Main content

InsideSales CEO predicts data science bonanza, part 1

Phil Wainewright Profile picture for user pwainewright October 6, 2015
Fast-growing sales acceleration vendor InsideSales has always seen data science as the key to efficient inside sales teams, its CEO Dave Elkington tells us

There's so much to be done in the world of predictive analytics. If you like data, life is good right now. Data scientists are the new kings.

Dave Elkington, CEO InsideSales
Dave Elkington, InsideSales

There's no question that Dave Elkington, CEO of sales acceleration provider InsideSales, is excited about the untapped potential of predictive analytics to discern hitherto undiscovered patterns in human behavior. He told me about his obsession last week at a London launch party for the vendor's EMEA presence.

We're in the middle of the Wild West of predictive analytics and big data. I'll go to one of my data scientists and say, 'See if there's a correlation between buying and the Superbowl'. And sure enough, we actually found there's a correlation between buying patterns in cities representing football teams and the output of the Superbowl. [By going back over the data] we actually predicted the winner, we even predicted the score range of the two teams. Because everything is intertwined.

That future has been a long time coming for Elkington, who founded InsideSales eleven years ago. Even before that, he told me, he had always had an interest in mapping patterns of human behaviour.

My undergraduate was in philosophy — InsideSales is practically my senior thesis.

I was particularly enamored with epistemology and set theory and learning theory. People are very pattern-based in how they learn. [Most people] drive to work the same way every day, [they] have more or less the same breakfast almost every day.

If you know those observations then you can accurately predict what's going to happen. If you know that, you can then prescribe how to optimize interactions.

It's not about gaming people by the way — that's the bad side of predictive analytics. The good side is personalization.

So InsideSales applies these philosophical principles to sales interactions, detecting the patterns that lead to better sales outcomes and then steering sales reps towards those successful patterns. For example, the analytics show that calls are more frequently successful when it rains, or when the local baseball team is doing well. So it makes sense to call prospects in those locations first.

Finding success patterns

Of course, when the company launched back in 2004, it didn't have that data. Elkington got together with Ken Krogue, who had built the inside sales team at time planning vendor Franklin Covey. They chased down the web domain and became a top search destination for businesses researching the then-emerging phenomenon of office-based sales teams that had started to use web leads, telesales, and web meetings to drive B2B sales.

Later, the company teamed up with Kellogg School of Management and MIT to research success patterns when handling web leads. For example, studies showed that the optimum time to respond to a web lead was within 5 minutes, whereas the average company took 46 hours. The optimum number of follow-up calls was 6 to 9, whereas sales people typically made only one or two.

The company used the research to help it build better processes into its application, each of which led to incremental improvements in its customers' success rates. But it was also building a predictive model, one that was able to learn from success and automatically add those incremental improvements. Now that it has reached scale, the predictive model is starting to snowball, as Elkington told me.

We have 60,000 users on our platform and up to 3,000 companies. Every time they use anything on our platform, we're anonymizing and we're aggregating it.

In this data structure, we have over 120 million unique buying personas. More interestingly, I have almost a hundred billion sales interactions with those 120 million people. A sales interaction's a conversation, an email, a response, a visit, a purchase. We're adding roughly five billion of those a month. The reason is because it's aggregate, it's crowdsourced.

The aggregation means that even the most bitter competitors can benefit from each others' experiences. In the HR software space, InsideSales has ADP, Paychex, Zenefits and Workday as clients. Two of them have a ongoing lawsuit between them. But they're all on the platform, said Elkington.

They're all helping each other to do better. A sales rep at ADP, if they make a mistake — none of my clients, including Microsoft, Salesforce and Dell, can ever make that same mistake twice. Because the algorithm learns for next time. We call it interactive cognoscence. It's getting smarter with every interaction.

Data potency

The data that InsideSales collects is especially potent because it's the most timely and authentic of the four main classes of data that people collect, Elkington explained.

  • One, legislative data. If you've paid your taxes, unfortunately, you're profiled by the government. The problem with that data is, one, we're incentivized to give as little as we can. We don't want to go to jail but also we don't want the government knowing any more than they need to. And it's not very timely. You don't get your driver's license every year, it's every four years or very five years. So it's not very timely and it's not incredibly accurate.
  • The next level up is brute force data. That's Dun & Bradstreet. You'd have call centers in the Philippines just hammering phones, calling companies. It's actually accurate because, when we volunteer data, we want to minimize it — and sometimes we want to falsify it — [but] when it's observed, [it's hard to] lie. So the accuracy rate when it's brute force is much higher, but its timeliness is actually fairly bad because you can't call seven billion people every day.
  • Social media brought that next revolution of data, which is crowdsourced active data or volunteered data. Facebook, LinkedIn, Twitter, every time we do anything we're posting that real-time. Its timeliness is the best, but its accuracy is almost the worst. You think about people's LinkedIn profiles. Is that actually their career? Or is it more the career they're trying to — curate, we'll say. Even Facebook, people maybe exaggerate around the edges a little bit about their lives. So the problem is, it's super-timely, but its accuracy is nominal.
  • What we're doing is called passive crowdsourcing. We're passively observing all of the things that happen, meaning, you can't lie because we're just watching what actually happens. What's different [from social and web analytics] is that we're actually crowdsourcing the activity of all of those sixty thousand sales people who use our platform, so we actually have the knowledge of the math. So it's the advantage of social media with the accuracy of the passiveness.

In part 2, we talk about InsideSales' ambition to become a predictive analytics platform and how B2B sales processes are evolving in the digital era.

Image credits: Finger touching statistics screen © Ruslan Olinchuk –; headshot courtesy of InsideSales.

Disclosure: Workday is a diginomica premier partner.

A grey colored placeholder image