Lights, Camera, Data

Nate Nelson: Hi everybody and welcome to the Data Masters podcast from Tamr. My name is Nate Nelson. I'm sitting with Mark Marinelli, the head of product at Tamr, who's going to introduce the subject and the guest of today's episode. Mark, how are you?

‍

Mark Marinelli: I'm doing fine, Nate. So not many people think of media and entertainment as an industry that is run by data. We think about Hollywood, we think about movies, actors, scripts. There's a ton of unstructured data there for sure, but as organizations try to leverage their data for competitive advantage, capturing, mindshare, et cetera, we don't really think of that facet of the media and entertainment industry. But the reality is that behind the scenes there's a huge infrastructure of very complex supply chain of data that helps shape what trailers you watch, what gets shown in your local theater, and what date these blockbusters are going to be released on.

‍

Mark Marinelli: So today we've got Eric Iverson, former CIO of Sony Pictures and former CIO and CTO at the top talent agency, Creative Artists Agency or CAA. He's going to take us on a whirlwind tour of all of the different facets of this data supply chain in the media and entertainment industry and to talk about how Hollywood has historically leveraged their data, and some of the more recent advancements they've made in being able to drive really interesting outcomes with the data that they are constantly collecting from us, their consumers.

‍

Nate Nelson: Let's get to it then. Here is Eric Iverson. Eric, could you introduce yourself to our audience?

‍

Eric Iverson: Absolutely. I'm so glad to be here today. I love this topic of data. I started working on data probably very heavily back in 1996, not counting all the work that we do in college of course, but when I was working at Accenture, which was Andersen Consulting at the time and I started working on my first athletics project back then and I kind of got hooked on it. As the years moved forward, I eventually moved over to Sony Pictures and pretty early on, really back in 2001 got very heavily involved in all of our data projects, basically digital and data projects and the reason for that was the industry was about ready to go to a major shift. I was working on some of the strategy projects then and two things were pretty obvious to all of us in the industry.

‍

Eric Iverson: One is that the industry was going to transform to all digital workflows, or if you think about that, all data workflows and the second guy is that this was going to be data that was going to drive those digital workflows. And we used to talk back then, which you could understand now, which is if you could imagine if all of your product is digital and everything that you're moving is all of these digital assets around, but none of it was described. You have a lot of assets that you can't find anywhere and you wouldn't be able to move them around.

‍

Eric Iverson: So that's when I got pretty heavily involved in all of our core data projects at Sony Pictures and really pretty much that was most of my career at Sony, was kind of focused on data and digital. Eventually becoming the CIO for so many pictures television and then moved over to CAA in 2016 and similar kind of opportunity, which was great to be able to take some of those learnings and then move them upstream with a very talented team that I ended up having the good fortune to get to work with and I have continued to do that in the industry. In between I have been involved in a lot of the industry initiatives in around data, which we'll talk about in a second. We'd talk a little bit about how media and entertainment works.

‍

Nate Nelson: So entertainment isn't the first industry that people think of when they think of big data. Take us through it then. In what ways are you leveraging big data and analytics?

‍

Eric Iverson: Yeah. It's such a good question. I think that a couple things, one for the listeners would be simply let's ground ourselves. If you are out there right now, not driving and listening to this, I would say take one second, pause this podcast and go watch a trailer. Your favorite film coming up, and as you're interacting with that trailer, I want you to think about the fact that every single thing that you are watching right there is in fact data and the way that that was made in its supply chain was driven by data.

‍

Eric Iverson: In media and entertainment today, our product is 100 data at the end of the day. And that's the way it is today, but that's not ... The industry's relationship with data goes way back and if anything, it's one of the industries that we've loved data about for a long time, but we never really thought about it that way. So if you go back all the way in history, we started tracking box office is kind of like a box score in a baseball game and this has been really important benchmarks for the way that we start to measure how everything works in our industry and you keep moving forward on that. Very, very soon after the invention of television we started doing things like Nielsen ratings and being able to do audience measurement and all of these things were kind of the box scores and really understanding what our audiences wanted.

