NBCUniversal drives media effectiveness through modernized TV

Hello, everyone and welcome. Thank you for spending your first day at re:Invent with us. My name's George Borghesi. I'm a global account manager here at AWS and have the pleasure of being your host for today's session.

NBC Universal drives media effectiveness through modernized TV, media, entertainment. Customers, of which we have many of you in the audience here today, are faced with complex use cases driven by changing consumer behavior. AWS is here to be your partner through that journey. We've invested in six key solution areas including domain level expertise, partner solutions, AWS services and resources to provide thought leadership and purpose built technology across content production, media supply chain, broadcast, direct to consumer and streaming, data science and analytics for media, and our most recently launched monetization.

Here to share more about how AWS has helped NBC Universal accelerate their monetization goals, please join me in welcoming Jessica Reed, VP of Product Marketing, who will later be joined by Eddie Lee, SVP of Enterprise Product. Thank you.

Hello. Thank you, Georgia. Thanks, AWS. Thanks, re:Invent. I don't know what's up, what's up re:Invent? It's Monday. We're kicking it off. There's more than eight people, I feel like we're already winning.

So I'm gonna walk through our agenda because who doesn't want to start without an agenda? So I'm gonna do a little intro to who NBC Universal is in case you don't know, I'm gonna do an overview of our One Platform which is our proprietary tech stack, much of which is powered by AWS, hence, ostensibly why I'm here. I'm gonna talk about where AWS fits into that tech stack. And then we're gonna have two specific examples of how we've collaborated with AWS. I'm gonna talk about in-flight optimization for better linear TV performance. I know could not sound more exciting, get ready. And then Eddie Lee, my colleague at NBCU is gonna talk about clean room integrations for better streaming performance.

So who is NBC Universal? We love a good logo soup slide. So let's kick it off there. You may know the NBC part of NBC Universal, which is our broadcast network. You may know the Universal part of NBC Universal, which is our movie studios and our theme parks, but we're actually a lot more than that. And so we own 12 television networks, 13 actually if you count Universal Kids, but we don't target there. Um that includes entertainment, news, sports, I mean, really everything right. We also have Spanish language content, which is a really important part of our portfolio with Telemundo Deportes and Universal.

And we own Fandango, which I think a lot of people don't know. And that's really a treasure trove of first party data when it comes to movie tickets and movie preferences. We're also part of the Comcast family. And so that includes Sky in Europe. And of course, there's our crown jewel streaming service, Peacock. And you can see also we've got the Comcast businesses, we have local, we have syndication. We've got not just Universal Pictures but also Focus Features and two different animation studios with Illumination and DreamWorks and our parks which are worldwide.

But let's get into the advertising side of the business because that's what I'm here to talk about. So for years, marketers have had this choice that they had to make between what you could think of as like the world of big media and the world of big tech. And when we think about big tech, historically, it's been mostly social platforms, tends to be user generated content focused, small screen tends to be a desktop, but especially mobile devices. But then on the plus side, there's these incredible treasure troves of first party data, really automated functionality, proprietary technology underlying everything and then measurement not just on delivery of campaigns but on real KPIs and outcomes.

When you think about the world of big media, which is where NBC Universal is traditionally thought of it's premium video, right? This is not user generated, this is premium world class, professionally produced culture defining content in a lot of ways tends to be on the big screen. Sometimes we call it the big glass, we like to have different words for everything, big glass, big screen television screen. But it's traditionally been transacted using small panel based data. So you know, traditional age and gender demographic data that is based on say 40,000 respondents extrapolated to 100 and 10 million television households in the country. Also, I would like to point out that not all of us live within the gender binary. I myself am non binary. So I don't fit into the women 25 to 54 demographic.

Manual processes just so much back and forth. Excel, spreadsheets, phone calls, emails. There may still be fax machines involved. I'm not gonna say there aren't and it's been traditional demo measurement really focused again on did we deliver to that age and gender demo that we targeted from the start?

But with One Platform, we're really bringing together the best of both of those worlds. And we think of NBC Universal One Platform as TV modernized. This is a video platform, the first premium video platform that combines the power of that professionally produced content, really culture defining moments across news, sports entertainment on every screen, desktop, mobile connected devices, national broadcast and cable television with the benefits that come from big tech.

