Home Analysis How improved data insights are delivering real business and UEX outcomes for...

How improved data insights are delivering real business and UEX outcomes for TV today

Share on

The power of data analytics (including with the use of machine learning and artificial intelligence) to transform television becomes clearer every month. And one company that is leading the charge to a more ‘intelligent’ future is Liberty Global, which used Cable Congress 2018 to give some examples of how its recently formed media insights division is delivering real-world outcomes. A better understanding of viewing behaviour has enabled TV3 (the commercial broadcaster owned by Virgin Media Ireland) to increase audiences, using smarter content promotion to drive tune-in for shows, for example.

Meanwhile, the Belgian cable operator Telenet has encouraged viewers to accelerate their consumption rate on VOD catalogue content where there is a danger that titles might disappear before they have finished a series. This provides a clear example of how the user experience can be improved. And the Belgian commercial broadcaster SBS (half-owned by Liberty Global) has been working with Telenet to introduce addressable advertising – and has been attracting marketing budget back to TV from ‘digital’.

A number of speakers at Cable Congress outlined progress towards a data-enhanced TV future, offered examples of specific outcomes that you could show to a Chief Finance Officer, and gave practical advice on how you implement a data analytics and insights strategy. They were led by Laurence Miall D’Aout, VP Advanced Advertising at Liberty Global. Her company’s data insights are built upon anonymized and granular viewing reports that can be used to understand consumption patterns and help channel owners understand who is watching shows.

Liberty Global insights can uncover where viewers come from (which other channels, for example) when they tune in to a programme and where they go afterwards – like if they also tune-out half way through. The company can work out the reasons someone stopped watching (like a big rival show that was scheduled half-way through the first). TV3 is one of the first channel owners to benefit (harnessing the local iteration of the insights service, called Virgin Media Solutions).

Miall D’Aout revealed: “We can help TV3 understand if a promotion drives tune-in. We can show a direct impact on the bottom line.” A corporate video explained that TV3 has increased audiences as a result of the insights and actions that stem from them, which in turn “adds significant value to advertising inventory”.

Liberty Global is now offering its data insights capabilities as a value-add to channel owners in carriage negotiations, fighting against demands for higher fees that Miall D’Aout said were the result of financial pressures which, in turn, are caused by audiences shifting from free-to-air broadcasts into the digital domain. The platform operator can help with an insights package that could increase viewership for their free-to-air channels and is trying to do this rather than pay more for the channels.

“We can provide data as a value proposition,” Miall D’Aout emphasised. “As an ecosystem, we need TV to remain relevant and we need broadcasters to make money. One way to achieve that is to help them recoup some of their traditional business and that is where our data could play a role.”

This is an important observation. Channel owners do not necessarily need addressable TV advertising to boost their business with data insights. That could be preceded (or even avoided) with efforts to boost the traditional linear audiences.

Telenet has used the migration of customers to quad-play as an opportunity to get them to create individual IDs for each household member, with their own log-ins and passwords. There are now 700,000 IDs and these can be used to analyse viewing at an individual rather than household level. One of the many use-cases for data analytics is to then work out which content assets deliver the best ROI. This in turn helps you to acquire the right content, and the next step is to market the content better to the people who are most likely to like it.

The ‘VOD-accelerator’ is a great example of what data analytics can achieve for the benefit of consumers, content owners and the platform. “Unfortunately it is a fact that as new content comes in, old content has to go out,” Kim Smets, Director of Marketing Insights at Telenet, acknowledges. “Viewers who were part-way through a VOD or catch-up series realised that suddenly they could not find the title anymore. That sucks. It turns out that we have the data and tools to solve this problem.

“From the viewing data we can identify whether you are following Series Three and we know the end date [when the shows are removed] so we built an algorithm that checks every day how far a customer has progressed and whether they are consuming the content at a speed that will allow them to finish the series in time. If they will not [complete the series] we target them with a one-to-one communication to warn them they only have this many days left, then ask if they would like to watch one of the episodes now.

“We have seen that we can influence the viewing behaviour of the customers we target. They start watching the episodes significantly faster and finish the series.”

Until now, Telenet has been using email alerts to encourage consumers to hurry up. The company is now thinking about using notifications in its TV app, or even one day putting notifications on the TV screen itself.

All speakers discussing data analytics and machine learning at Cable Congress in Dublin emphasised the need to deliver concrete outcomes and not build a ‘data lake’ just for the sake of it. “You are not creating any value until you start solving customer problems,” Smets pointed out. “Our approach is to deliver value on a case-by-case basis.”

Miall D’Aout provided some examples of what a real outcome looks like. “Can I predict which customers are going to pay their bill on time, or which ones will end up in the higher quartile in terms of their revenue value, or who is going to click on my VOD asset?” She said the point of machine learning in television is that you should see patterns that traditionally computers could not spot for you.

Jacques-Edouard Guillemot, SVP Executive Affairs at Kudelski Group (which includes NAGRA and Conax), said a data analytics solution (and his company offers one) should give you a real-time view of the key performance indicators for your business – the most important of which remains usage. He offered some examples of outcomes you should expect, like predicting the value of any given piece of content you want to acquire – knowledge that will help in content negotiations.

