DTVE
Universal discovery has long been an
industry goal, but will the scale and complexity of content always keep a
global solution out of reach?
Consumers can’t
be expected to know everything that’s available to them in the vast pool of
content in the world. They need help and the service that does that best will
be well on its way to success. Since discovery comes in many forms, from
accidental to recommended, and a million things in-between, the next frontier
for broadcasters and operators inUser this area is about making sure consumers can
view what they want irrespective of where that content is coming from.
Unified or universal search and
discovery has been an industry aim for a number of years but the need for it
has increased recently as more and more of us view content from on-demand and
streaming sources alongside and often in preference to linear broadcasting.
“Being able to do this in a simple and holistic
way really solidifies the need for new technologies such as conversational
voice and personalised recommendation systems,” says Tivo’s senior
director, international marketing, Charles Dawes. Unified discovery is
deemed critical if the industry is going to keep on delivering the content that
consumers desire. Dawes says: “Consumers don’t know, and shouldn’t need to
know, where the source of their content is. They don’t need to understand terms
like linear, DVB, IP, broadcast and OTT when all they want is to enjoy their
favourite video on demand.”
Universal search and discovery breaks
down the walls between TV providers making content more easily findable and
watchable. Laurent Van Tornhout, vice-president, product and marketing at
Zappware says the goal is “full integration in one UI experience and
multisource – aka universal – search.”
Such a goal moves beyond basic search.
“Going beyond basic genre and delving into dimensions of content such as mood,
theme, setting and character attributes [means] a new generation of navigation
and recommendation offerings will suggest the TV shows and movies most likely
to resonate with individual viewers,” says Simon Adams, chief product officer
at Gracenote.
Further, cross-media content discovery
will enable viewers to find and watch music videos on TV or identify musicians
or athletes in movies. Gracenote’s datasets and Gracenote IDs, which enable
universal search on platforms like Apple TV, Roku and Samsung could be key to
this.
In parallel, short-form and user-generated
content is on the rise and equally relevant as content choices. “Even long-form
content is being chewed up and spat out as bite-sized chunks for the more
‘attention challenged’ viewers,” observes Anthony Smith-Chaigneau, senior
director, product marketing, Nagra. “Shows and movies have endless amounts of
related video content that are accessible to fans. Looking at just the
explosion of gaming and the massive online following of Twitch users, Instagram
influencers and Facebook video content that is competing for one’s time, you
realise that content discovery solutions of the future will have to evolve to
not only provide access to alternative content but may be driven by it.”
Value of universal search
Universal search should allow the viewer
to fully understand where a particular piece of content is, how much it costs
and whether they have access to it with their existing setup. Yet the current
trajectory of content silos “make global search and discovery difficult”, says
Smith-Chaigneau. “Having a source- and service-agnostic way to remove those
ambiguous groupings addresses the real pain point that needs to be addressed.
There is no Google for TV in the mainstream although many are trying to be
this.”
If there were, the value of intuitive
navigation and universal discovery, which is already “sky high” says Adams,
“would go even higher”.
“User-centricity would bring added value
for the telecom operator as the content aggregator for all sources of content,”
agrees Van Tornhout. “For the consumer, the value is that, done right, their
experience is simpler and seamless – meaning they spend more time enjoying
content rather than searching for it.”
Unified search and discovery could bring
the operator “a wealth of new opportunities”, says Dawes, that are built around
a foundation of having a connected, engaged consumer who doesn’t need to go
‘off platform’ for their content. Dawes continues: “Operators have long known
that giving access to all the content the consumers desire via their remote –
aka their ‘brand in the hand’ – is key for their business and helping consumers
understand the value they bring.”
Simon Leadlay, director, pay TV business
development at You.i TV highlights the relative poverty of experience for most
broadcast sports fans. “Broadcast TV still scores highly among sports viewers.
Yet the discovery UI found in most pay TV services doesn’t cater well to that
market,” he says. “I would expect a huge potential for subscriber stickiness
and satisfaction for an operator to combine traditional sports video
broadcasting with a plethora of on-demand and OTT sports entertainment news,
stats, team information, and replays of past games and special events. Sports
fans are finding some of this rich experience via direct-to-consumer
platforms by their favourite leagues, but very few have a similar
offering available from their operator. There’s a lot of room for improvement
here.”
According to Leadlay, the best kind of
discovery systems should be quietly knowledgeable, unobtrusive, and yet always
recommending something. “At the same time, I am personally worried about the
risk of over-recommendation: where I’m never offered anything new, fresh, or
out of my comfort zone,” he says. “I think it’s important to strike the right
balance between seeing familiar and new content.”
Leadley suggests that the next frontier
is likely to see “increased commoditisation of what advanced recommendation and
discovery systems are already using”, namely more machine learning to drive
truly individual recommendations based on time of day, location, device, and
previous activity.
If it’s agreed that the system would
need to know who the viewer is and understand the individual’s likes and
dislikes, then ‘discovery applications’ would have to be redesigned.
