How to Improve Your Cloud TV Service KPIs Interview

Turning Data into Actionable Insights: How to Improve Your Cloud TV Service KPIs Interview

in our last webinar, Kaltura´s and Jump´s  experts, Jerónimo Macanás, CEO of Jump Data-Driven and Gideon Gilboa, executive vice president of Products and Solutions at Kaltura’s media and telecom division had a chat on how you can turn your data into actionable insights and improve your cloud TV service KPIs. In case you couldn’t make it to watch it, here you have their interview:

“Jerónimo:  Thanks a lot, and welcome everyone.  It’s actually a really fun panel because it’s the first time I am going to act as an interviewer. I hope you guys have fun and learn a bit about us. So, let’s get started. So, Gideon, what did you have for lunch today?

 

Gideon: That’s a great question. I must say. I had data. I’m having data for breakfast, lunch and dinner. That’s all I eat. On a serious note, I had hummus. Very healthy.

 

Jerónimo: Ok. Enough joking. Let’s get down to business. We received a lot of questions that are somewhat similar, and I’d like to summarize some of them: What do you think the current main driver for the convergence of data and TV in our industry is?

 

Gideon: Yeah, that’s a big one. I think maybe let’s start with a little bit of history. You know, in my mind having been in the in the TV space for the past almost 20 years (I only look young), TV and data were very strange concepts. Actually, everything to do with data on TV until not so long ago was like black voodoo or something.  In the early 2000s TV measurement in big markets like the US was partially still done by diaries. People actually went every night at the end of the day and wrote in a diary about what they watched, and that was the basis for TV ratings and measurements for a long time. And I remember that, until not so long ago, it’s easy to forget—maybe 10 years ago—the

discussion was actually about how to get data. You know, I was working for a different company at the time, and we were working on how to get data from PSTN modems connected to set top boxes and trying to compress the data. So, if a decade ago the question was how to get data, the question now is: okay we have the data, what are we going to do with it? and obviously, how are we going to improve our cloud TV service KPIs?

I think that the driver for the convergence – now that the data is available – the driver for the convergence of data is going to be all about the competition. In the US alone, if I go back to the same market, there are now over 200 OTT services that are streaming video where all the data is available from all the devices. In order to differentiate, people understand that it’s not enough to have good content and the right pricing strategy. They also need to have the right data and to understand who they want to target, who they want to acquire, which subscribers they need to invest in. So, I think competition will be the big driver for convergence.

 

Jerónimo: Absolutely. And that links to the second question. We are seeing video service providers that have been investing a lot of money and effort in creating a lot of infrastructure for gathering data and for storing data. So, focusing on this, they are now wondering: what should I do now? how can I improve my cloud TV service KPI´s? have a lot of data, most of the data is updated. Where do I go from here? Video service providers need to transform the video service experiences, their service, into more data-driven entertainment experiences to make the most out of their vast amount of data. So, what do you think the main elements for this kind of transformation, that ultimately is a strategic transformation, are? It’s not just operational, or product or technology driven. It’s more than that. At the same time, what do you think the main benefits and challenges of such a strategic transformation are? 

 

Gideon: I think, well, it’s interesting. It sounds trivial, but I would say that the first element is, boy you have got to have a strategy. I mean, you’d be surprised: many people – not to disrespect anyone, but they approach us at Kaltura as a leading cloud TV platform … we get a lot of opportunities from different parts of the world …  where they approach us as a technology provider before they even have that strategy. They start talking about APIs and latency and video codecs and CMSs and all sorts of boring technology, things like that before they realize what their strategy is. So, what do I mean by strategy? It is like: what content you are trying to offer, what are your business, and marketing goals? As an example, let’s say one of your key marketing goals is engagement. Okay, what are the KPIs you are going to use to measure engagement on your service? Time spent and duration and so on and so forth. And then, only then, you figure out, okay, how do I get ….two things: one a holistic data infrastructure and a data management platform that actually allows you to get to those KPIs. Because guess what: sometimes those cloud TV service KPIs are combined from multiple sources and they’re coming from different places and they require external data. And then once you have that the data management platform you also need the holistic platform that actually lets you act on this data, because you know it’s great to have the insights but you also want to make an impact on the users. So, I think these are the three things, to summarize what I know was a long answer, have the right strategy and the right KPIs, and build the infrastructure around that and not before. And have the platform to act on that data infrastructure. Does this make sense?

