 Hi, this is George Gilbert on the ground at the Marriott Marquis at the Data Science Summit in San Francisco. I'm with Andy Twig, Chief Scientist at Inside Sales, one of the hot companies that's come out in the last few months. Andy, why don't you tell us a little bit about the evolution of Inside Sales and then let's dig into the technology and the application that's got behind it. Okay, sure, thanks. So Inside Sales has been around a number of years. I actually joined as part of the recent acquisition of C9. So that was completed in May this year. Inside Sales is basically a sales acceleration platform and it provides a whole bunch of tools for sales teams, primarily Inside Sales, as the name suggests, in order to generate better quality leads and convert these leads. And they use a lot of predictive analytics and big data and machine learning to do that. C9 was more focused on the enterprise side. So we had products that would look at companies' pipelines, sales pipelines, and we had an application around sales forecasting that a lot of people used. So the acquisition brings those together to sort of expand the reach from the top of the funnel down more into the sales opportunity space. To paint that picture, at the top of the funnel, it would be improving the quality of the leads or improving the effectiveness of the Inside Sales force to bring a lead closer to the point of where it becomes an opportunity. And then the machine learning and predictive analytics would be applied towards figuring out so what's the likelihood that any one of these is going to convert? Is that right? That, I think that's a pretty good read, yeah. So the software that Inside Sales had before, the PowerDyla, this is really focused on if you're a company of a large Inside Sales team, making, say, 100 dials a day, you want as many, two things, you want to make as many dials per day as you can. So you want to cut out the time, researching prospects and that kind of stuff. But you also want to make calls to the most likely, to the leads which are most likely to convert to opportunities. Okay. All right, so that's a pretty clear explanation of the application. So tell us about some of the building blocks that made that possible. I think the most interesting building block is the data itself. So Inside Sales has around 80 billion, I believe, data points on all sorts of sales interactions. So these are data points associated with emails that people have made to reps or meetings or phone calls, all that sorts of stuff. This is a really huge data set and it allows us to do a lot of stuff. Now, that data set sounds like it's a lot of personal data. How is that harvested and how is it cleaned up? It's, first of all, it's all anonymized. So there's no possibility of leakage there. It's harvested primarily through the applications themselves. So when companies and reps are using the tools to do their day-to-day business, these things are collecting a lot of data about how people are using it. For example, when there's an email sent through the tool, it will register that and it will register some interesting features about the email and did the guy respond? Did he not respond? How long did it take? How long was a call? Did he even answer the call? So this isn't syndicated data. It's instrumenting the use of Inside Sales itself. Correct. Yes. Okay. Inside Sales really owns this data set. Okay, okay. Interesting. So then now that you have this rich data set, what analytics are used on top of it to make the forecasting and the prioritization more effective? We use a variety of tools. We use a lot of different machine learning algorithms, random forests, some deep learning, neural networks. The main product around data science and using the data set is actually called Neuralytics. So it implies some use of neural networks, which we use. Can you tell us a little more about the combination of the tools that you're using and which contributes in what way? Could you elaborate a bit more? Okay, so it sounded like you're using different machine learning models. And so why would you help us understand what each contributes? And for instance, in the case of Neuralytics, what's its role? So the role of Neuralytics is as a platform to basically make the best use of this big data set for the applications that we have inside Inside Sales. So for example, for the Dining application, this Neuralytics really provides a strong foundation to let the application know, how should I order these leads? Which leads should I call right now? That sort of stuff. And for our forecasting applications, it allows us to determine, well, I've got these 10 opportunities or 100 opportunities in my pipeline. What's the probability that this given one will close this quarter? How much is it likely to contribute towards our revenue at the end of the quarter? So it does a wide variety of things. We've heard of dozens, if not more, B2B marketing applications over time. Is this a generational shift where the other ones who are mostly sort of tracking applications and this is much more on forward-looking? In other words, tracking were more like backward-looking historical performance reporting. And this is more like let's use some intelligence to make better decisions about what we're going to do going forward. I'm not an expert in B2B marketing. I would say, for example, lead scoring. There are many lead scoring applications. Maybe these come under the same umbrella. The main difference I would say between lead scoring and inside sales is that the lead scoring application, like you say, could be a plug-in to Salesforce and it might append each lead with a score. Or it might say as a dashboard, like you say, historically, you convert these sort of leads with this ratio. What inside sales does is it really takes out a step further, right? Because it says it's more about how can we use this to drive actions and you have an application there in which you can actually drive action, the people using it to dial. And because the machine learning has as its output predictive models, those go and get operationalized in the application directly without user intervention. That's right. Yeah, they're embedded within the applications, yeah. And have you guys started thinking about future applications of this sort of technology to go even steps further? Yeah, I believe that it does have applicability elsewhere and I think that's something that we're working on. I mean, available. Working on, but can't talk about. We'll have a big booth at Dreamforce, I believe. So come back then. Oh, we'll have to come back then. Good. Great. Thanks for filling us in. That was very exciting. All right, thanks a lot. Cheers. George Gilbert on the ground, Data Science San Francisco.