Smith-Chaigneau says: “All of the streaming video services are currently
obviously inadequate in coping with the sheer volume of content available.
Everything seems to be a Netflix clone. The technologies driving the content
like recommendation engines will also have to be redesigned to be more sophisticated.
Even how we ‘think of content’ will have to evolve for us to come up with the
right solution for the future TV watcher.”
Recommendations made to an individual
could be increased in relevance by adding profiling input from other online
sources. This can happen “provided that the back-office of different online
systems can feed each other technically as well as from an opt-in point of view
by the end-user,” points out Van Tornhout. “In terms of personalisation you’d
need to bring the most relevant content and content suggestions based on
profiling, context and device used at that moment by that user. The service
would also need to aggregate different content sources to be able to bring the
‘fat tail’ as well as the ‘long-tail.’”
Vendors began embracing the need for
universal discovery across the consumer electronics and operator spaces a while
ago. Tivo, for example, lays claim to being the first user experience provider
to integrate Netflix into an operator when it helped Virgin Media launch the streamer
in 2013.
“For operators there was a need to bring
unified discovery to their platforms even before the current trend of adding in
multiple video content sources,” says Dawes. “Tivo now integrates multiple
video sources into a single experience that is led by the content rather than
the service the content comes from. When a consumer searches for their
favourite show they are presented with the show first and where to get it from
second.”
Gracenote’s global video data offering
is built on a unified data schema “that provides a consistent and standardised
format for all international data,” says Adams. The Nielsen-owned firm is also
developing the next generation of descriptive metadata and images in its
Advanced Discovery suite. “These solutions help providers deploy all-new
personalised recommendation capabilities, visually rich guides and even
voice-driven experiences,” he explains.
AI and metadata building blocks
The second phase of unified discovery
will make greater use of metadata powered by machine learning tools. “Metadata
is absolutely critical when you are delivering a compelling discovery
experience,” Dawes says. “It is no longer good enough to have the simple time,
title and channel that some EPGs deliver. Today’s discovery systems, especially
those which are voice driven, need to have a high-quality baseline of
information about each piece of content.”
Tivo refer to this as the ‘Metadata
iceberg’ where a small amount of visible data is underpinned by a huge amount
of non-visible information. Sitting on top of the metadata is a personalised
recommendation platform. This goes above and beyond a simple rules-based
recommendation system into a system that can be personalised based on time,
day, context and an innate understanding of how individuals personally consume
content.
“Access to a wider range of metadata
that allows users to pivot from content to content is essential to providing a
rich expansive experience,” says Nagra’s Smith-Chaigneau. “However, if you look
at Netflix, it puts tremendous effort into creating its own metadata and
manipulating it accordingly. For example, it has shown that different users are
drawn to different types of movie posters. Users on the Netflix platform see
different movie posters based on their previous viewing habits, whether they
have binge watched, and other parameters discovered by the system.
There is also the concept of dynamic
metadata which goes beyond static images of posters and actors. But the
constantly changing social commentary and related media content is also
becoming an important facet to the primary video. There will also be frame and
scene content tagging for age appropriate viewing which all highlights the
importance of metadata.”
Enabling anything from poster quality,
size and colour to clear and well-crafted information that ensures any and all
platforms can present it for viewers to effortlessly discover and consume is an
increasingly complex task but metadata is foundational.
“As the world of content merges there
are now exceptional foreign titles found in the mix so there are language
issues to overcome, translation, subtitling and a whole swathe of cultural
aspects that also need to be taken into consideration,” observes
Smith-Chaigneau.
Artificial intelligence and machine
learning are increasingly playing a greater and greater role in automatically
building out content metadata, be it from assessing and assigning a confidence
level when information is initially imported, through to keeping tabs on how
the importance of relationships between different pieces of content changes
over time.
“AI-assisted metadata will soon become
the norm,” says Smith-Chaigneau. “AI can afford more depth to metadata services
where it could, for example, compare the synopses of two titles and identify
differences that could then aid or automate editorial management of those
pieces of content.”
With more and more user data points
available and with much more sophisticated methods of interrogating it, “we
have to evolve past simple usage-based recommendations toward
hyper-personalisation,” Smith-Chaigneau says. He suggests leveraging social
indicators and ML to better understand what drives users to content “especially
as the experience crosses over into more and more personal devices, with unique
behaviours.”
On top of all this, there’s a clear
trend toward a voice-based discovery model. The sheer scale of available
content renders the tools of the linear world, such as a grid or tree-based
content catalogues, insufficient going forward.
“Voice input frees you from the
limitations of single character entry and allows you to discover content in the
way you think about content,” remarks Dawes. “The system should be able to
respond to simply saying ‘What’s on TV tonight?’ and put together complex
queries like ‘What’s the film with Peggy from EastEnders and Kenneth
Williams?’. It should ‘know’ you implied ‘for me’ and give you a set of
personalised results from across your entertainment choices that isn’t limited
by linear TV.
