38
BROADCAST TECHNOLOGY 2018
“Our hope is that, as an industry, we are
able to move away from simply offering
what’s worked elsewhere to offering
more nimble, tailored models that are
more befitting of the myriad needs and
wants of the 1.8 billion viewers in Asia.”
— Peter Bithos
CEO, HOOQ, & an APB Panellist
Asia demands
more than just
‘on-demand’
BY PETER BITHOS
T
he explosive growth of Asia’s
middle class and rapid increases in
Internet penetration offer significant
opportunities for over-the-top (OTT)
service providers, but these factors alone
will not drive growth in the industry.
Media consumers in Asia need OTT
services that are as diverse as the region
itself, and this means catering to a wide
range of socio-economic, cultural and
linguistic needs.
“Build it and they will come” may
have worked for a fictional baseball
field, but it is probably not going to
cut it for OTT. In an age where media
consumers have more choice and control
than ever over what is in front of their
eyeballs, decision makers at these kinds
of companies (full disclosure: I am one
of these decision makers) need to reckon
with the fact that our critical competitor
is not really FTA TV — it is the Internet.
Where previously Internet access was
governed entirely by who had access to
a computer and the requisite networking
facilities, today, consumers from virtually
every socio-economic bracket in Asia are
going online using their smartphones.
Now consider this: in Indonesia, the
minimum daily wage is under US$10.
In Thailand, it’s about $15. You get the
picture. To afford some of the subscription
and transactional video-on-demand
services out there, a substantial number of
users in Asia would have to pay roughly
a day’s worth of wages in order to gain
access to content.
Many viewers in these markets simply
are not looking to spend weeknights
bingeing their favourite shows. Often,
they simply want access for a day here
and there, for example, when they transit
from their hometown to the city they work
in or on public holidays.
For us at HOOQ, it was with these
kinds of situations in mind that we
introduced nearly all of the FTA stations in
Indonesia on one’s mobile phone for free,
all the time, or, for premium content, one-
day ‘sachets’, which for US$0.25 per day,
provide a 24-hour block of time where
customers have access to our full range of
on-demand premium content and pay-TV
channels. No longer is premium content
restricted to those few who can afford
(or have the payment mechanisms for) a
monthly subscription. This dramatically
An
Supplement
changes the entertainment landscape for
90% of Indonesians (instead of just the
10%!) who can now watch what they want
(even premium local and Hollywood
content), where they want and at a price
that is affordable.
We conceptualised this daily pricing
approach in response to the consumption
patterns we see in South-east Asian
countries, where consumers purchase
household essentials such as shampoo
and detergent in single-use sachets. At
the same time, we saw similar mobile
data consumption patterns in South-east
Asia’s pre-paid mobile market, where 80%
of the population prefer to top up credit
on a daily or weekly basis, rather than
subscribe to a post-paid mobile plan.
But it is not just economic factors that
need to be taken into consideration for the
OTT market here. The other area where
we believe OTT providers need to be more
flexible is in the content they deliver.
It goes without saying that there is a
market for Hollywood blockbusters just
about everywhere, but we need to ask
ourselves: are we delivering these big-
name films in ways that take into account
not only regional diversity, but also intra-
market diversity? Recently at HOOQ, we
have been focusing on not just localising
content, but “ultra-localising” content,
which means doing things like developing
Kannada- or Telugu-dubbed versions of
the Hollywood movies we distribute in
India (languages that have 44 million and
81 million speakers respectively — not
exactly a niche offering!).
Outside of localising Hollywood
content, more support for local films and
shows is needed to better tell the stories
of the peoples and cultures in the region.
Sure, The Big Bang Theory is undoubtedly a
hit show, but we believe it is important to
deliver popular series like this alongside
content that speaks directly to audiences’
day-to-day experiences and interests. An
Indonesian horror series, anyone?
It is an exciting time to be growing
the OTT industry in Asia, and it is great
to see consumers here being given more
options in the genres of entertainment
they consume. Our hope is that, as an
industry, we are able to move away from
simply offering what’s worked elsewhere
to offering more nimble, tailored models
that are more befitting of the myriad
needs and wants of the 1.8 billion viewers
in Asia.
Mixology and the
content rabbit hole
BY SHAD HASHMI
I
was in a trendy, new-age mixology
bar. There were no menus, no rigidly
defined drinks and your imagination
was the limit to the experience.
