Personalisation has been one of the big buzzwords in the media industry over the past few years, writes Andy Fry. But what exactly does it mean in the context of the TV business? What is possible, who are the biggest beneficiaries and where are the major pitfalls?
As a starting point, it helps to consider personalisation of content and advertising separately. Looking first at content, personalisation is a way of ensuring individual end users get to see more of the content that is right for them.
As myBBC launch director Andrew Scott explained in a blog: “By finding out more about you and what you like, we can make better content, make it more relevant, and bring it to you more effectively.You can receive recommendations for programmes you might like. You can receive alerts about world events, hot topics and things you’ve told us you’re interested in. You can also get other benefits like starting to watch a programme on one device and picking up where you left off on another.”
Clearly this all benefits the consumer, but it also benefits the company delivering personalisation. According to Scott, data gathered across October to December 2016, showed that “people who were signed in [to BBC services] spent over 20% more time watching, listening and reading content on the BBC’s websites and apps each week than people who were not signed in”.
The implication of this number is that the BBC’s ability to deliver personalised recommendations made its services stickier and enables it to justify its annual £3.7bn licence fee.
Sarah Rose, director of consumer insights at Channel 4, is equally enthusiastic about personalisation. In a recent interview with research firm eConsultancy, she said: “We’ve got five years’ worth of first-party data from 15 million registered users, for whom we have age, demographic, email and postcode, and viewing history.” With “hundreds of millions of data points” at its disposal, Rose said C4 on-demand platform All 4 has reached the point where it can “make recommendations to every user based on individual history”.
The BBC and Channel 4 are (fundamentally) free-to-air broadcasters, but if you translate their data-led approach to personalisation into a PayTV or SVOD context and there are obvious business benefits. Accurate content recommendations might mean more transactional revenue. And, at very least, increased customer satisfaction may mean lower churn.
This explains why the past year has seen several companies announcing advances in this arena. TiVo, for example, released “a fully-integrated personalised content discovery platform” that provides operators with “the most comprehensive suite of content discovery features”. Voice search, for example, brings up “highly-relevant” personalised search results, highlighting the top ten programmes that the viewer is most likely to be interested in, based on their viewing behaviour and interests.
For TiVo, whose mission is to deliver the “Ultimate Discovery Experience”, personalisation lies at the core of its business. Charles Dawes, Senior Director, Marketing at TiVo, says the company’s aim is to “provide its customers and their consumers with the ability to find and consume their favourite content, whenever and wherever they are”.
He adds: “By providing a personalised content discovery experience, service providers can help the consumer navigate through the ocean of content to discover the right piece of content for them considering the time, their location and their device.”
SVOD platform Iflix has been making similar noises, revamping its user interface to deliver a more “individual customer experience”, with users able to select their favourite channels and receive recommendations based on their tastes and viewing history.
Iflix subscribers can also add TV shows, movies and channels to their Iflix homepage, and use the ‘follow’ feature to track playlists curated by hundreds of popular influencers and celebrities. “For the individual user, sifting through immense catalogues of programmes and finding what you want has become more challenging than ever,” says iFlix CEO Mark Britt. “Our new personalisation features now allow us to solve this dilemma by delivering recommendations specifically curated for each local user across our business.”
Orange Group’s Viaccess-Orca (VO) is also driving forward the personalised TV experience in PayTV. In September, leading Israeli PayTV provider Yes selected VO’s Compass content discovery platform to build new levels of personalisation into its SVOD and TVOD services, as well as offering subscribers tailored recommendations.
Through its collaboration with VO, Yes has integrated Compass into a new user interface for its VOD services that automatically recommends content to viewers based on their personal tastes and viewing habits. Additional Compass capabilities, such as Compass Continue, a watching engine, and a Viewer Series engine further enhance the viewing experience. “To keep our competitive advantage, it’s critical that we continually improve our customers’ experience and provide services that are tailored to their preferences,” explained Yes chief marketing officer Ori Gal.
Jeffrey Weber, CEO of ZoneTV, talks about bridging the world of OTT and Pay TV with Media+Networks content lead Niall Hunt.
Content personalisation is not, however, a simple process, says Noriko Matsuoka, who is responsible for emerging tech in the Technology Strategy and Architecture team at the BBC. While its benefits are evident, it does come with a range of creative and commercial challenges. “I don’t think anyone in the industry has come up with a perfect personalisation solution yet for several reasons,” she says.
The first is to do with the complexity of the technical architecture required: “There are so many variables it isn't easy to know what to prioritise. To give you just one example, we have several versions of BBC iPlayer across BT, Sky, Virgin, mobile, smart TV and so on. So how should we respond when new technologies are coming to market all the time? Like many other companies, we are constantly in discovery mode.”