The user interface, of course, shouldn’t
be the destination for the user; that should be the video content. However, as
Leadlay underlines, a great user interface can make the process of finding and
watching that content truly satisfying, “and we all know that a poor user
interface leads to frustration and failure.”
Ideally, the simpler the UI, the better.
It should be intuitive, get you to where you want to go in the least number of
clicks and to all intents and purpose be transparent to the user. Arguably
no-one has yet alighted on the absolutely right design mix.
UI as entertainment brand
“It is sometimes hard to break the mould
of consumers’ TV UI habits – removing the EPG proved to be wrong for some
platforms who soon added it back to the mix,” says Smith-Chaigneau. “Not having
channel numbers also proved to be a no-no for one European platform which had
to rapidly add them back in because a lot of viewers had become used to that
form of navigating. We have tried 3D UIs and other fancy tricks but, just like
PowerPoint, all the fancy fonts and animations simply interfere with the
objective of getting the message across.”
The industry has experimented with many
different discovery mechanisms from personalised to promotional, curated,
edited, serendipitous, trending and logical recommendations. It has
incorporated cognitive search and voice options and now it is incorporating AI
in the background.
“We blend all of these together and try
to create a simple and refined way of presenting these in visual form so the
consumer feels they are being well served. From a product design standpoint,
the necessity to have a connected platform and flexibility across platforms is
the most interesting tool. We have to think of our products holistically and as
complementary parts of an ecosystem. We have to leverage more distributed tools
and standardised development libraries,” says Smith-Chaigneau.
In general, it’s difficult for the UI or
UX to be the entertainment brand by itself. The experience is reliant on the
content and usually it is those content brands which are seen as the
entertainment brands by consumers.
However, you could argue, as Dawes does,
that ‘TiVo Experience 4’ is one of the few entertainment experience brands that
exists in its own right. “As an established consumer platform in the US,
with a loyal customer base, TiVo has won multiple awards and been featured in
popular culture on many occasions. Consumers choose TiVo for its extensive
functionality, continued innovation and ability to get your key entertainment
choices in one place, be they linear or streaming services like Netflix,” he says.
Zappware’s Van Tornhout agrees that if
the ergonomics are right “and if relevant enough for the end-user as a gateway
to discover all the relevant content across content sources” then the UI could
become an entertainment brand. “The UI should be grandma-proof as well as
millennial-proof.”
It’s arguable too that some of the
latest OTT services view the UI as an integral part of their package and
branding. Disney will surely look to create such a UI as its brand is already
synonymous with entertainment, as opposed to a service provider that delivers
multiple entertainment brands in a single UI.
“Just as Google became the brand of
internet search, there is definitely an opportunity to be the face of content
discovery,” says Smith-Chaigneau. “There is a magical blend of form and
function that, when applied to obvious pain points, can become ubiquitous.”
In focus: Curated or algorithim-based
discovery?
While the sophistication of algorithms
will only get better, editorial or curated discovery will retain a role in the
creation of a ‘more human-led’ approach to any service.
“One of the drawbacks of early
recommendation systems that just relied on techniques such as collaborative
filtering was that they would drive you further and further down an
ever-narrowing set of content choices,” says Tivo’s senior director of
international marketing, Charles Dawes. “There was no facility for companies to
promote content that was important to them commercially or for the system to
understand the difference between a customer needing to watch something they
like right now or them looking to broaden their content horizons.” TiVo says
its Personalised Content Discovery platform has the ability to blend not only
editorial and algorithmically-based recommendations and predictions but also
include new recommendation paradigms such as sponsored discovery – where paid
placements are served to the right, receptive consumer. “This will deliver a
superior experience that enables more video viewing than ever before,” Dawes
claims.
It is the vast amount of legacy content
in addition to new content that presents the real challenge of scale to manual
processes. “By leveraging both human editors and ML, we’re able to create
datasets with the breadth to cover existing and new content and the depth to
describe that content with increasing specificity,” says Simon Adams, chief
product officer at Gracenote. “This is critical to positive user experiences.”
“We know that there’s a need to go
beyond editorial and algorithmic recommendations, using AI to better predict
not just the type of content, but getting to why we ultimately watch what we
do,” says Anthony Smith-Chaigneau, senior director, product marketing, Nagra.
There will be a blend and balance of the
two recommendation forms. For example, says Laurent Van Tornhout, VP, product
and marketing, Zappware the first three items could be editorial, related to a
trending topic, and the rest of the suggested items on the ‘bookshelf’ could be
algorithm-based.
Nagra’s Smith-Chaigneau says: “I will
never forget talking to an Uber driver about his TV consumption and he said
that after about six months of using a particular OTT service he went home
after a shift and realised how boring and same-same his TV viewing had become.
The machine had led him down a silo of similarity and he was bored with what
was on offer to him. He wanted a reset-button so that he could flush out the
system and start all over.”
