The ability of Data Science to offer unique insights into data is not restricted to traditional IT organisations anymore. With the promise for increasing the efficiency of bottom line operations, analytics has seen a quick expansion and adoption scale in retail, manufacturing, logistics and media sectors.
Service providers across verticals have been baking machine learning in order to gain insights into their processes. Moreover, they also enable finding new knowledge for many different parts of the customer sales funnels.
To know more about how a big media company uses data science, Analytics India Magazine reached out to Viacom18’s Sidharth Kedia, Head – Strategy and Data Sciences.
Sign up for your weekly dose of what's up in emerging technology.
AIM: How has been your analytics journey so far?
Sidharth Kedia: My analytics journey started in 2000 when dashboards were being digitised by GE and I was fortunate enough to have been part of those initial seeds being sown for what would become one of the biggest and fastest growing industries today and in times to come.
Thereafter, I setup the analytics practice for a BPO, whereby the buzzwords were SPSS, Business Objects, Object Oriented Programming were the buzzwords and most of the front end was developed using .NET. Fast forward to two decades later and now the same world is called Data Science.
Data being generated is of a magnitude difference and hence Big Data. Business decisions have become so complex that it is humanly impossible to model outcomes and hence, the need for machine learning. Even new programming languages (Python, R) and front-end visualisation tools (Tableau and Power BI) have been developed.
The computing power has shifted from local machines to the cloud, giving a fillip to newer business models for AWS, Microsoft and Google. The journey has been interesting, to say the least. However, this is really Analytics 3.0, the way I look at IoT and this is the inception point where the hockey stick will kick in, in terms of the importance of data backed decision making.
AIM: Having worked with startup and now Viacom18, what are the differences you see when it comes to analytics adoption or technical maturity?
SK: The difference between start-ups and a large company like Viacom18 is best described under four distinct categories:
- Resource Availability: Building an analytics function puts a considerable financial strain on an organisation as it entails massive investment both in terms of people hiring and infrastructure. A large company can readily deploy resources, once this has been identified as a crucial need for the company. The same level of capital allocation is usually harder for start-ups and hence the ramp-up of analytics tools and models is much slower.
- Resource Quality: An analytics organisation is as good as the people it has been staffed with. A good analytics team comprises of Engineers, Economists, MBAs and even PHDs. Better the institute that the talent comes from, the more the struggle for a start-up to attract talent. There is enough empirical evidence that a premier school recruit (more often) would choose Viacom18 over a start-up. A large and stable employer brand is what primarily lends to this.
- Agility: This is the key parameter whereby start-ups leave behind large organisations. Organisational structures and hierarchies make it harder to manoeuvre any new initiative in a large organisation as too much time is spent on democratising the idea and in consensus building. As against this, a small organisation is more nimble and agile and it is much easier to change direction or incubate a disruptive idea in the organisation as the primary decision maker is one or a few people at best.
- Ecosystem Access: As a large and well-established organisation, the brand acts as a calling card and access to the ecosystem players, large and small; local and global becomes that much easier. In an era of increasing collaboration and dependency, wherein greatness can’t be achieved by any singular organisation, large firms like Viacom18 score over start-ups merely by the access and reach their brand can get them.
AIM: What are the various analytics solutions that you have developed or worked on?
SK: Analytics solutions that Viacom18 have developed are broadly under two buckets, namely Television and Digital.
In TV, the models largely focus on optimising both viewership ratings and revenue. In digital, the primary objectives of the business are around customer acquisition and retention. All AI/ ML models are built aligned with these business objectives.
AIM: How did your tryst with media and entertainment begin? How do technologies such as analytics and AI driving innovations or streamlining processes in these areas?
SK: Like any tryst, mine began more as serendipity than by design. In 2007, I co-founded a Private Equity fund, focused on growth capital. While the focus initially was on TMT (Tech-Media-Telecom), it soon became focused on pure play Media & Entertainment as an industry, a practice I built for the fund and was associated with evaluating and investing in media companies for close to a decade, before moving full time into a media company (Network 18 group).
The world of media, as we know it, is undergoing a significant transformation. Like most other businesses, new age media companies have shown that value explosion is unlocked by owning the consumer and knowing his/her media and content preferences at a micro level. This can be achieved only with tools like analytics and big data.
Also, with not only the quantum of data but the extent of unstructured data that is being generated, a meaningful analysis of the same can be done primarily with the help of AI and other tools like NLP. Analytics is increasingly going to be the key differentiator aiding companies in fields as wide as revenue augmentation to customer satisfaction enhancement to cost reduction and in some cases even identifying new business opportunities.
AIM: How are you adopting analytics at Viacom18? Please give some use cases where analytics has benefitted the organisation?
SK: Analytics is being adopted in both digital and broadcast universe at Viacom18. Here are two illustrations in both these universes that are currently being treated distinctly:
Broadcast: Traditional broadcast media’s business model is primarily B2B, whereby the lion’s share of the revenue comes from advertising. Advertisers for most mass channels are looking at the viewership that a 10-second slot delivers and their aim is to maximize their ROI by optimising reach and frequency.
The slot viewership, in turn, depends on two factors, namely – Program viewership and retention of viewers at the time of the break in between a program segment. A break model has been built at Viacom18 that maximizes retention during a break, effectively increasing per slot and overall revenue for the company.
Digital: Voot, the digital offering from Viacom18 is one of India’s largest avods (Advertiser funder Video on Demand) platforms. We have built a machine learning model to look at content affinity amongst other factors and predict age and gender. This model over time has reached a very high level of accuracy and has resulted in significant revenue upside, as a result.
AIM: How are you adopting a data-led approach for content creation, segmentation and targeting? How has it proven to be beneficial in these spaces?
SK: Consumer segmentation and targeting are at the very heart of what any digital OTT platform does and we at Viacom18 (Voot) are constantly building better and sharper models for the same. Our digital recommendation engine is being built on the principle of segmented targeting.
In TV, we constantly analyse consumer preferences and thereafter use a mix of the traditional and digital medium to promote the same. The individual-level data is not yet available for the tv consumers, however, the same is “work in progress”; once available, it will make our tv consumer preferences much sharper and significantly boost our marketing ROI.
AIM: What are the challenges of analytics adoption in the media industry?
SK: There are primarily two challenges that the media industry faces:
- Absence of Granular Data: Unlike digital, TV data is grouped at the demographic level and broadcasters are dependent on BARC for the same. While a lot of analysis can be done at the operational level, the consumer level is limiting.
- Using creativity as a Shield: Any change causes some resistance and media industry is no exception to this rule. Older the practitioner in this industry, more vehement opposition to using data backed decisions. The most commonly used weapon against this change is ‘creativity’. The creativity of the media industry not only assumes primacy but is also equated with gut and relationship-based decision making, thereby discounting the power of data.
AIM: What does analytics and data science hiring scenario at Viacom18 look like? How do you ensure smooth hiring in data science which is often considered a big challenge?
SK: There are primarily two things we are doing to build our hiring pipeline:
- Brand Viacom18: We continue to deliver high impact models that help build brand Viacom18’s prowess in data science and helps us attract talent and build our name in the data science circles.
- Internship Programs: We have identified three premier institutes whereby we are nurturing talent by institutionalising internship programs therein. Our endeavour is to increasingly hire through the PPO route.