The rapid emergence of voice as a mainstream industry standard is one of many developments that adds to the challenges around personalisation. Deciding which voice systems are going to emerge as commercially viable is one consideration, she says. The issue of accents and dialects is another (here’s an amusing video on this subject). “A public service broadcaster like the BBC has to deal with scenarios such as when a device like Echo or Google Home comes with a default setting that may not include the BBC, hence we have to negotiate to make sure the license fee payer is getting what they have already paid for,” says Matsuoka.
Alongside questions of technological specification, says Matsuoka, there is a massive issue around how to build a platform that is accurate and empathetic enough at an individual level to make meaningful recommendations. She adds: “We have data coming out of our ears. But how can you judge an individual’s mood well enough at any given moment to know what they want to watch? And how do you know that their ‘funny’ is the same as yours? Netflix spends a fortune on recommendation but most don’t have those resources.”
TiVo’s Dawes says: “when a user asks, “What’s on now?”, they expect the results to be unique to them based on their previous viewing habits, even though they haven’t specified ‘for me’.
There’s also a commercial contradiction that hampers personalisation. Expensive content like drama, films and sports requires large audiences to generate ROI, so it’s inevitable that platforms will want to market them regardless of whether they are a true reflection of an individual’s stated interests.
For example, PayTV broadcaster Sky’s Sky Q platform has a ‘personalised’ section called My Q that sits next to its general TV Guide. However, the only truly personalised element of this is a section that remembers what you have been watching and allows you to dive back in where you left off. The rest of the My Q interface is the promotion of new movies and series that don’t necessarily have a connection with the end user (an example would be big budget series Britannia).
Dawes says: “Comparing recommendations to personalisation is not like comparing apples to apples. While recommendations do not need to be personalised, TiVo believes they need to be.”
He explains that TiVo’s approach uses two types of recommendations; predictive and suggestive. Dawes adds: “Predictive recommendations give consumers the best programme option for right now – based on factors including time and previous behaviour – while suggestive recommendations give more expansive content choices based on consumer viewing habits. When implemented correctly, operators can also leverage user behaviour to deliver targeted promotions, turning a content discovery system from a cost to a revenue generator.”
Speaking to The Drum ahead of CES last year, Campbell Foster, director of product marketing, Adobe Primetime, also raised this point: “If you look at HBO, none of the recommendations are personalised,” he observed. “They are recommending what they want you to watch not what you want to watch. They just kind of swing products that they think will boost their margins… They have a lot of programmes in their back-catalogue that have dedicated audiences that people just don't know about because they are pushing Game Of Thrones so hard.”
There is a fundamental point at play here – which is that media companies need to decide what level of personalisation is optimal. A user’s viewing behaviour may indicate they have a soft spot for 1960s Korean romances, but that doesn’t necessarily equate to a meaningful ROI opportunity for the platform they subscribe to.
For Dawes and TiVo the answer is simple: “users who interact with a personalised experience consistently consume 30 percent more content than those who are using a non-personalised system [TiVo White Paper: Success Story:
Personalised Content Discovery Drives Higher Content Consumption for a Leading Pay-TV Provider], which more than justifies the investment for the operator. Better promotion of paid content drives incremental revenue the operator would not have captured without personalisation.”
Data, data, everywhere
On top of this is the question of data capture, says Matsuoka. “People might like the idea of personalisation but they are also anxious about handing over data because they don’t know how it will be used or whether it will be sold.”This isn’t helped by the fact that most consumers have already been exposed to a wide array of inappropriate or irrelevant “personalised” commercial message online (pop up ads that try to sell us something we’ve just bought for example). “There’s also a lot of scaremongering in the media about devices listening to us,” adds Matsuoka, “which inevitably makes people a bit more hesitant about voice for example”. All of this is being played out against a backdrop of greater regulatory scrutiny, “which is a huge problem for the BBC as we protect the license fee payers data so that it can’t be sold by third parties”.
Having said all this, Adobe has products that it claims are moving the agenda forward. Its video personalisation engine, Adobe Primetime Recommendations, “harnesses big Data to increase viewer engagement and viewing time”.
In the US, “Primetime Recommendations learns from 200 billion online video consumption points across most US households that stream TV and film content. Deep contextual insights, such as how much and what content a viewer has watched, yield personalised TV recommendations.” To alert viewers to recommended content, Adobe Campaign can also orchestrate email campaigns and notifications based on preferences,” Foster said.