 

Jerónimo:  Perfect sense.

 

 Gideon. Good, good. Of course, you know this is easy to say, but then the real challenges start. I guess you, Jerónimo, have been in the trenches, so to speak, on the bleeding edge of getting data into those services. So, let me ask you: what are the challenges you’re witnessing around collect collection of the data?

 

Jerónimo: It’s a tough topic. You’re right: essentially once you have a strategy in place, a second common mistake that we have seen in several customers and at other opportunities we’ve seen in the market is: Well, now we know the strategy and now we want insights right away – or as you mention actionable insights. We want to know what is happening. We want to know what’s going to happen, and then we want to take action on our data and make an impact on our users’ behavior to improve our ROI. Basically conceptually, this makes sense.  But the reality is that our customers first need to go through a journey before they can get to this point. The implementation of a data-driven strategy is a journey, and you need to start at the basics, the foundation just like if you are building a house. 

So first, the main advice we give when people ask us how they should start and what the challenges they’ll face when retrieving data, we tell them all that comes a bit later. First, it’s about fixing your data fragmentation. There is huge fragmentation in data sources. We all know what the video ecosystem is like, and unfortunately there is not just one single piece or just one single vendor or just one single data store that you need to work with to manage your business. First there is creating the single source of truth. That will be a fundamental building block for your data strategy. Once you trust your data sources, when you trust this single source of truth, the rest will follow. It’s true that this is a difficult task. Traditionally  video service operators particularly have been investing a lot in creating this big data capability, two-year-long projects, millions of dollars invested,  and at the end, they oftentimes wind up frustrated because it’s a long and costly process. Cloud and cloud data technologies can now really help simplify the process and help complete this first step.

The second challenge is about building unified insights. This does not necessarily mean that you have all your data in one repository. It means that you really have unified insights because you correlate data. You need to create meaningful relationships between data so you can really look at your video business in a holistic way. So, my second piece of advice to customers facing this challenge is once they have the data infrastructure, they should think about how to really have this holistic view. Otherwise it’s going be very difficult to move to the more advanced phases along the data journey like using advanced analytics or predictions or AI, etc.

Probably, the last challenge is really related with what you mentioned, Gideon. It’s about making data easily available within your organization. That’s a very common mistake:  there are a lot of reports; there’s a ton of information available, but people cannot really use the data available in very creative ways. And it not just about using data as it is meant be used. You can use data to reinvent your business models. You can use data to redesign the user experience. If you are creative, you can use data for tons of things. It is also very important in some way to “democratize” the organization’s internal data, make it available to everyone who should be able to access it.

Lastly and on the more advanced part of the journey, you have to take into consideration that when you move to the world of predictive analytics, you are not dealing with data science as you knew it. You are working with BIG data science. The scale of big data (versus just data) science is important in the streaming industry. When you try to build machine learning models for whatever predictions – churners or trial user conversions, whatever you need or think you are going to need to build models – using data from  millions of subscribers, you will need to make sure that you have the capabilities to manage the data scale. This is not a trivial thing in data science. And of course again just to finalize: this is a journey, so you need to start step-by-step What is important is that video services start that journey now, begin the process today.

 

Gideon:  I think that makes a lot of sense. The other thing is that companies like yours have done actually, I think, made it simpler for companies to kickstart this type of project. From what I’ve seen, you don’t have to run a multi-year, millions of dollars project in order to get results. You could get things much faster with the capabilities today that companies like yours have.

 

Jerónimo: Totally, that was the essence of our company, actually. To try to build up the best practices in data from the best companies that had built capabilities in-house, let’s say Netflix, Amazon Google basically. Then put all this in a SaaS platform that people can use in a very easy way, that they don’t require a huge investment, and they don’t really need years of projects.

 

Gideon: There’s one thing that always comes up, and I think it is a fair question to ask, and we were asked it again here. I’m interested in your view.  You talked about the journey. People always talk about quantities. Like: how do I deal with the amount of data; how much data should I say I need to delete; do I need to keep years and years of data;  what actions should I take to deal with these volumes of data?