You told the waiter what spirit you
fancied, what tastes you liked and then the
producer behind the bar concocted some
madness. I specified whisky and chocolate
and was duly presented my custom
lubricant: a variant on an Old Fashioned
— with German chocolate, instead of the
predictable, Angostura bitters.
It tasted great — familiar yet
different. Old yet new. The perfect gift —
unexpectedly getting exactly what you
wanted, without knowing you wanted it.
While I sipped on my ‘new’ Old-
Fashioned, I thought about TV. In the
world of linear TV, serendipitous discovery
was the norm: a function of the platform.
You used to stay tuned, not turn the
dial, on a channel. After the programme
you had tuned into ended, you were
automatically exposed to something new.
In a branded channels world, you got that
different but still the same feeling.
We also got the same ‘box of chocolates’
sensation when we flipped across channels.
You never knew what you were going
to get. Yes, I concede, it was not always
good and sometimes you just kept flippin’
but, on hindsight, I have mostly positive
memories about these experiences.
Back to the present: I log onto a
subscription video-on-demand (SvoD)
platform and I feel like I am holding onto
a freezing-cold metallic drink shaker as the
icy, rational logic of artificial intelligence
(AI) via a recommendation engine curates
my view.
Every recommendation I am presented
with is largely a result of some shaken but
not stirred combination of two algorithms.
The base ingredients (the algorithms)
are ‘Content-Based Filtering’ and
‘Collaborative Filtering’. I will focus mostly
on Collaborative Filtering as Content-
Based Filtering uses metadata about the
content to recommend similar titles.
Collaborative Filtering looks to build
a picture of you — and it wants to get to
know you intimately. Unfortunately, our
preferences are not as simple as whisky,
chocolate, gin and juice. People are more
nuanced and collaborative filtering takes
time to build a picture of us.
The machine-learning algorithm
scrapes through our activities to scrounge
for information on the shows we have
watched. Did we finish the episode? Did
we abandon the series? It then attempts to
be even more explicit asking us for ratings,
a thumbs-up or down, and may even ask a
series of questions to refine its view of us.
For example, when we are asked to rate a
show on a video-on-demand (VoD) service, it
is not so that other subscribers get a number
to determine the quality of the programme
but the rating is a direct representation of
what we like.
As systems get more advanced the
recommendation engine also looks at
implicit signals beyond what we watched.
To build a picture of us, the engine looks at
implicit signals like what we are browsing,
whether we stopped and read a synopsis
and what items we scrolled to.
All this learning, questioning and
analysing takes time but, in the end, the AI
knows you and your tastes as you are now.
Extrapolating the algorithm, it also knows
similar information about everybody else
on the platform. The AI then makes the
following core assumptions: (1) I like what
people like me, like; and (2) the past predicts
the future.
Then, the AI bartender comes in and you
are presented with a collaborative filtering
curated world that can be described in these
phrases visible on the user interfaces of most
SVoD platforms:
People who watched ‘Show X’ also watched …
People like you watched …
Or, to be specific: “Our algorithms predict,
with 90% (say) accuracy, that people who like
James Bond also like Mission Impossible.”
Over time, these are self-reinforcing or
self-perpetuating as we are shown (and
hence watch) more and more content that
people like us like, and as past behaviour
is used to predict future behaviour we end
up in this zone where we are never shown
anything really different, or new, from what
we have seen before. The platform closes our
window of new experiences.
A recommendation engine can take you
down a content rabbit hole as each action
re-enforces your existing taste profile and
embeds you even deeper within your taste
cohorts. It is thus a self-fulfilling prophecy
you cannot break out of. You are who you
are tightly defined to be and your tastes
never mature or change. You are who you
are once, for all, forever.
That is a dire, depressing and sad reality.
It almost eliminates the human characteristic
to learn, change and grow. It instinctively
seems wrong and does not really work with
something that is supposed to be uplifting,
fun and imaginative like content.
Our world is changing and we may
not be ‘linear’ for much longer and our
futures are inextricably tied to AI, pattern
matching and the rise of machine learning
but we simply must find a way to mix in the
human. We have to rediscover that human
element that adds variety, spice, variance
and randomness, and that magic mixology
of ingredients that makes content and its
discovery so special.
“As systems get more advanced the
recommendation engine also looks at
implicit signals beyond what we watched.”
— Shad Hashmi
SVP, Digital Development Global Markets, BBC Studios
& an APB Panellist