Our target audience does not watch content thrown at it randomly

Data is almost turning into the Scam 1992-BGM  for this OTT Major. Seldom does one get the itch to compare a developer to Achint Thakkar. But Sumon Mal, Vice President of Backend Engineering at SonyLIV, gives you that vibe when he explains just how deftly, and precisely, his team and tech partners strum up possibilities using various strings of data at this OTT major. I almost asked him about the next season of ‘Rocket Boys’, but it turned out that there is a lot more thrilling than that which he could give a peek of in this chat. And it looks as nail-biting as some of the shows we gulp hungrily. Binging on data, on the right couch can mean a lot. Let’s press the ‘play’ button.

What role does data play in the OTT space? What tools do you use?

There is content on one side, and there is a consumer on the other side. Data helps us manage both in a very intelligent way—from content management to customer management based on subscriptions, packages, choices, taste and real-time consumption. With a wide global reach, there is a lot of diversity on that front, and data becomes quite useful for the insights we need. We realised that a structured data strategy would not help much, so we leaned towards the unstructured models, and MongoDB came as a category-killer top choice when we were looking for NoSQL solutions.

Tell us more about this in-house CMS project built on MongoDB.

We are not selling content but experience.

The in-house CMS system, aka Blitz, provides fast, error-free publishing, while it brings a one-stop solution for operators. Earlie, it was juggled between three different systems. It brings a series of validations in place to ensure error-free publishing, while we are using a moderate number of metadata for each content, its actual performance comes when required bulk updates are required. On Mongo persistence, overall data handling is robust and simplified.

What has changed with this shift?

It has been cost-effective and performant, and it helps us to innovate on many use cases. In the technology industry, nothing is fully secure or predictable, but when things go down, MongoDB has given us support in critical situations. That’s what matters to us, and this makes it more than just a well-chosen solution on the price front.

You begin your journey with TCS and AWS.

How well does this move tie in with those incumbent environments?

It is a fresh in-house project with MongoDB. We created this as a service-based company and explored how intelligently we can publish content, make it market-friendly and performant. This CMS needed strong pillars of metadata with different geo-location, season, and episode-level insights. I come from a developer background, and with that lens too, MongoDB did not disappoint us. It began with CMS search-latency improvement, which was quite compelling. With Mongos’ right index optimisation, it not only reduced the size of the data set but also improved the query time.

Has the new data approach been scalable? And does it flow well with the device heterogeneity that’s common with OTT consumption in India?

Yes, we have a database in multiple regions and deployments. That’s the beauty of NoSQL. As to device diversity, yes, this model helps us to grow fast and on demand. The streaming aspect is taken care of in various formats and platforms, and it aligns well with various device configurations.

Your platform has sports and KBC kind of content, which can have their spikes and, hence, low-latency and availability requirements. How do you manage that?

Some of these scenarios are planned, and some are handled with auto-scale-ups. MongoDB utilisation and reports also help in categorisation of various services, giving us visibility on consumption and scale patterns.

How much does data help in monetisation and revenue strategies?

Data means everything. Knowing what a customer prefers to watch, when and how long—these are valuable insights. It helps to drive personalisation in content. It also helps to elevate search accuracy and work on metrics like Video Load Time (VLT) and CTA. We are not selling content but experience. We need to know customer priorities based on search and viewing data—whether the preference is for KBC or mythology content, or towards Sports. To serve your data, we need data.

Data means everything. Knowing what a customer prefers to watch, when and how long—these are valuable insights. It helps to drive personalisation in content.

Should, and how much, OTT platforms help in spam regulation – the way it is being emphasised by some Telco players?

If you are referring to phishing, etc., our application is fully secure when it comes to payments – with end-to-end encryption. But the transactions that happen through payment links and with different subscription plans – those are the points where most frauds happen. As of now, there is not much that can be done there in the OTT world. But what we do ensure is that immediate payment alerts and information about transactions are given to our subscribers when something related to payment happens.

Are ads still a tight-rope – you cannot make them intrusive, especially in OTT, but they are an important revenue model? When do we get ads that we do not skip?

It is about how a person reacts to an ad – and that matters. Sometimes ads are an interruption, sometimes not. Whether a person likes a given ad or not—that’s valuable data for us. We use placeholders, and we keep a close eye on social media for various preferences and feedback. Ads are an important revenue source, so we value viewer preferences and behaviour very significantly and keep track of that. Ads and personalisation are ever-evolving—the content and creativity part is something that is beyond the ambit of technology.

Does the OTT space have any corresponding metrics as TRPs? Are measurement standards in place?

We do have other frontline KPIs and data on what shots, duration, and watch-behaviour a viewer gives preference to. These KPIs and derivative metrics help. VST, RAI, Add target, Watch time, CTA and a few others are what the industry has now.

How much of your content strategy is shaped by data?

Our target audience is very niche. It does not watch content thrown at it randomly. It chooses what to watch and comes to our platform. From Marco to Rocket Boys to Scam 1992—the specific content resonates. And we learn from our audience’s tastes.

Are trends like dual-screen content, passive-viewership content going to be fads?

I believe that content should be engaging. If one goes for subscriptions, then it is better to have content that truly grips your attention.

Would Bandersnatch-like moments happen again at a wider level? How far are we from interactive content? Is it a technology issue or a content issue?

It is an unknown territory. We need to experiment and adapt to respond. We do not know yet how people will react—especially if the interactivity is forced at crucial moments in a program or movie. That said, it is not a technology problem. It is a high-risk area,  though.

What’s next?

AI is not the future; it is the present. We are focusing on AI-assisted development and automated advancements, especially in areas like validation, and MongoDB is collaborating in these initiatives.

AI is not the future — it’s the present. We’re advancing AI-assisted development and validation at scale with MongoDB as a key partner.

pratimah@cybermedia.co.in