We have this amazing first party data that we're collecting across all of those properties that I showed on the previous slide. We're increasingly leaning into precision targeting. So that 1:1 addressability as well as targeting through national linear television, which I will explain a little bit more in a minute, automated buying. We're really increasingly building programmatic pipes to all of our inventory. In fact, all of our live inventory across Peacock and our streaming properties is now available programmatically. And we started that in the World Cup last year, we gave programmatic access to this really premium culture defining experience and outcome based measurement. We're really really focused on showing how buying premium video drives outcomes for our clients.

And the result of this is something that we think of as kind of the new social CTV, which is that big screen experience is now super personalized and we've scaled it and made it as relevant as a social feed, you log into Peacock, what you see and what I see are gonna be different, right? Those recommendations are gonna be different. The ads are gonna be different, all of which is because we have this foundation that is driven by data. It creates a better experience for consumers, but it also drives better engagement for advertisers. And we are an advertising organization. So that's kind of a big deal to us.

So how do we do this? We're really breaking with legacy in four key ways. So starting with data and identity, we're going from this small data that panel I talked about and these really broad targets to using big data sets, including our own first party data and precision targets, like really going down to that person level on the activation and automation side.

This is, you know, kind of following the flow of an ad campaign. Once you have your audience, you have to activate that audience. And typically, that's been done platform by platform screen by screen in a really manual fashion. Even on the digital side, it's been a very manual process. If anyone knows Donovan Data Systems, which is now MediaOcean, which maybe you do, maybe you don't. Um they're on the sell side. On the buy side, I'm on the sell side and I worked there when they first introduced like an electronic insertion order. And it was very exciting. Um and we're still catching up in a lot of ways on the TV side, static and cluttered ad experience.

You don't want to see the same ads over and over again. Frequency capping is really important but also creating a really organic ad experience which we're able to control in our streaming platforms and especially Peacock. So we've really created these interactive and high impact ad units. We've also reduced ad load and we're able to control for frequency across our full portfolio.

And finally, we're taking measurement from this panel based bi platform approach. So what did I deliver in my Nielsen women 25 to 54 demo on linear. What did I deliver on streaming. Ok. That's the old way. The new way is using these census based data sources, those big data sources across platforms so that we can see the delivery everywhere that viewer is and then tracking that back, not just to reach and frequency and impressions, but actual business outcomes.

I know we're all about pillars, we're very into pillars. So our four pillars translate them into four product pillars. So it starts with what we call NBC Unified. This is our first party enterprise data platform and Eddie is gonna talk more about how we created data collaboration through NBC Unified. But this is our high fidelity data and identity that we own. That's our first party data for precision audiences at scale.

And we also have a marketplace of third party data to augment with. Ads Smartt is what we call our activation and automation suite. And this is really about data driven activation of those precision audiences on every screen and increasingly making sure that those processes are automated. No more emails, no more calls, no more faxes, assuming fax machines still exist, which they do because doctors ask me to fax some things which is confusing because I'm not a doctor. But that's another story.

NBC Ad Innovation, which is those innovative ad products and formats that really improve the consumer experience. They're often shoppable, they're driving down the purchase funnel and they really drive ad effectiveness and then our measurement suite which again is about holistic measurement across screens across platforms, both for reach and for impact. Think about it as you know, impressions to impact.

And our advertisers have been adopting One Platform in droves. In 2022 alone, we had 400 data driven linear and cross platform campaigns. Over 1500 audience based streaming campaigns. Over 5000 programmatic deals, over 750 Ad Innovation placements on Peacock. And we enabled over 1600 outcome based measurement studies and that's on the buy side and the sell side. So we can do measurement on behalf of a client or the client can do the measurement themselves using data that we provide to them through our data collaboration and they're reaping the benefits, right? If it's not working, what's the point.

So for our self service buying options, this is one of the ways automation is so essential. Um and especially on the programmatic side, campaigns go live way way faster from up to two weeks, down to two days. The creatives when we use those ad innovations are much more impactful, 24% on average higher purchase intent using those ad innovations when we do data driven targeting on television, which is really key for us. Target CPMs go down on it by an average of 16%. So that's 16% better cost efficiency for our clients when they use data and their sales go up 16% on average sales lift from those data driven linear campaigns.

