kentoh - Fotolia

Sky mixes Adobe machine intelligence and intuition

With data collected from the Sky Q box, the broadcaster can figure out who’s watching what and why customers’ viewing habits change

Sky is using Adobe Experience Manager to provide cloud-based omni-channel orchestrated customer communications.

The broadcaster’s head of digital analytics, Rob McLaughlin, says it is using the Adobe software to increase revenue, retain customers and deliver customer service more efficiently.

Speaking to Computer Weekly about the work Sky has been doing with Adobe to better understand its 11 million customers, McLaughlin says: “With a diverse group of customer demographics, we don’t have a customer archetype. Customers want a whole set of different things.”

Given the rich diversity of the customer pool, McLaughlin says common marketing approaches such as customer segmentation can only take the business so far.

“We have a long history in creating personas, and we are now working with Adobe on a programmatic approach, using machine learning and artificial intelligence to connect individual customers with relevant experiences,” he says.

To do this, Sky identifies its customers through their login. With six to eight devices per household, it also uses a device graph fuelled by Adobe Marketing Cloud’s identity service.

Customer identity is also useful in helping to deliver a good user experience for customers who leave Sky and return at a later date.

“We are using machine learning and artificial intelligence to connect individual customers with relevant experiences”
Rob McLaughlin, Sky

“When a customer has authenticated in the past, we hold their profile. If a customer hasn’t authenticated for six months and then comes back to us, we are able to connect all their previous browsing history to their account,” says McLaughlin.

The customer’s ID is available via Adobe Marketing Cloud, enabling Sky to analyse exactly how each customer is consuming its services, from high to low levels of engagement. There is a sizeable role for machine learning in this area.

Recommendations from historic data

“We deploy machine learning to trawl historic information on what potential products might be effective for customers,” says McLaughlin.

This kind of product recommendation engine, combined with traditional regression analysis, contextual layers and real-time signals such as browsing or viewing behaviours, is the essence of Sky’s approach to customer analytics, through what McLaughlin describes as Sky’s “central brain approach”.

This goes further, as Sky can also tap into network analytics from its set-top boxes. “We are a subscription business, and our devices are physically in people’s homes,” says McLaughlin.

Arguably, the Sky box was one of the early home internet of things (IoT) devices. The network data it provides helps Sky deliver a better viewing experience for its customers. “When there is contention on the network, we can proactively tell the customer that their network is reaching a limit,” he says.

Given that the Sky Q platform is a broadband router, McLaughlin says it is able to run diagnostics regularly. It provides a Wi-Fi mesh network across people’s homes, which shows Sky exactly what is being watched on which mobile device, enabling the company potentially to deliver targeted adverts. Now that the Q box is voice-enabled, there is an additional opportunity to understand who is watching what on TV.

Combining human and machine intelligence

However, according to McLaughlin, not all of the clever stuff is being achieved through artificial intelligence. “We believe in human and machine intelligence,” he adds. If children’s TV is being watched, it makes sense to promote to kids, he says. Machine intelligence is not really needed for a decision that is intuitive.

But by making the data available to Sky’s algorithms, McLaughlin says it is starting to make use of the viewing data being collected, to try to uncover trends that are not as intuitive. “We have been doing this in partnership with Adobe and using GCP [Google Cloud Platform] as our big data platform,” he says.

Read more about AI and analytics

  • With customer data growing in ever-evolving volumes, we look at how companies are using data analytics to stay in control.
  • Research shows that customer experience will boost business, but IT and the people across the organisation need to embrace data analytics.

In the past, if habitual usage tailed off, McLaughlin says Sky would concur that something had happened. “When someone stops watching box sets, they are probably enjoying something else from Prime or Netflix,” he says.

Sky now uses partnerships, such as the one it has with Netflix, to provide a fuller service for its customers. McLaughlin says Sky has been looking at developing second-party relationships, such as offering Netflix content on the Sky Q platform. “My team does not believe in a 360-degree customer view. Where there are powerful signals, we work with a second party because it is worth doing,” he says.

The company is also forging partnerships with utility providers. “There is huge churn when customers move house,” says McLaughlin, so such partnerships almost always start as a data sharing agreement, where Sky and the utility look at shared customers and then try to figure out what they can do jointly. This is not necessarily data analysis, he says, rather the two companies looking at how they can develop something proactively targeted at the shared customer.

Read more on Artificial intelligence, automation and robotics

Data Center
Data Management