Artificial Intelligence came up in virtually every single conversation at Cable Congress this week. The difference from last year’s show was that companies are beginning to put some clout behind their AI claims and operators are progressing in applying algorithms into their content businesses.
Telenet’s Director of Market & Consumer Insights, Kim Smets, gave the example of an algorithm built in-house by Liberty Global’s Belgian subsidiary, allowing it to see if pay TV subscribers who have begun watching a series on catch-up, will have enough time to finish watching the full series before episodes are pulled from the catalog. The AI system then pushes a notification saying there are only so many days remaining, do you want to watch this episode now?
We feel this is a very cautious entrance into the AI space by Telenet and is doing a disservice to what these machine learning algorithms are capable of. However, this use case is a start and from here Telenet can apply these algorithms vertically across its business, beyond content assets. Operator caution is understandable, considering 70% of current AI ventures have demonstrated zero return on investment, according to a panel at Cable Congress.
It’s unclear how Telenet’s content AI ties in with its recommendation software, or if they are entirely separate, but John Gee, Chief Business Development Officer for ad targeting company Alphonso TV, said, “Set top data has lag and automatic content recognition data is real-time. From this, you can feed recommendation engines – there is a trend to change this.”
Liberty Global’s VP of Advertising and Data, Laurence Miall-d’Août, claimed that 99% of the value in AI is from very simple data mapping, such as where will a consumer click on a website or what products do they want. However, she gave the example that criteria for targeting a rich, eccentric male over the age of 60 would bring up the Prince of Wales as a match but would also fit the criteria for the Prince of Darkness Ozzy Osbourne. Therefore, Miall-d’Août’s advice for the cable industry is to also apply elements of common sense before diving into the business of targeting consumers using AI technologies.
Another challenge for operators is how do they, as legacy companies, attract talent away from perceivably cooler internet companies? “We need to employ 18 year-olds who can code in their bedroom and make them data scientists, but in the correct environment, so they don’t churn. Putting them in a smaller business within an operator’s footprint means they will churn in 2 months, which has happened to us,” said Miall-d’Août.
In addition, Miall-d’Août confessed that as legacy businesses, cable operators are poor at unified data management, while claiming that the likes of Google, Amazon and Facebook don’t know how cable subscribers behave – so operators must be very careful with managing and acting upon subscriber data.
A valuable case for AI beyond content came from US business consulting firm Sand Cherry Associates, which reportedly helped an unnamed US cable company with very high churn rate, reducing churn over a five-year period by using predictive AI analytics, according to VP Michiel Sanson. Investments made it easier for the cableco to innovate and manage its subscriber base, calculating who is most likely to call the help desk and why, for example, and then pushing the most relevant offer to subscribers before they make the call, as a way of reducing churn.
We heard a similar business case at Cable Congress 2017, when Swedish operator Com Hem described a radical consumer satisfaction turnaround driven by data mining techniques based on machine learning capabilities and visualization tools. Com Hem CTO Thomas Helbo told Faultline Online Reporter at this year’s show that the operator has now seen the full impact of this on the organization, extending the method to field technicians to derive specific data from specific customers. Today, Com Hem is collecting even more than the 360GB of data every day it declared one year ago, although Helbo could not specify a figure.
Discussing AI and data mining led to the question of how operators can harness the power of voice in pay TV. EVP of Layer 3 at T-Mobile US, Jeff Binder, said, “I think the remote control will be gone in a few years,” as he described how T-Mobile US has prepared for the future by connecting with Alexa, Google Home and Siri. T-Mobile US is well-known for being a disruptor, yet Binder conceded, “We don’t have a connectivity challenge, we have a consumer packaging and marketing challenge.”
Arris CEO Bruce McLelland added, “I wasn’t a great believer in voice at first, but it really has been transformational.”
From voice to virtual reality, the panel quickly decided the industry is yet to see the killer market case for VR. McLelland mentioned Arris has several PoCs for VR, but the company is not ready for commercialization in the next few years, while Pedro Bandeira, Head of Product Development at Portuguese operator NOS, said augmented reality in sports is an intriguing case and NOS has PoCs to see how consumers will react, but NOS is around 5 years away from anything it can call a product.
Concerning how existing TV technology can cope with VR, Binder highlighted that today we are dealing with set tops with between 3,000 and 10,000 DMIPs. For VR, set top environments will require much more processing power.
Finally, content strategies were discussed in depth in Dublin this week. A debate called “Is Sports the Only Real Pay TV Driver?” resulted in a resounding answer of no from the audience and panelists. The argument was that pay TV has always been about serving a variety of content to subscribers, including documentaries, lifestyle, drama, news and sports, which together hold more value than live sports alone. Skinny bundles are a testament to this, a topic which was ignored at Cable Congress, with operators talking openly about embracing change – which is not reflected by their businesses, in reality achieving change at a snail’s pace, before it’s too late and cord cutting has taken its toll.
A cable show is naturally going to more be more conservative about emerging technologies than an event like MWC last week, but interestingly the conclusions around VR were harmonious.