‍

Eric Iverson: So for a long time in the history of media entertainment, we've been enamored with data. All of the studios have always had research groups that have been very focused on that and pulling data and understanding what audiences like and don't like and using that information to actually market the film and even alter production. One of my favorite films of all time, because I'm a baseball player, is The Natural, and to give you an idea of how much of an impact actually data can have or research can have into that is the entire film The Natural, the ending was rewritten based on research and data. They went out and did panels and discovered that people didn't like the sad ending that was based on the original book. And so they actually rewrote it to a happy ending and it's now known. I mean I don't think anyone could watch the movie The Natural and not imagine the happy ending at the end of The Natural. But if you went back and read the book, that's not how that story ends.

‍

Eric Iverson: So we've had a very tight relationship with data a long time. You start moving into our digital age and all of a sudden now you find it pervasive pretty much across the entire industry.

‍

Nate Nelson: Right. Nowadays I imagine that because the entertainment industry is so big, you must have so many different players, lots of stakeholders and data silos that may not have been made interoperable with one another. How do you get diverse data silos to work together and what good can come from doing so?

‍

Eric Iverson: It's such a good question. You start looking at the supply chain and you hit the nail on the head, which is that, it is an industry with a lot of players in it. So we have all these players and by analogy, we have supply chains that have to move from phase to phase. So if I took it a high level, you would say, great. For those companies that are the beginning of this process, like the CAA, they're interested in trying to figure out what are all the opportunities out there that we can match make. They're either creating opportunity or they're matchmaking opportunities with the world's greatest talent. That's kind of one example of that. And so you can imagine they're part of the supply chain is, the simple question of what are all the opportunities?

‍

Eric Iverson: If you ask that question on a global scale, that's a pretty difficult question to answer. Imagine that, like what are all of the opportunities to potentially produce something for all the audiences on planet earth? Where are the talent and where are the opportunities? That's a difficult question to answer. CAA has these amazing relationships and that's one of the things that it does incredibly well. So obviously tracking all of that and making sure that they have the best information is really important. But in today's world that's moving so quickly with so many productions and entertainment opportunities, being able to track that on a global scale and matching that against your account is super important to be able to be that organized.

‍

Eric Iverson: So that's the first thing. And you can imagine there's data providers out there that provide some insights into that, that you can get from different places and you have to glean that. So an example might be coming out of the US but going into some European territories, you might have to start doing tracking things like asking the question of like, Hey, who's been producing some really interesting film or pictures in some territories across the world on some smaller channels that are becoming bigger as that opportunity grows? And so you had instead of just sending scores of human beings out there, you also might be able to leverage some data sources to start understanding that.

‍

Eric Iverson: Then once you finally get all of that packaged up so that you can get people together and produce things, you move into production. This is where it gets so much more complicated than I think most people understand. The production process in and of itself has two things that make it hard. One, is that there's so many components to that and today it's gotten a lot more complicated and in addition to that, the stakes are higher and that we're trying to drive the timelines down as quickly as possible. Think of it is like a manufacturing supply chain, what does that look like in storytelling?

‍

Eric Iverson: And in the spring telling you this, you have stories that you have to start off with, but then you're having to assemble all of your teams and you're storyboarding things and you have to figure out your set design and you have to go find locations for places and you're going to cast, try to get the right casting for all of these. These are all things that are happening early on in this pre-production leading into production. And as you can imagine, each one of these things has lots of data behind it. You went to any of the studios, you're going to find that like Sony they're going to have an entire library of what they have for costumes. There's costume databases and they have location databases for where all these things could be shot in different locations. You have props, everything else that you could imagine that you could use into a production.

‍

Eric Iverson: That is just now the tip of the iceberg. As we've moved into digital production now you start having all kinds of visual effects. You have lots of shots, because you can imagine, what do you think happen when all of a sudden digital photography came out compared to film photography? We got a lot more footage than we ever had before. So when you had film photography, you know you could shoo a shot during the day and it's going to cost you a decent amount of money, you're paying for the film. But once the price of actually shooting on live film almost vanished, and now we're just down to a hard drive space. The cost of shooting lots and lots and lots and lots of footage, when that went through the floor, the amount of content we produced went through the ceiling.

