At less than two years old, Sky’s digital analytics department has been given relatively free rein to test out artificial intelligence and machine learning – all the while keeping a focus on customer retention.
Headed up by Robert McLaughlin, the team has settled on a system which relies on what he refers to as “the big brain and the little brain”. The former gathers historical data to predict what a current customer might do next, before advising a targeted marketing strategy focused on driving retention (through engagement), revenue (through product recommendation) or the delivery of customer service.
Its ‘little brain’ counterpart, otherwise known as the less-sexy “contextual layers”, maps empirical customer-specific information onto this prediction. McLaughlin explained: “the big brain might say that within the next 12 months [a customer has] a high propensity to take buy our broadband product. But then the little brain might see that same customer looking at our sports product”.
In cases such as this, the small brain has the power to override the big brain when it comes to informing its media buys, and serve up an ad reminding the customer about their interest in Sky Sports, rather than Sky Broadband. The information gathered from the small brain here is also fed back into the big brain to inform future decisions.
The sophistication of this machine learning – and its real-world success – is almost entirely dependent on the quality of Sky’s first party data, of which it has a lot of. But for now, it’s taking these insights and using them to target predominantly its existing customer base, rather than as part of a wider marketing outreach strategy.
“Most of our AI activities happen with existing customers,” said McLaughlin. “That comes from a more general focus of our business today. We don’t want to lose customers and we don’t want to have to win them back – that’s a costly process for us – however in the UK market we don’t necessarily expect to grow our customer base beyond where it is today.
“That orientates us to focus on customer marketing – the revenue opportunity is in building the relationships of our existing base.”
This AI is also being applied in Sky’s conversation strategy (pre-empting whether or not a Sky user would be likely to speak to a chatbot rather than a customer service advisor, for instance), which in itself forms part of a wider goal to understand how, when and where to best target customers.
“One thing we’ve started looking at is the right time of day for specific products,” explained McLaughlin. “On Friday nights and Saturday nights, people are interested in movies and sometimes sports. But on Tuesday mornings, they’re less likely to be interested in sports and movies and perhaps more likely to look at fixing their broadband connection.
“That [insight] feels human, but it’s massively born out of data and we’ve started acting on it.”
McLaughlin, who has “great relationships” with equivalent teams at BT and Virgin Media, is hesitant to say whether Sky’s advances in digital analytics, AI and machine learning are ahead of its competitors. But, he is grateful for the freedom he’s been given to build his team and its capabilities.
“To steal some terminology from machine learning, we explore and we exploit,” he said. “It’s pretty much still unknown where the best applications of AI are going to be within a business. We’re not too caught up with whether something’s going to deliver uplift, whether its sales, service or retention. That mind set puts us in a slightly different category to our competitors – they’re trying to work out where to start, but we’re like, ‘let’s just explore’. I don’t believe it’s possible to be sure where the value is, but I want to find it.”
He added: “Culturally, Sky enables us really well. We may be [in the business of] media, entertainment and telecommunications, but at our heart we’re a tech company. My team and I are lucky to exist in that culture.”
McLaughlin was speaking at Adobe Summit in Las Vegas.