In a report sponsored by Adobe, research firm IDC claimed that the integration of Adobe Primetime could generate a 5-year ROI of 385%. Based on surveys with programmers that use Adobe Primetime to support their delivery of primetime video content, IDC attributed an increase in viewer numbers, viewer engagement, subscription revenues and advertising to the tech.
The overall message, then, is that personalisation of content is probably a good thing, but that there are many potential mis-steps when you try to second guess what a consumer wants.
As a starting point, companies should seek to integrate simple, non-intrusive features (such as basic recommendation or the ability to pick up content easily on any device, at any time). They should also ask whether personalisation is really the best place to spend their budget. After all, a big driver of TV continues to be the social appeal of shared experience.
Exploring this theme further, Anthony Smith-Chaigneau, senior product marketing director, NAGRA, used his blog to suggest that too much personalisation might lead to monotony.
“Algorithms have their limitations. Just because a viewer likes a show or a genre doesn’t necessarily mean they want to watch it over and over. In fact, in many cases the machine will need to be told to stop doing that. TV and film have always relied on variety, which is what we get today (even if we don’t like all of it). By relying on algorithms to bring us content, it’s possible there will be something out there we never come across.”
According to Smith-Chaigneau, “there are instances where viewers don’t rely on recommendations and look for content that doesn’t necessarily align with their viewing history (this could be because of mood, time of day, the company they’re keeping or any other number of influences). There needs to be an element of chance and discovery in selection. To address this and offer better recommendations, NAGRA is developing solutions that incorporate all the techniques available from algorithms to new fun and interactive solutions for discovery.”
One possible area worth exploring, says Matsuoka, is whether it is possible to shift the industry’s mindset with regard to data capture. “Often it’s a one-sided communication where media companies are selling to people but not listening. You might get closer to true personalisation if you ask people about how they structure their day and what they want when. Of course, that raises other issues – for example people often say and do different things. And you need to avoid a situation where you constantly question them. Maybe we could ask people how they felt right at the end of their viewing experience.”
The point that Matsuoka is driving at here is that there is also a role in the personalisation debate for empathetic personalised messaging. While most people detest being spammed there is an opportunity to use voice during viewing downtime to ask questions and establish preferences.
Looking at personalised advertising, there’s no question that some platforms are experiencing the benefits of developments in this direction. Sky Media, for example, recently reported that it had broadcast the 10,000th ad campaign using Sky AdSmart, a proprietary tech that gives advertisers the ability to target TV ad campaigns to specific audiences.
According to Sky Media: “Every two days a brand that has lapsed or never advertised on TV before selects Sky AdSmart to reach their target audience in a trusted, brand-safe environment. The technology has 1400 targeting attributes up from 400 at launch, allowing niche and local businesses as well as mainstream brands to benefit. Sky AdSmart advertisers have a 64% return rate, indicating that addressable TV is working for businesses and successfully expanding the TV ad market in the UK. This is expected to continue as a third of audiovisual advertising in the UK will be addressable by 2022.”
The 10,000th campaign was for banking group CYBG’s new mobile banking app, B. The client up-weighted the execution using regional targeting to serve ads to households in Birmingham and London with young professionals, high income families with children and middle-aged families. Helen Page, Group Innovation and Marketing Director at CYBG, says: “AdSmart makes TV trackable and measurable and we have seen real benefits from the start. We’ve used AdSmart to support the promotion of our digital banking app B and it’s helped us to target specific audience segments while providing the insight needed to follow through to conversion.”
The general view is that this kind of addressable TV advertising is going to grow. US research group eMarketer, for example, expects US addressable TV ad spend “to reach USD$3.04 billion in 2019, more than double its 2017 level of $1.26 billion”.
Ovum, meanwhile, reports that 52% of consumers polled like it when services are personalised to them and their interests (though that does raises an obvious question about the other 48%).
There are some challenges however. In a March 2017 survey of 150 US addressable TV decision-makers funded by AT&T AdWorks and conducted by Advertiser Perceptions, cost was named by nearly half of respondents as an obstacle to greater investment in addressable TV. Two in five advertisers surveyed also named limited scale or reach as an obstacle.
Furthermore, while this kind of tech is of benefit to advertisers and platform owners, it's not clear whether ‘personalisation’ is really the right term for it. Echoing the point made above about content promotion, this is not an example of media owners responding to a consumer’s moods, needs or budgets – it is more about reduced wastage. While addressable advertising may eradicate wholly irrelevant ads, it won’t reflect where an individual is in the buying cycle unless it asks them.
A further risk with addressable advertising it is that tighter targeting may make media owners feel entitled to increase the ad load in areas they previously avoided. So consumers will end up with ‘quite relevant’ ads in places where previously there were none (eg around on-demand content). The ultimate impact of that will be to make services feel more impersonal.