 

Jerónimo: That’s a good question. Honestly, I think that it’s been changing over the last one, maybe two years. What’s sure is that that question has come to the table very often over the years – because data volumes equal cost, essentially, right? It doesn’t matter if it’s on premise or cloud. Of course, on premise overall requires more investment, but with cloud, at the end, there is a cost component to the data. The reality is that the cost has been going down and down and down. There is a threshold when – depending on your  volumes of course – there is a threshold where the process of managing data that you don’t want to ingest versus ingesting all the data that your video service is generating,  starts to pay off. I would say 70{1e556b0dda4010577b784b4e8e746da95ecb739eaec1707a25fd3098bb696a30} of our customers – and some of them are really big – decided to just ingest all data sources. I mean all that data is not necessary.  Ingesting every single piece of data that your service generates isn’t needed,  but everything around audience registration data, content catalog, playback activity, commerce events, subscriptions, instructions, ad monetization, and UX activity – all that becomes mandatory core ingestion data portfolio.

Honestly, what’s most important is not what data you ingest. What’s important is what data you are going to focus on to really create these meaningful relationships between data points, gaining the insights. And that’s entirely related to the maturity of your video business and completely changes the focus. So, if you are a new OTT service or video service and you’re in launch phase, your focus is growth and that means that you’re going to focus on conversion, right? So, you need to focus – make the most out of your insights – around how to optimize your investments in acquisition channels. So: which channels should I invest more in? Which ones pay off or have faster campaign payback? All these questions are important. How can I focus on those trial users that are more likely to convert into paying users? So you need to focus your efforts both not just on marketing activities, but also on which data you will focus on. When you move along the journey, when you are further along in more of a growth phase, then your focus is retention.  You need to keep the subscribers that you have acquired. For that it becomes very relevant how you treat things like churn.

I have an interesting a use case that we recently finished, and we are still seeing the results. It is for a big broadcaster with a direct-to-consumer service. They were very worried about the end of Big Brother. They thought that when Big Brother ended, they would have massive churn. So, instead of focusing on their entire audience base with hundreds of different variables, with our help these guys focused on understanding the reasons for Big Brother lovers to churn, comparing last year’s results with this year’s data. We ended up with a very interesting result. Now what they are doing is driving people that were Big Brother lovers to other content that might also be relevant to them. As a result, they have held onto a lot of their viewers, compared with last year, when they didn’t take targeted action. Those are the kind of things you should focus on in the growth phase: retention. And then of course when a service is mature it’s all about engagement. In that phase you need to focus on the quality of your content catalog, what are the different audience segments that you have, how are you going to treat these audience segments differently, etc. In summary, depending on the maturity of your video business you should focus on certain data points more than others. They are not all the same.

 

Gideon: Agreed.

 

Jerónimo: So, let me review some of the other questions we received. Some of them are about the measures service operators can take to make their data operational and improve the usability of the data. It looks like a lot of people are worried about how they can fill this gap. They have the data and want to know what’s the next step? How can they jump in? What’s your take on this?

 

Gideon: I think in general it relates a lot to what you just said, because I think it is a fear, right? You’ve got this great data lake and you’re ready to jump in and swim in the data lake but you’re not sure how to dip your toe in the water. So, I think to me … one comment and it’s easy to say as a vendor …. but I truly believe in getting help, get outside help. Because sometimes – and I’ve seen this, especially on the data side – there’s this fashion, “you know let’s just build our own infrastructure and see what happens.” And it’s not easy; it’s not always trivial. You can shortcut. I know that when we started as a vendor, our internal journey using vast amounts of global data that we have from over 50 million monthly active users across the globe, multiple businesses cases AVOD, SVOD, ATV. We needed help. We couldn’t get the value out of the data initially. So, that’s point number one. It goes back a little bit to what we said earlier, what I talked about with regards to goals. So, understanding what your business marketing goals are. To your point about which phase that you’re in the maturity of the service, but I want to add another layer to help understand the concept of usability of data.  It’s the personas: the different personas within the organization that require different types of data and even different types of infrastructure for their data. So a couple examples: your operations people might be interested in quality of experience data, because all they care about is if the playability percent is high, and that quality of experience data actually requires some near real-time infrastructure for ingesting real-time QoE data from the player and player events, and so on and so forth. If you understand that you have these people in your organization and that’s part of what you need, you will also need that infrastructure and understand how these people work. Like: the person that is in charge of part of the operations team wakes up every morning, and what are the five metrics that he or she looks at. And how to make that data accessible.