‍

Eric Iverson: And so now we have so much more content in the production process than we ever had before. And then now you're trying to layer that on in workflows with visual effects and all of the editing work. And then as you can imagine is we're going to lead into distribution, the marketing teams want access to material earlier and earlier and earlier. And your audience sees this, I mean how many people, whether it was Star Wars, imagine your last big film that came out last year. You don't want to see a theatrical trailer. You get teased with an early theatrical trailer and then you somewhat expect that pretty soon they're going to have another one that's even better. And then another one that's even better.

‍

Eric Iverson: That's your marketing teams trying to get access to the production material as soon as they possibly can. Those shots are a composite of live film and editorial and visual effects trying to actually get into that right content that has the right access. Boy, that can be quite a challenge. And for the listeners out there, this is all driven these days largely with data, because data and metadata for all of these assets that are going through these workflows, these things are actually defined, like access control is defined by data. So the question is like great, I just shot 35 shots of this one scene. But two of them are the ones that are interesting. We don't want people looking at all of the other 34 we want them to look at the two that are out there. What's defining that metadata about the shot?

‍

Eric Iverson: So we need the right data moving through those supply chains and later on there may be all of these other legal clearance questions that are going on. So legally is this something we can use in marketing or not? Or if you have big talent that is on a production, can they have or can they not be able to intervene? And do they have a say in whether or not it gets used. So these are all things that actually happen intra supply chain that's just in production.

‍

Eric Iverson: And then you finally get over into kind of distribution and imagine, now I'm going to simplify all this. Someone is now dropped off the final package, and I'm going to oversimplify this because it plays out more complicated in media entertainment supply chains. But imagine this final product gets up and it's like great, here's Star Trek 20 and we've just produced it for United States and the whole thing was shot. All the actors spoke English and we're done. Well, if you can imagine you are not going to distribute that just in the United States. You're going to do distribution everywhere in the world where possibly you have Star Wars, Star Trek camps.

‍

Eric Iverson: So if you're going to basically take that finished product, you now have to consider, I need to have that available in every language in the world. Does that mean I have dubbing in every language of the world? Am I doing subtitling for every language in the world? And do I have assets and marketing asset? And when the opening scene comes up and it lists the title of the film, what's that going to be? Because I could tell you right now, Spiderman isn't Spiderman in every country of the world. The word doesn't even mean anything in some places.

‍

Eric Iverson: So all of these films have to have unique title everywhere in the world where they're shot, which means they're going to have unique assets tied to each one of these things. So just one film is going to be super complex. I'll use the original Spiderman, because we did some very specific asset analysis. Just for that original film, in major components, not all the small minor components that you would do in digital effects, but on the major components like, Hey, the film title's change, we have this trailer, had over 3000 core assets just to composite into the film in distribution. And so you add all of that up.

‍

Eric Iverson: Now you have to have all this metadata and data that's tied to each one of these assets so that ... Imagine if your goal is that I'm delivering Spiderman into Spain and I need to give them all the right Spanish metadata and they need to have all the right Spanish assets and they need the right title of that film, that version of the film that actually has the Spanish subtitles or dubbing with it. That is a tremendous amount of data orchestration that you're going to see. And that doesn't happen just in broadcast, that's happening in DVD and digital markets. And as you could imagine, every company, whether it's Google or iTunes and whatnot, has different standards for what they liked to see getting to them, and so they could do their work in exhibition.

‍

Eric Iverson: And I could tell you that categorically because Sony was on both sides of that supply chain, both in the distribution and it did exhibition, so it had digital exhibition in Crackle, had 155 linear channels at the time. And so we would sometimes be delivering content to ourselves. And so you see this kind of digital supply chain all the way through and has all these complexities, which drives to this reason why we have all of these data organizations trying to provide standards in media and entertainment.

