Artificial Intelligence Applied to Sales: What’s Under the Hood of Neuralytics?

2017 is the year of Artificial Intelligence. We’re now using AI in every aspect of our lives, from entertainment to navigation and shopping. In our professional lives, AI has the potential to improve results significantly – by increasing revenue, minimizing time wasters and by making our jobs easier.

For years, artificial intelligence has been a scary concept. We all remember the eerie reply given by HAL 9000 in the “2001 – a Space Odyssey” movie: “I’m afraid I can’t do that, Dave.”

It’s time we demystify the concept of AI and understand how it can work for us to improve business outcomes.

At XANT, we fundamentally believe that salespeople can sell more by incorporating both data and science into the way they work. We accelerate the sales process using Artificial Intelligence (AI) technology.

I’ve  spoken to Michael Murff, Principal Product Owner at XANT for the Neuralytics software, who showed me how AI works to make businesses thrive.

Michael Murff had been working within PayPal’s predictive analytics group for the last 12 years and recently moved from Silicon Valley to work on the XANT Neuralytics project.

When it comes to data and analytics, there’s more questions than there are answers, but we want to get into that today. At a high level, Michael, when you say Neuralytics what do you mean? What is the definition or the idea behind Neuralytics?

At the highest level, Neuralytics is lots of sales data plus AI models , which, together, produce scores. These scores are provisioned to clients through our predictive applications like Playbooks. Our application stack is how we communicate our predictions to our customers to accelerate their sales.

Tell us a little bit about predictive analytics and AI and how it fits into the sales world.

Artificial intelligence has the potential to fundamentally change the way we work today in sales. The old way of doing selling was very manual, there was a lot of paper involved and it was a relatively inefficient process.

We have so many tools and predictive technologies today that can help us accelerate sales and make it more efficient and effective.

Talk to us briefly, Michael, about the concept or predictive analytics. Is there a difference between predictive analytics and AI?

Predictive analytics is an umbrella term for many methodologies for processing data and deriving insights. Predictive analytics may also include things like descriptive analysis or summary kinds of data analysis.

AI are a set of machine learning based models: neural networks, random forest, deep learning–these are some of the buzzwords you hear.

If you think about what’s happened over the last several decades, we’ve seen a very significant shift from reactive to prescriptive kinds of analysis using data and models.

If you hark back to the 80’s and even earlier, companies were using pretty small data sets of several thousands of transactions. Analysis was retrospective and sales operations  didn’t have a lot of automated systemization in this process.

Data sets today are much larger. Companies are able to answer a richer set of questions, and analyses are happening in near real time. This evolution of data has enabled some very significant changes and now we can automate sales and sales interactions.

For most people, AI is a newer concept as they think about using data in a sales approach. I’d love for you to continue our conversation and explain a bit more about AI-driven selling.

Neuralytics enables salespeople to sell more by incorporating both data and science into the way they work.

It is is the only AI platform for sales built on 10 years of cross company’s sales interactions. This provides huge insights for us to make very accurate and reliable predictions.

Neuralytics has sales specific algorithms addressing key questions across the whole sales funnel which are then integrated into our sales applications, like Playbooks.

The predictions and scores are used to deliver insights and gather more data. This creates a virtuous cycle such that the next generation of models we release are even smarter.

What kind of data is available to the Neuralytics systems?

Neuralytics is based on customer relationship management (CRM) data in part, and that includes: leads, accounts, contacts, opportunities and forecasts, and those data are going to be snapshot as they change. Once we sync up with the customer and connect our pipes we’re going to see the full history of their data – we also make sure the data is secure.

We can derive insights like: call time, duration, the results of the call, if it was the correct contact, if it was a voice mail or a bad number. We capture the source and the recipient phone numbers, as well as call recordings, which are encrypted. Add to that all of the e-mail data, email templates and the tracking insights as you interact with customers.

Additionally we have sales activity data, so the actual outcomes from those interactions, the number of calls, the number of e-mails, appointments that occurred, demos and data about appointments and closes.

We also collect forecast data (quarterly, year to date, and near real time forecasts) through our Predictive Pipeline product.

Finally, part of our secret sauce is we’ve taken many external data sets. Whether it’s crawling the web or connecting with third party vendors to gather things like firm characteristics, economic indicators, web details and stock prices – we collect all this information and factor it in.

That kind of enrichment process allows us to combine CRM data with the call data, the e-mail data, the outcomes and the forecasts to really build a very powerful data set which we’re calling Neuralytics.

I’m interested to get into a little bit of how you can apply this to real-world sales, but I want to make sure we hit this concept. Can you talk about these external data pieces and how they fit into this Neuralytics model?

Our models use five classes of information.

Histographic is a record of all of the sales interactions made for a lead or account. It includes the number of calls, emails, texts or other interactions – and their timing. These are histographic variables which can be very rich and powerful and predictive.

Firmographic information includes: how large is the company, its industry, the revenue and funding events that have occurred.

Demographic information describes key attributes about populations of leads, accounts, and contacts. For example, we create specific variables around the contact title and the lead source.

Geographic information: We use geographic factors like location, region, and other geolocators.

Psychographic variables: These are unique variables to Neuralytics that proxy for consumer sentiment. We consider information like stock data and macro economic GDP changes.

All of this information is factored into our model building process. These variables that have proven to be quite useful for many of our customers.

So how do you get sales performance lift, using all this data? How do we close more sales using AI-powered systems?

We have some fantastic numbers to prove that Neuralytics is working and that the models are performant.

With this system, all of your leads and accounts are scored. You have a number saved on a range of one to 100, where larger numbers are more likely to win. As you go down, those are less qualified leads or accounts. The leads with the top 50% of the scores are what we call an AI-recommended item. These scores that are highly predictive. On average and over time they will perform much better than simple human intuition – or just calling your items in random order.

We’re seeing in those upper bins a very significant lift over the baseline rate (what you currently have without using modeling).

With Neuralytics, our customers have seen:

• • 2 to 3X lift off base contact rate for AI recommended leads (compared to non-AI-recommended leads)
  • Nearly 2 to 3X lift off base conversion rates on AI recommended leads
  • 2 to 3X lift in close rate
  • 30% lift in revenue
  • 85% larger deals

Once we provision a model, using the score correctly is a key element. Our customer success and  account management teams will ensure that you get the best value out of these scores as you implement them in your Sales Operation environment.

Can you give us a summary of how Neuralytics supports some of the other parts of the XANT platform?

Neuralytics fits into a system of growth that’s automated and enabled by data. Our Neuralytics engine is going to sit on top of the customer CRM, and pass scores to a set of applications that run on  different parts of the sales process.

We’re going to use Playbooks as tools in the pipeline building process that are leveraging those Neuralytics scores. You are going to to be able to answer questions like which deals to focus on, what actions to take, and what can I forecast.

If someone wanted to experience Neuralytics in a more tangible way, what would they have to do?

If your interest is piqued, we can offer a predictive scan. This is a free demo of how models work for your data. It’s very simple.

You input a sample extract of CRM records where you define the success criteria what you want to predict. That information will go into our Neuralytics engine and produce a set of outputs, which includes scored outcomes as well as model results.

This way, you’ll be able to see is this product working for you, it will show you how Neuralytics enables your sales and accelerate your process.

If you are interested in how this works, you can contact XANT for a predictive scan. We’re always interested to learn more about your sales process and how we can optimize it.