InsideSales is an interesting company to watch. On the one hand it’s a cloud software vendor that focuses on its primary sales acceleration product. On the other hand it’s an AI platform that is using its customers' crowdsourced, but anonymised, data to provide predictive analytics.
Whilst other companies are trying to figure out how to integrate AI and machine learning into their enterprise offering, it’s something that InsideSales has been doing for over 10 years. In fact, CEO Dave Elkington wrote his thesis on the idea of an AI platform that uses anonymised, but networked, data to create links and spot patterns.
Elkington was in London this week for a company event and I was fortunate enough to grab half an hour with him to discuss the company’s progress, it’s ambitions and to have a frank conversation about the state of ‘AI’ in the marketplace.
Elkington doesn’t mince his words when it comes to this subject and he had some harsh truths for other vendors in the market that are claiming to use machine learning and AI technologies.
Whilst the consumer market has AI “intertwined with everything”, said Elkington, what’s interesting is that most workplaces have no predictive analytics at all. Whilst we may trust Netflix to serve us the content we want, or for Google Maps to predict our routes, or for Spotify to recommend us some songs we may like, when we get to work we revert to manual processes and guesswork.
It’s crickets at work. What we’ve found is that users today have passed that chasm of trust for using machine learning and AI, but the enterprise seems to be really lagging.
However, that hasn’t stopped vendors jumping on the AI bandwagon - and much of the offerings you currently see in the market are just “buzzwords”, claims Elkington. But why has the enterprise been slow to tackle this? The answer is, of course: data. Elkington said:
Right now in the market, people are using predictive analytics, machine learning, AI and big data very synonymously. They’re not synonymous at all. Machine learning is a subset of AI. Machine learning is a set of algorithms that automate the answer to a specific problem. It’s a set of math that solves a hypothesis of artificial intelligence. There is no true artificial intelligence out in the market right now, it’s a buzzword. In the past 12 months it has become the buzzword du jour.
The question is - why is it all over consumer apps and not in the enterprise? It’s actually a very reasonable answer. It’s all about data. To have machine learning or AI to be truly effective, you’ve got to have legit huge amounts of data. The consumer space they have that. In 2005 Amazon was doing 84 million unique customer transactions a month. Enterprises don’t have that many customers, so being able to optimise around that is just not pragmatic.
We are in the hype cycle right now of enterprise AI, so there is a lot of BS. It’s not really real at this point. It’s not about the math, that’s been around for 60+ years. The math has been around forever, what’s different now is data. The consumer guys have had data for 15 years. The challenge is that enterprises don’t have data.
InsideSales was able to solve this problem by using the crowdsourcing approach to data, where it lumps all of its customers' data together (albeit anonymised) - whereas traditionally, companies stored their data in isolation. Elkington says that it’s nearly always impossible to provide value back to a company based purely on their data alone. You need a lot. He said:
You really can’t do that, there isn’t enough data. The question of who’s going to win and who’s not going to win in the market - it’s companies that can negotiate contracts with their customers that say we’re going to anonymise your data and do cross-customer analysis.
Given that enterprises typically feel safer storing their data in isolation - heck, even ‘cloud’ companies have been signing agreements with large buyers to keep their data in isolation - this isn’t an easy conversation to have with customers. Elkington said:
It’s hard. There’s a bit of a cold war stand off at times. When we’re negotiating, they often want the benefits but don’t want to contribute. And we have to be like ‘no, it doesn’t work that way’. It’s a really hard thing to do.
The potential of the platformInsideSales’ AI platform is called Neuralytics, and it drives every product that the company offers. Elkington explained that most companies focus on the five Vs of data - volume, velocity, variety, veracity and value - but this thinking is in fact flawed. Instead, Elkington argues:
It’s about how you store and normalise the data, more than it is about having the data there. The most important thing about data is the interrelationship of the characteristics and the way that you’re storing that data. We’ve got this theory around how you store and structure this data. We put a service on top of it, with an algorithm set in it and we’ve just opened this thing up and we call it our predictive cloud. We’ve got a couple of customers and it’s only recently we have opened this thing up.
It’s always about the question. The questions we are good at answering are about human behaviour. How people would behave in a given situation. The assumption we’ve made in our neuralytics engine is that people are the centre of the universe, that’s what we care about.
Now, this is interesting. InsideSales, a cloud company that traditionally provides sales acceleration software, because of the amount of data it holds, and the way that it uses that data, is now offering it’s AI platform as a completely separate product. This is what will take InsideSales far.
Elkington explained that one customer using the platform is a “large infrastructure-as-a-service” company, and they’re using the predictive cloud to predict utilisation more accurately. They’re putting their data into the cloud and have increased the predictive accuracy of their utilisation from 30% to 70%.
Separately, a global entertainment company has been using the cloud to determine what events to invite their VIPs to - to create a more personalised service. The cloud has been able to determine interesting details, such as, that some VIPs won’t travel more than 1,200 miles in winter. And that if you invite a VIP more than four or five times, and they say no to these invites, they don’t feel like they’re getting a personalised service and become disengaged.
So my question to Elkington was: when does the sales acceleration business become inferior to InsideSales’s predictive cloud? He said:
Right now we are chasing a huge market on the sales app. In a given month we are doing six and seven figure annual transactions every month. We are growing as fast as we have the ability to process and support those customers. But we’re also beginning to increase the volume of interest in customers on this predictive cloud. Over the next year or two I would watch us.
My next question to Elkington was: if you were to be acquired, can the predictive cloud be applied to anything? For example, if InsideSales was acquired by Salesforce, or Microsoft, could the cloud be applied to their whole businesses. Diplomatic in his response, Elkington said:
It is designed to be very universally applied. When I started the company, my ambition was to build that. And I wanted to figure out how to drive value from this. We started with the design of a very universal machine learning platform that should be able to be universally applied.
Data scientists aren’t a thing
Finally, given the topic of discussion, Elkington had some rather interesting thoughts on the skills marketas it relates to ‘data scientists’. With some reports claiming that businesses will need one million data scientists by 2018, it’s fair to say that demand for the position is high.
However, Elkington is far from convinced about the concept of data scientists. He believes that people with a bit of a background in maths and stats are putting that on their CV to boost their salaries, and then are faced with the daunting prospect of managing AI projects that are beyond their understanding. His words serve as a warning for those looking to build up their ‘data scientist’ capability. Elkington said:
Data scientists aren't a thing. If you have a PHD you have to have some experience with regression. You then put ‘data scientist’ on your LinkedIn and all of a sudden you're earning $200k. Then what that means is that you’ve got historians with a PHD working as data scientists that are managing AI for heart monitoring applications. Or driverless vehicles. That's worrying.
You’ve got half the market that isn't actually a data scientist and then the other half are small AI start ups that you're just handing your data to. If you’re after ‘data scientists’ you should hire mathematicians, statisticians, computer scientist grads. Most importantly hire people that can build. Academics like to theorise and think, but it’s better to hire someone that can actually build something.
I love Elkington’s honesty. A very intelligent guy with a lot to offer on the AI market as we know it. I’d put my money on InsideSales having a lot of success in the coming years. Why? Because the more data you have and the longer you spend analysing it, the more intelligent your systems will be. InsideSales is 10 years in, while the others are just getting started. And I would guarantee that in 5 years time we aren’t talking about its sales acceleration app as the main focal point. The interesting bit will be the predictive cloud.
One to watch.