‍

Eric Iverson: So media, entertainment, I personally worked with more than a dozen standards organization dealing with trying to figure out how to make this more common. So for us [inaudible 00:17:45] and for us to really be able to try to interchange data from one to the other. But as you can imagine, that is a hard road. It's been a long road. All of them have made good progress and they're all working hard and it gets us so far. But when you have just in the US over 600 TV productions with all of these independent companies every year going on right now, as you can imagine, it's pretty hard to get the standard. So you're going to end up with messy data flowing through the pipes. And at the end of the day, when you get to exhibition, it's got to be right.

‍

Nate Nelson: Can you give me an example of how your work directly impacts the bottom line in the movie industry? So maybe a story where your work helped push the profit margins for one film or cut costs for another film.

‍

Eric Iverson: So I'll use as an example. There's this concept, if you've heard it out there, I'll explain it. There's a concept if you ever hear from medium stream called day and date, and what that means if you ever hear that word, day and date is that the day and the date that a film or a television show is released, it is released everywhere in the world all at the same time. Now, as you can imagine, that has one benefit of fighting piracy. Right? So if I can get ahead of the pirate, then I have an opportunity to be first. And then that actually theoretically drives maybe some more people to actually be in a subscription or a paid for scenario. That's the general concept.

‍

Eric Iverson: In a market, for whatever market is, I'm releasing it. So you could actually somewhat theoretically actually have day and date multiple times. If it was a film, I could do day and date theatrically, which means it's in every theater or most every theater in the world. We'll put a caveat on how large the market is and then it could later on be day and date we're releasing it globally all at the same time in television. And why would it be that it would be difficult to do that? Well, if you go back the snip before it, if I release it everywhere, I mean all the metadata has to be done. It means all of the subtitling has to be done. It means all the dubbing has to be done.

‍

Eric Iverson: All of those steps that you could in a previous world wait and do later, like let's get the US product out and then we'll go make all those dubs and subtitles and fix all the metadata and determine the releasing and all of that. So there's a lot of complexity and try to figure that out. So what I exposed is actually two things in the same moment. If we use this releasing strategy being really, really clever and smart when you release a film, which is a releasing, like picking the date. Picking the date is a big deal in the film world. If I just took a step back and just asked you to imagine for a second, what are all the data points that you would want to know if you were to pick a date? You're going to pick ... You're going to be responsible.

‍

Eric Iverson: Let's say Star Trek 20, but Star Trek 20, again we'll use that example. The most amazing and sought after film of 2020. You're just going to imagine that. I'm making this up, just making sure everyone knows I'm not broadcasting a product called Star Trek 20. Make believe title. So imagine all the information you would want to know. You would want to know, great, when can the film be done? You're going to want to know what audience affinity does it have and when do they typically go to the theater. You might want to know what weekends have other major films coming out that I would compete against. You might want to know time of year.

‍

Eric Iverson: I'll use as an example, there are times of year in certain countries like in Europe where the summers are beautiful and gorgeous and everyone wants to be outside. People don't stay inside watch television and they have different patterns for going to the theater. So there's all of these things about the audience and the product that you would want to know for picking that date. And if I gave you an example of how important that picking that date can be, and this was absolutely a data influenced decision.

‍

Eric Iverson: If you go back, one of the most profitable films of all time at the Sony library is actually the original Spiderman. If you compare time, give it a fair comparison in time. One of the best performing films of all time, it set box office record at the time and I encourage everyone to go look it up. They picked a really clever date and they were the first ones to do it, which was picking a date before Memorial day weekend and no one had done that before. If you look at the day of the analysis you're like, Hey wait, this is actually a really big long weekend and it doesn't have a lot of competition. Let's release it. Then going into it, a lot of people thought that was a very strange decision. Now you'll realize that look over the last 10 years, there's a lot of big films that get released over that weekend, so that weekend isn't as good anymore as green field as it was when they made that original decision.

‍

Eric Iverson: So you can see just in this called releasing area, just releasing the film. You have both workflow components that are critical for you to play, both offense and defense and you have analytic information for you to be able to make good decisions about when you release the film. And that absolutely can have an effect on the needle. So hopefully that was a good example. I definitely could give you a couple others that would certainly make the point about why getting it right is important.