So where does AWS come into this pretty much everywhere? Um AWS is foundational to our infrastructure. You can see across every pillar we're leveraging AWS. Um as I said, I'm gonna talk about linear the linear side of this one case study and Eddie is gonna talk about clean room integrations. But really across the board, we are leveraging AWS and especially S3 to ensure that our architecture is scalable and not having that on prem and having it in the cloud just means everything is faster for us, right? Like all of the processing is faster, but it also means we can go to market faster.

Ok. So some examples, all right, before I get into this, how many people know how national linear television spots are scheduled? I see one hand, two hands. Ok? So all of you basically, we've got experts on linear television. I don't need to explain anything.

So linear television, by the way, I'm using the words linear television, how many people even heard those words before today? Ok. So not very many. So when I say linear television, I'm talking about what's thought of as broadcast and cable television, right? So like what you watch live as it's happening or on demand within say three days or if do people still DVR things i don't know on your DVR. Um so that is linear television and it's called linear because it's a linear schedule, right? Like, there's something at eight, there's something at nine, there's something at 10, something at 11. Um so it's a linear schedule of programming as opposed to streaming, right where you kind of watch what you want when you watch it, that includes both what you watch through a cable box and what you might watch through if you have Hulu Live or YouTube Live, like all of that is still, if it's that broadcast, that live broadcast, that is linear television.

And typically, as I said, it's been sold on age and gender demographics. And the process like a client will come to us and say, I want to reach, you know, I'm just gonna keep harping on women 25 to 54...

"And I want to reach them in this much on Bravo, this much on E, this much on Sci Fi. Like they come to us with where they believe that audience is ahead of time and then we build spot schedules manually around that. So there's this initial media plan that's like this much percent in these places. And then we have traffickers who go in and manually in a system, put those spots in different places like Monday 8 to 10 on Bravo.

Obviously super efficient, super scalable, really built for the future of television. Our goal with this AWS and NBC in-flight optimization is to really change that right. We wanna transform what's called TV log scheduling to behave more like a digital ad server. So the digital ad server, everything is in real time, right? Those decisions are happening in milliseconds, if not less time than that. And using viewership forecast data as opposed to kind of taking what the client gives us and going from there.

And we wanna be able to automatically move these TV spots around to our audience, are audiences are actually watching so we can deliver their campaigns in full. Television is not at a place right now, linear television, where you can do real time. It's not gonna be real time decision, but the closer we can get to that, it's just gonna drive much more efficiency for us and for our clients.

So where does this fit in? It does not fit into innovation, which is confusing and just pretend that you don't see that pillar name on the slide. This is really about delivery. So it's kind of in between that activation and measurement pillar because this is once you've built your media plan, you wanna follow that audience where they are. And so this is where AWS is coming in.

So this AWS powered log scheduler places TV spots more intelligently to meet those audience impression goals. So if you look at CNBC, Bravo, E, Golf, I mean, you see all the networks, we wanna be able to fluidly move across those networks to reach the audience where they are as opposed to x percent on Bravo in this day part, x percent during Below Deck.

Um we really want that to be much more fluid, much more like how digital content and advertising delivery happens. Try not to break it again.

So our partnership with AWS, there's some really strategic opportunities here that are gonna drive these really important outcomes for us. So it starts with reducing liability. Liability is a term that I found really confusing when I joined a media company. I was like, are we lawyers now? Um but this is where we under deliver campaigns and we owe clients something right? Audience deficiency units or ADUs, we owe them, make goods, we owe them money back or we owe them units to take place where we under deliver. And we really want to reduce that.

We also want to increase advertising revenue by really optimizing these campaigns. We are an ad sales organization after all. We wanna reduce manual operations the way that ADUs and make goods and even just trying to change spot schedules in flight right now happens is through emails and phone calls and spreadsheets and taking spreadsheets and recoding them into a system requires a ton of approvals on every side it's just not efficient.

Um and we really wanna make this more automated and smarter. And speaking of smarter, we want to improve the quality and accuracy of these placements. So on the outcome side with the in flight optimizer that we've built with the AWS, we're really simplifying how we're doing these jobs and replacing those manual tasks with automation.

We've built a modular adaptable architecture that's going to ensure we're future proofed. As I said, fax machines, not so future proofed, we want more visibility into delivery and where we're track, we want where we're placing our spots, which this is gonna enable. And then again, we want that flexibility, this flexibility is so key to be able to place ads across state parts networks and eventually platforms.

