Playing the long game: Using data to drive new sales
Rob Symes, founder of The Outside View, talks to Liam Ward-Proud about how predictive analytics has helped his firm find new customers, but why it won’t solve all your organisation’s problems
IT’S OFTEN difficult for businesses to know how best to use the sheer volume of data they generate. Information on interactions with customers, the performance of key salespeople or business units, and even data from public records and social media sites often sits idle. According to Rob Symes, chief executive of The Outside View, this is set to change over the next decade. His company sells predictive analytics solutions to other businesses, helping them find ways to streamline processes, and replace the guesswork of salespeople and managers with conclusions based on hard data. And The Outside View has implemented predictive analytics itself – the firm “eats its own dog food”, as Symes puts it. He tells City A.M. about using big data to find new sales leads, and the teething problems the company faced in rolling out new technologies.
How have you incorporated data analytics into your business model?
We see data as the new oil – it’s sitting in businesses, often completely unused. A lot of companies have invested in customer relationship management (CRM) systems, for example, but they’ve sometimes got no idea what to do with all the information that has been gathered. Others talk about data visualisation, self-reporting and forecasting in businesses, but even these uses of data don’t get to the extent of what’s possible.
One way we use data ourselves is through predictive analytics, helping to identify new customers. We take data from a huge amount of sources, including Companies House and our own historical sales records, to find the firms that are likely to buy our products.
But then this is cross-referenced with a vast amount of information about the individuals working within the companies. It’s not just the firm as a whole, we’re also looking for the people who are likely to respond to a sales approach. We call it micro-targeting.
This involves taking data from news feeds, LinkedIn, Twitter, Facebook, as well as anything that’s in the industry press for the sector we’re looking into. We also have historical data on the kind of people that have bought our technology in the past.
Specifically what are you looking for in all this data?
One type of person we’re seeking is an early-adopter – someone who’s curious about the way the world works, and is likely to be interested in our services. Another way of putting it is that they’ve read Moneyball (Michael Lewis’s book on the use of statistical analysis in baseball), and not just watched the film. If they’ve been interacting on LinkedIn around that topic, then clearly they’re of interest to us.
It’s about fusing all these different sources of data together to try and understand what the person is interested in, why they’re going to care about our services, and how we can use this information to better target them.
A lot of the data you’ve referenced are non-quantitative, like posting a link on Facebook. How can you process this type of information?
We class that as unstructured data, and processing it involves taking something that isn’t easily tabulated, and making it so that you can interrogate it.
To do this, we use sentiment analysis, which means analysing the content – the key words used, and the tone of what is said in the Facebook post or Tweet. We can score someone’s excitement levels when they’ve just posted a link to an article, for example, and then use that information to judge whether they’re likely to be responsive to a sales call.
How successful has the data-led approach to sales been?
Before we used this system, the biggest problem we had was actually finding people who might buy our services. Our salespeople would spend a lot of time researching, trying to find the right people to call, but less time on the phone actually selling. This wasn’t an efficient use of time.
Now we’ve got the automated research system in place, our sales call times are about 33 per cent longer, which is a fairly good proxy for the quality of the call – people are more interested in talking to us. But we’ve also seen the conversion rates go up, in terms of converting contacts into calls, and then calls into actual sales. Our growth statistics reflect this.
Are there any other areas you’ve used data analytics to make improvements?
When our salespeople are on the phone, we use voice recognition software to automatically transcribe the calls in real time. All this information is automatically stored in a database. We then run a sentiment analysis on the text in order to get an idea of the quality of the call, and all this is fed back into our predictive engine in the cloud so that we can guide the salespeople in real time on what to say to improve the call and boost conversion rates.
What challenges did you face when you were rolling out these predictive analytics technologies?
Initially, we found that there were some dead ends. For example, we’d call a lot of people in the early days who had expressed an interest in Moneyball or analytics, but it would turn out that they were working at the wrong level of the business, their department was irrelevant to what we were selling, or they had left the company.
The key point people need to understand with predictive data analytics is that it won’t solve all of your organisation’s problems or give you all the answers immediately. It’s about probabilities – over a period of time, the results will get better.