‍

Nate Nelson: Yeah, I'd love to hear them.

‍

Eric Iverson: So I had this experience once at a phony, because I was at a horizontal role and I was working with two groups. I was in a location where the channel operation group and a distribution was in a smaller office distribution group. We're literally sitting a floor apart and as they have for years, through an atrium. So literally you could look up, if you looked up, you could see the person working in the channel operations group, could actually see the person in the licensing group above and kind of two different business units.

‍

Eric Iverson: So the person in a network channel group has now licensed content from Sony pictures distribution. You following me? So this is the channel group. They have a channel like ABC and a little bit just imagine ASN and which is one of the Saudi channels. And they've licensed content from the distribution group. Maybe we'll use just for fun, we use a film or a television series that was produced by Netflix actually that Sony pictures was distributing, which is House of Cards. So internationally Sony distributes House of Cards, which might feel weird today because Netflix does international distribution. But back when they originally produced that show, they did not have a presence in international markets. So they want to make money on it. So they went over to Sony and said, great, you guys license it out for us.

‍

Eric Iverson: Imagine you have the distribution group having House of Cards, a channel that wants to license it. They license it now from Sony. And what actually was happening is that the Sony distribution group sent out a package like they would send to anybody else in the world. So we had all the metadata and it went off, it went through the legal group and the sales group. And when it got to the channel operations group, they literally started over from scratch. Like they did it if they received any piece of content from anywhere else in the world and they're like, great, so what show is this and where can we find the metadata? Oh, we better look that up on INBD. Let's go find.

‍

Eric Iverson: And literally the company has every single piece of information ever about this title. And we're having this conversation, and as we were going through, so I'm like, have you ever asked downstairs whether or not they just give you the data and they're like, Oh, I don't think, I don't know. I don't think they'd give it to us. And so we set up that conversation and they're like, yeah, streak it out in the data. And just like that, we were looking at the data flow, but that's just giving you a sense, inside of a company that's very well organized, it's still difficult because so many of the other parties out there just don't operate at that level of sophistication.

‍

Eric Iverson: So those folks that are in exhibition most downstream are actually used to having to deal with data coming in from every single source imaginable and it's all over the place. A big studio probably gives them really good data and the mom and pop shop that's doing a production in Budapest, Hungary may not have the same level of standards or even video quality standards. And so the industry has actually created these kinds of like data wrangler and asset wrangler positions. And they've had that for a long time. I know that that's something that you see now, a lot of companies they actually have like a data wrangler or a data management team, but these groups have actually existed for a long time in media and entertainment to try to deal with all of the different data they get from all these different sources.

‍

Nate Nelson: That was Eric Iverson. I'm back here with Mark Marinelli. Mark, any last word that you can leave us with?

‍

Mark Marinelli: The key takeaway for me was how complex the data supply chain really is for media and entertainment and how balkanized so many of the relevant data feeds are across silos. Everyone listening to this podcast has some degree of data disarray in their organization, but to hear what Eric had to deal with and how he dealt with it brings some confidence that this is a tractable problem rather than try and apply a top down standardization for all these different data sources, which would have been impossible given the level of fragmented ownership and collection in all of the systems that he has to gather data from. Eric was able to accommodate the variety and gain tremendous leverage for his business by being able to corral all that data together.

‍

Mark Marinelli: So I think there's really a lesson in there for all of us, regardless of how insurmountable the challenge may seem and how many different constituents and stakeholders we have and the data that this actually is doable. And the degree with which we can provide extremely targeted analytic outcomes from all of these data is something I think Eric gave us some high confidence in and some good examples.

‍

Nate Nelson: Okay. That'll do it for this episode. Thanks to Eric Iverson for sitting down with us and thank you, Mark, for your insights.

‍

Mark Marinelli: Thank you.

‍

Nate Nelson: This has been the Data Masters podcast from Tamr. Thanks to everybody who's listening.