On the other hand, the marketing side is more interested in consumption and maybe the technology there also has more machine learning or other types of AI to predict churn, like the examples that you mentioned. So, I think that’s one thing. Look at the personas and what they need. The final thing is: think about the receiving-end, like ultimately who is going to act on the data, and how. So, for the QoE persona, for example, the operations person: what are they going to do if they find something wrong in the data? If they find that the playability is going down, are they going to make some changes in the player? Are they going to check the CDN, and what are they able to do? So make sure there is a path, an end-to-end path. The same goes for marketing. The marketeer wakes up every morning, and what do they need to do? I mean, can they actually create offers based on the attributes that the data is showing them? Can they target people? Can they create promotions? And do they have the right tools? I would say these are the main things to make data really usable in real life.

 

Jerónimo: That’s an interesting point, and the I think it is related to the accessibility of data. Accessibility means different things for different personas. Maybe the tech guys are going be very interested that you open up you know your data lake for them to query and look and navigate through the data as you mentioned, and then discover very specific things. The marketing guys are probably just expecting a daily report in their mailbox that tells them how their business looks today. That aspect of accessibility is crucial so data strategies don’t fail.

 

Gideon: Agree.

 

Jerónimo: Do you have any take on the importance of the timing related with data? I mean we do see many organizations that create these extensive reports of 300 KPIs in an excel file. Then it takes a couple of weeks to generate it, and by then, nothing can be done. You have lost the window opportunity. Where do you think the market is moving in this respect? Are there any trends you are seeing when people or your customers require data, “trial data”?

 

Gideon: I think again it goes back to what I said earlier. We are seeing it, as the platform in places where it works well, we’re actually seeing feedback that comes in to us – based on real-time insights to help our customers optimize things. I’ll give you an example: one customer looked at – it was actually more quality of experience data – and they looked at the data and they saw that in certain regions (this is a developing market) it takes a lot of time for the first video to load, because establishing the DNS connection in that market is difficult and with analytics, deep player analytics, they were able to understand exactly how the networking layer of the player is working and what’s going on there. We could then provide a kind of “fix” or an update that started to establish the connection as soon as you open the app and before you click on the first video. So within a matter of a couple of weeks we were able to give them a release that solved the problem that was very acute for the business based on data. And that’s a good example. Of course, there are bad examples too. The other thing is really how you close that loop. How do you, and this is something that you and I have discussed a lot, how do we continuously make sure that we get some data? We make some change; we get some information about what that change was; we iterate. It’s easier said than done.

I think in many, many cases, it’s easier said than done, but I think that’s an important thing. If you don’t have the ability to do that, to add a new business model, to add a  new promotion, to add a new discount, to target a specific segment …  the data insights are not going to help you, and if you don’t have the right data insights and the right source of truth, to go back to your earlier comments, then whatever tools you have also won’t help you.

Where do you see, when we’re talking about how people need to improve their services, where do you see the focus? Where should it be? We talked a little bit about QoE, we talked about marketing. Where do you think it should be?

 

Jerónimo:  I really believe, of course, that any video service provider’s value chain step can be optimized with the right data, with the right use of data. From the examples that you mentioned of QoE but also for example there is a lot of room for improvement in encoding optimization. We have seen that during this pandemic. And of course targeted marketing, as one of the examples that you mentioned. So select specifically a group of people who you know a lot about and then be very concrete with the actions that you do, and track back for four months. There’s a lot of room in there for improvement.

But again, I will think today it’s mandatory to optimize your core business processes in video business. That is acquisition, retention, and engagement. We have seen a trend change in the last, I guess, one year, last twelve months. At the beginning I remember that when new video services kickoff they were very – or just – worried about quality of experience, or the user experience in their apps. More and more, when we find new customers, they come with the lessons learned; that from day zero they really need to manage their business using data. They don’t think they can put that in a second phase.

For example, linking to one of the previous examples, more and more we are finding new customers who are asking us: “hey, Jump, how can you help us really assess in which acquisition channels to invest, because I don’t want to invest my whole budget, my whole yearly budget in one specific channel. I want to be able to really, week by week or month by month at the latest, shift and then allocate the money in the right acquisition channels.” This is something that you can really look at with data. You can look at data and how long it takes to pay back that campaign, which channels drive a higher customer lifetime value compared with others. These kind of things are really there.