“I would like to be able to measure the efficiency of my marketing campaigns so I can improve the next one,” he added. “I do not want that capability limited to one or two campaigns a year, either, but available for every campaign.”

For Pedro Bandeira, Head of Product Development at the Portuguese cable operator NOS, the key output is knowing what customers are using and what they value, partly because this helps you to define your service roadmap. Thomas Helbo, CTO at the Swedish cable triple-play provider, highlighted improved network operations as a key benefit of machine learning at his company.

Com Hem is using advanced data analytics to give the network operations centre (NOC) a better view of what happens in the network and the ability to prioritise resources when improving operations or responding to problems. Michiel Sanson, VP at the strategy and consulting firm Sand Cherry Associates, noted how smart analytics can reduce the number of calls you get to a contact centre.

Focusing on advertising, Liberty Global’s Miall D’Aout said better analytics is already attracting advertising budget back from digital platforms [like Google and Facebook] to television. “They [brands] went to digital because it had scale and data and because it offered attribution and because it is easy to trade. We have the scale and data and we have the potential for attribution. Do we have the ability to trade [using data-driven insights and targeting]? With TV3, our 50% of SBS and Telenet, yes we do.”

She pointed to the availability of addressable advertising on SBS (with Telenet) in Belgium. “In just a few months this has changed how SBS positions itself in the market. They are getting money back that had moved from TV to digital; it is flowing back to TV.”

Miall D’Aout thinks there is a big opportunity for platform operators to help channels to keep monetizing their ad-funded content as viewers watch more catch-up programming. She contended that the first-party data that cable operators have access to is excellent [in terms of its quality and integrity] and compares with or betters anything that the GAFA (Google Apple Facebook Amazon) companies have. She thinks data-enhanced TV will keep money in this industry.

Liberty Global operating companies are customers of their own insights business: Telenet has been using data analytics to improve its own consumer advertising, optimising TV campaigns and integrating them better with digital advertising to drive improved outcomes. Kim Smets noted that the company has segment-specific value propositions that are designed for different kinds of consumers, and which can now be targeted at a household level.

Guillemot pointed out that data analytics has been around a long time, but machine learning and AI allows you to scale operations and use the reasoning to enable mass-personalisation. “The ecosystem becomes smarter, little by little,” he observed. He provided a list of actions for cable operators that want to harness and analyse data more effectively. The first step is to collect the data, of course, and ensure that it is good quality.

Miall D’Aout echoed this point. “I have worked with companies where the data is hardly collected or it is deleted because it is too expensive to keep. You need a rigorous approach to data collection.” She also highlighted the need to unify data. Liberty Global is creating a single data platform, she pointed out, an initiative that is helped by the creation of a single (RDK-based) set-top box platform for Europe (EOS – which is marketed as V6 in the UK).

All customers on the footprint will eventually feed into a single data collection point instead of the 15 data collection points (from 15 different platforms) the company has today. “It is important that in future we have unified data so everything can speak together to really harness value. You cannot change an organisation if you are using data in siloes,” she declared.

Kudselski’s Guillemot offered the advice (to other operators) that all the different data sources must be compatible with one another. The systems and models used by data scientists must also be transferable, and not just the data itself, so that insights can be used across departments.

There are some serious organisational and personnel issues that must be tackled. Guillemot said you need a framework where multiple departments can work together on one specific business question that can be solved by data but also by intuition and ‘art’. If you try to keep your AI and data skills in each department you will not achieve the exchange of information and the critical mass you need to deliver sufficient value-add, he argued.

He said you need to gather data analytics skills into a data centre of excellence covering business analysts, data scientists and data architects (who know how to find the data in systems). “These people are hard to find and it is important to group them and make them work together,” he suggested.

Guillemot added: “Often AI is under the leadership of IT, but it needs a specific mindset to make AI work, so you need IT to adopt a more service-based mode.”

Laurence Miall D’Aout also highlighted the need for centralised data operations. “You must create a group where the data scientists and coders feel motivated and work together.” She said a 20-year-old coder will leave within months if they are isolated in a small unit, separated from the rest of the organisation (or their peers). “They need to feel a part of something bigger even if you are incubating the data division until it becomes a key part of the overall business.”

Liberty Global is creating a centralised unit for its data scientists and they will work collaboratively with the local operating divisions across all markets.

Sanson at Sand Cherry Associates suggested: “As you bring in more data scientists and analytics people, it is important that the rest of the organisation knows more about data, too. They need to know what questions to ask of the data team. We need everyone more involved in analytics.”

Guillemot believes the implementation of a data analytics and insights strategy must be put into the hands of top management and also left there. “There is a tendency for executives to try to push this down the organisation, but we believe it has to stay at the top where you can ensure the arbitration that is needed between the different siloes.” He noted that you should be ready to use external partners, too.

Photo: iStock/PeopleImage

Share on