speaker 2 (Insights Demo)

walk me through exactly what that call flow looks like when it's going as intended based off of the human interactions.

5:06

S… Speaker 2 (Insights Demo)

So the first thing we do is authenticate the customer. The next is we identify the issue.

5:10

S… Speaker 2 (Insights Demo)

The next is we resolve. We also have some optional management offers and call wrap up.

5:16

S… Speaker 2 (Insights Demo)

Now, one of the things that you can see in here in addition to the flow is these little wrenches that helps us understand when data might be needed.

5:24

S… Speaker 2 (Insights Demo)

And the primary reason for this is so that we can start to automate that particular topic through AI agents.

5:31

S… Speaker 2 (Insights Demo)

So one of the use cases for this, in addition to understanding how that particular call flow typically goes, is evaluating this process for automation.

5:39

S… Speaker 2 (Insights Demo)

So if you look here in the top right, we have an automation readiness evaluation that is comprised of things like resolution rate,

5:47

S… Speaker 2 (Insights Demo)

how typically this call flow goes according to this plan, excluding any deviations.

5:53

S… Speaker 2 (Insights Demo)

And it helps us understand if this particular process can be automated.

5:57

S… Speaker 2 (Insights Demo)

So we can see our automation readiness is high here in the top right, which means this is probably a good starting point for automation.

6:03

S… Speaker 2 (Insights Demo)

And the last thing we can do from here is get started on an AI agent prompt.

6:07

S… Speaker 2 (Insights Demo)

So if I wanted to start building out or testing an AI agent, this gets us a head start into activating that particular process.

6:15

S… Speaker 2 (Insights Demo)

So we don't have to start from zero. So this does jumpstart that process.

6:19

S… Speaker 2 (Insights Demo)

It's not designed to complete the process for us. So this is not a plug and play format, but it is more of helping us navigate where to start when we're mapping out this call flow.

6:29

S… Speaker 2 (Insights Demo)

And this brings us back to the wrenches, which are going to tell us when an AI agent might need to access an API or external data source to be able to conduct that activity.

6:39

S… Speaker 2 (Insights Demo)

So if we're reviewing an account activity, we need access to the customer's account activity.

6:46

S… Speaker 2 (Insights Demo)

If we're going to file a dispute, we need to understand the dispute process.

6:48

S… Speaker 2 (Insights Demo)

If we're going to enroll someone in paperless billing, we need to be able to modify the account with an API.

6:53

S… Speaker 2 (Insights Demo)

So it gets us started thinking about how can we use this data to drive outcomes with automation,

6:58

S… Speaker 2 (Insights Demo)

in addition to viewing what we previously saw around the conversation reasons and the outcomes.

7:05

S… Speaker 2 (Insights Demo)

The next place that I want to go to is kind of a follow-up for how we drill deeper into a specific topic.

7:13

S… Speaker 2 (Insights Demo)

So we use what we call AI Analysts, which is a conversational engine that will allow business users to ask natural language questions.

7:20

S… Speaker 2 (Insights Demo)

So the typical users for this can range anywhere from executives to management to product owners.

7:27

S… Speaker 2 (Insights Demo)

It basically unlocks the data for pretty much every persona that has a leadership role.

7:33

S… Speaker 2 (Insights Demo)

So it generally lowers the barrier to entry, and you can start asking natural language questions based off of the business outcomes that you'd like to drive in the data.

7:44

S… Speaker 2 (Insights Demo)

So I can ask a question such as, what call reasons are driving negative sentiment?

7:50

S… Speaker 2 (Insights Demo)

And what that'll do is help walk me through the prompt that will ultimately end with an analysis here that we're going to show in a moment.

7:59

S… Speaker 2 (Insights Demo)

So as we run this analysis, it's going to take roughly 60 seconds to run.

8:03

S… Speaker 2 (Insights Demo)

And for the sake of time, I am going to pull up a previous one just to make it a little bit more brief.

8:08

S… Speaker 2 (Insights Demo)

But it will pull up the analysis for us and generally help us explore that particular topic based off the question that we asked.

8:14

S… Speaker 2 (Insights Demo)

So if I pull up my previous version of this question, which is what topics are driving negative sentiment,

8:20

S… Speaker 2 (Insights Demo)

we can very quickly see a couple things as a result of this input.

8:24

S… Speaker 2 (Insights Demo)

So we get this kind of summarization.

8:27

S… Speaker 2 (Insights Demo)

Here are the top five reasons that are driving negative sentiment.

8:30

S… Speaker 2 (Insights Demo)

And you'll notice that the lens on these is very much context and are focused.

8:35

S… Speaker 2 (Insights Demo)

So it's not just saying here's what's driving this particular thing.

8:38

S… Speaker 2 (Insights Demo)

It is also kind of explaining the why, which helps us explore and dive deeper when combined with data and topic discovery.

8:47

S… Speaker 2 (Insights Demo)

So we immediately know the top five reasons for negative sentiment.

8:51

S… Speaker 2 (Insights Demo)

We have some quantification happening here in the middle.

8:54

S… Speaker 2 (Insights Demo)

I can jump to find evidence if I need.

8:57

S… Speaker 2 (Insights Demo)

So here's some examples of negative sentiment on unauthorized transactions.

9:02

S… Speaker 2 (Insights Demo)

And it gives me the example why it selected it and exposes the reasoning.

9:07

S… Speaker 2 (Insights Demo)

If you hover over the AI reasoning as to why this particular call was nominated as an example.

9:12

S… Speaker 1 (Insights Demo)

So I can quickly do deep dives.

9:14

S… Speaker 1 (Insights Demo)

I can find examples.

9:15

S… Speaker 2 (Insights Demo)

I can see how frequently something's happening and generally go from knowing nothing about a topic to knowing a bunch about a topic relatively quickly.

9:24

S… Speaker 2 (Insights Demo)

Now, the last thing that I want to show here is our ability to ask follow-up questions.

9:29

S… Speaker 2 (Insights Demo)

So within any topic, we can also drill down and ask a follow-up question to expose even more data.

9:36

S… Speaker 2 (Insights Demo)

So one of the questions that I ask here is what can I do to improve these transactions?

9:40

S… Speaker 2 (Insights Demo)

We know that unauthorized transactions are driving negative sentiment.

9:43

S… Speaker 1 (Insights Demo)

What can I do to improve them?

9:45

S… Speaker 2 (Insights Demo)

And AI analyst comes back with some recommendations around what we can do to potentially start improving this particular topic.

9:52

S… Speaker 2 (Insights Demo)

So this is one of the avenues through which we transition from just having data and a hypothesis as far as what might be driving.

10:00

S… Speaker 1 (Insights Demo)

negative outcomes to expediting the data analysis process to understand the why of how that

10:06

S… Speaker 1 (Insights Demo)

particular interaction can be improved.

10:08

S… Speaker 1 (Insights Demo)

So hopefully this was a good intro to insights where you can understand how we can go from

10:14

S… Speaker 1 (Insights Demo)

having a context and interactions that are previously unstructured, structuring them

10:19

S… Speaker 1 (Insights Demo)

to understand call reasons, understand outcomes, and eventually get to the point where we understand

10:23

S… Speaker 1 (Insights Demo)

the why negative outcomes might be happening and start to address it at a process level.

10:30

S… Speaker 2 (Insights Demo)

Thanks, everyone. If you have questions, please feel free to reach out. Thank you.

Rezumatul

Întreabă Al despre acest text