This is one example I think is really relevant. And then of course depending again on the maturity of the service, how you help these video services to convert better and retain better and engage better.  This is all about getting the insights, being able to predict once you have a certain number of months. People think that you cannot make predictions until you have, say, three years data. That’s not true. Actually three years’ worth of data doesn’t help with predictions because customer dynamics and user behavior can change a lot.

From three to six months you can really start to use predictive analytics in a right way, by getting insights, making predictions about what is going to happen, and then iterate with marketing campaigns, assessing the performance of those campaigns. That’s where I would focus if I were launching a service, using data to create this cycle of data analysis and data action that really drives the decisions that we make. Just to give an example on this as well, and I think that’s a very important point I think we should discuss:  people think that AI is all about automatic algorithms that are making decisions for us. I don’t think that is the smart combination.  We have seen today, probably most of the audience has seen it, the announcement of HBO Max. Essentially they are differentiating from Netflix in that they are going to combine AI and human editorialization all together to create a more human-focused media service.

Of course, with a lot of AI behind that, but putting a human touch and combining both, I think that is the probably the “secret sauce” to really make the most out of data. You need to have the data, this augmented intelligence, but at the end the experts have a lot a lot to say. I think that’s a common fear when we talk to customers about AI. As if it’s going to replace everything. I do not agree and I’m an AI company.  That’s my key on where to focus.

 

Gideon: Yes, I agree in the last point. I think one thing that we haven’t discussed so much here is that the idea of ….  we talked a lot in the last half an hour about how data helps the streaming service providers, but it should also help the end users, and ultimately find the content they like. I think that in some way, at least talking now as a consumer and not as a professional in this industry, there is always a feeling that most people, most services have not completely cracked this notion of recommendations, content recommendations. I think part of it is to do with a) the right balance between machine and human – it can’t be just human and it can’t  be just machine. And b) the ability to make those, speaking about usability of data and some of the topics we discussed, to create something that’s a lot more real-time in terms of experimenting, seeing the response, and adjusting. At least in my view, going into the future we call this cognitive TV. The idea is that there will be a lot less content. I mean in my perfect world the concept of the rail that everybody is used to – because folks like Netflix have taught us to get used to that and maybe others – is irrelevant anymore. There’s just going to be one content which is either something you are continuing to watch or if you’re browsing, there’s one particular piece of content that’s right here, right now you’re most likely to want to watch and it’s just there for you.

Do you think we can get to do something like that?

 

Jerónimo:  Yeah, fully agree with that concept. I think there are, again, some steps to reach that level. But definitely one of the concepts that are more hot in terms of personalization right now is contextual personalization. So now most of the AI personalization engines are built on top of how similar you are to other individuals and how similar each piece of content is between them.  So mixing both then, you receive a recommendation. But the reality is that now we are going to have much more in the future, in near future, much more channels as inputs of data that can improve recommendations.

Imagine for example once voice controls and voice interactions are much more massively used,  there’s going to be tons of data that you’re going to receive about hours people, sound, emotions, and words they use, and how they ask for specific content. All these things can be put in the in the cocktail, in the equation, and are going to improve that level. In other things, like how are you watching content? Are you alone? Are you at home? Geolocalization  … there are many other things that aren’t there yet. They are going to happen, there’s going to be a big shift in the that real-time personalization that you mentioned that now seems very algorithm based.

 

Gideon: Yeah, agree. So just before we get into the last part of this, maybe we talked about some of the good things. Let’s talk about the bad and the ugly. What are the biggest mistakes you see customers do when it comes to their data strategy? What would you say, “don’t do”?

 

Jerónimo:  Well first of all I would say, if you want to start big data or a data project implementation, the first thing I would say is get upper management support. There are a lot of companies that say “okay I want to do things around data” but they don’t have strong management support.  And that is a source of big mistakes and a reason for failure. That would be my first point. My second would be that you need to make sure that you rely on modern technologies. Everybody knows that I mean big data; it’s been a big shift in recent years. Because of these cloud capabilities that we all are now using, so you need to rely on modern data platforms. Otherwise you’re not going to be flexible and you are not going to be able to create this ingestion of storage processing and unified visualization cycle and these need to be your pillars. And if you don’t have these in the right way, you’re going to fail any way. It won’t matter how much you invest. But that would be my main advice: get management support together with the strategy, as you mentioned; and second, relying on modern technologies. Today, it’s not valid to rely on technologies from five years ago if you want to stand out in this competitive landscape that video services are facing.

 

Gideon:  Yeah, I think you’re totally right. I would add one more thing maybe, which is: think before you act. Because here’s a story: one of the customers we were working with, they had this big – and this was in the early days actually before even like a big cloud big data technologies – they were working on this huge project to identify churners, and there are people with potential for churn – and this is a pay TV operator – and they were working very hard to find those people and they managed to pretty accurately identify who are those are going to churn. And think about it, it’s not always that difficult. The problem was that when they did the experiment, they actually approached a lot of those people and they actually caused churn because they called the customers and said, “Hey we’re not pleased you’re not using the service so much; Let us offer you some kind of a discount.” And the guy says, “wait a second I’m not using the service, why am I paying for this?” So they actually pushed some of the customers into a decision and made them aware that they’re in this decision junction and actually got the opposite result. So, I think that you always have to actually be careful before you act on the insights. The data and the insights do not replace your strategy in your thinking.

 

Jerónimo:  That’s the human touch that we previously mentioned.

 

Gideon: The theme of our discussion, absolutely. That was at least some of the questions. It’s been fun so far, but I’m conscious of the time, and I wonder if we have any audience questions that we want to address.

 

Irna: yes, actually we do. We don’t have as much time, so I think this one is it actually a for you Gideon: How can you use the data to take real-time actions because, while there is time to process the data, but when it comes to real time actions what will be your recommendation?

 

Gideon:  I think I think it goes back to something we said earlier which is: you don’t always have to take real-time actions. It really depends on who the persona is, what is their role, what is the type of data you are dealing with. In the case of quality of experience, if you suddenly have playback failures that go through the roof, obviously you have to take real-time actions. So that is the sort of data where you really need to take real-time actions. I think that when it comes to a lot of the marketing data or user experience analysis, actually not taking real-time actions – speaking about the last example here on churn – it might be a good idea. So you sort of get the batch of the data analyzed. Do you understand? You figure out creating the marketing campaigns which also takes planning and time. So, I think that is around marketing data. When it comes to recommendations, obviously as we discussed the infrastructure and the ability to have real-time experiments and improvements based on content of your service to the users. Also you may want to insert your own, what we call the boost mechanism, so when your service needs to boost some specific content, to combine that with curation and with real-time recommendation and experimentation, this is an important thing in a complex ecosystem to measure and that should be taken into consideration.

 

Irna: thank you for that. We have another question on the monetization of the data from the audience: Do you see ways to monetize the data acquired by the platform through intelligence or analysis, basically by selling the insights or analysis to other platforms?

 

Jerónimo: That’s for me? Ok .. that’s an interesting point. I think there is a lot of a concern right now about data privacy. And all the aspects, especially in Europe, we’ve been in for the last couple of years talking a lot and improving a lot our systems to meet the GDPR policies and now the US is becoming more and more stringent as well.  I think of course you need to meet those requirements because any user’s data privacy is the most important. But saying this, and when you work at the right aggregation level of data there are of course a lot of opportunities, I think, not just for us as a platform, but also for customers. So they can really use data to create specific products for users. Like, for example, the way is Spotify is doing this.

Probably some of the audience knows this,  but at the end of the year Spotify every year sends you a very, very detailed report per user about how much music you have consumed, the tracks that you mostly watched, how long you’ve been listening to specific genres. So maybe there is not a strong direct opportunity for monetization with end consumers, talking about d2c services, but there are a lot of opportunities to  create additional value add products around your core offering that are going to have a huge part of engagement with the user. So, definitely there is a lot of room to use data in that sense.

 

Gideon: I totally agree. If I can add a couple of quick points: monetization doesn’t necessarily have to be, as I think Jerónimo just said, doesn’t necessarily have to be a direct sell of the data. I think that there are two aspects where you can actually see ROI on the data. It depends a little bit on your role. If you’re super aggregator, if you’re a pay TV or telco that has a new cloud TV offering and you’re aggregating content, having data to actually inform your content strategy and make the right content acquisition decisions is super important and can actually have a direct impact on your bottom line in terms of spending money on the right content, and at the same time, having data and – not necessarily selling that data – but using it to … as they say data is the new oil … so

using it to oil the channels, the communication with content providers and advertisers, by providing even greater insights if they use your platform could help you in that relationship, as well, which may have a commercial impact. So, these are two other examples we have seen in the market.”