Our focus for this first phase has been linear optimization, which again is that traditional TV. But eventually we want to get to a place where this is fluid across screens. So we've intentionally designed this to support growing and evolving needs for NBCU. Again, it's modular. We're talking about, you know, loosely coupled microservices as the foundation of the architecture makes it more resilient. It's more flexible.

It allows for us to really optimize technical solutions for specific workloads. We're using data science. I would argue, you know, the smarter the system can become the more it learns the better those placements are going to be. I don't know why I would argue that it's because I'm a lawyer, right? We're talking about liability and it's flexible and it's scalable.

We need to be able to grow how we do this across day parts, across networks across platforms. And this solution has really been architected to ensure that those capabilities are gonna be possible so you can see the foundation of all of this is AWS from start to finish. And we've used both AWS as a foundation as well as open source solutions where it's made sense. The goal really being the right tool for the job.

Um and all of this is auto izing, izing. That's a new word optim and izing. That's what we're doing here at NBC Universal. And you can see down on the NBC one platform how foundational to our tech stack AWS is.

So our goal moving forward, we're in a crawl phase right now. These are creepy icons that I regret selecting. Um, but in the crawl face right now, which really is replacing that legacy TV technology with modern tech that understands television rules and can be more ad server like in its decision, this is focused on the traditional linear television model with demographics 2024.

Our goal really is to make this complimentary to data driven linear media plans by leveraging AI to really predict where those audiences are gonna be so that we can place spots in the right pods to achieve our impressions goals. I said I would explain data driven linear television. I know you've been waiting with bated breath. You've been like, what is it? Tell me more?

So as I said, this way of buying has been very manual and it's been very prescriptive from our agencies and clients with data driven linear television, which NBC Universal was actually the first to bring to market back in 2016, I joined in October of 2015. I'm not seeing, I'm not saying that there's causation there. But is there correlation to modernizing television? Maybe you're welcome.

Um but seriously, it's taking true audiences. So behavioral data, psychographic data, applying that to viewership data and the infrastructure of this, I cannot emphasize enough is S3 and really AWS is the foundation for our data architecture. And we're able to match those consumer audiences to television viewership data.

And instead of having a schedule that's predetermined, we find where those two audiences match and we predict where those audiences are gonna watch television and auto generate a media plan as opposed to having a video plan that's been provided to us by a client. So this was revolutionary and kind of still is a lot of television is still transacted based on traditional demos.

So our goal, the way we do that right now is it's all pre flight optimization. So it's before the campaign actually starts running. And there's not a lot of in flight optimization right now again, because clients get really attached to schedules and it requires a lot of approvals to move things around. But we wanna make those schedules smarter, we wanna be able to do work in between when the campaign starts running and when it ends.

And by work, i mean, we want processes that are automated that will find that audience wherever it is and be really fluid across all of the television properties. And then 2025 and forward will be in our run phase, which is both about shifting spots across tv to achieve campaign goals. And that could be impressions, but maybe ultimately something more all of this idea of run of tv.

So even within linear optimization, clients will give us guard rails of like x percent on each property x percent on each day part, we really wanna be able to run across everything that we have. And so if you've heard of the run of network on streaming, this is like run of everything NBCU has on TV.

And then ultimately to shift impressions across platforms with this idea of one of premium video. So if audiences are shifting from linear to streaming, we follow them to streaming. If they're shifting from streaming to linear, we follow them back. And so that's really the long term goal when we talk about cross platform optimization and it's really leveraging all of the work that we have done with AWS.

And this initial crawl phase took less than six months to get off the ground, which is pretty mind boggling. So that's it for me. I'm gonna hand it to Eddie Lee, who's gonna walk you through AWS Clean Rooms.

Alright, thank you, Jess, son. Um so I'm gonna talk through the second example of where we've partnered with Amazon and AWS for building out one platform, specifically what we're gonna talk about here is how we enable the data collaboration between Amazon and NBCU leveraging both of our respective first party data and in order to enhance audience targeting and maximize media effectiveness.

So one of the important things here to point out is we're doing a collaboration through a clean room. So clean room being a way partners can come together and share data in a safe, secure and privacy minded way. Um so making sure that there are the right privacy hooks that you know, you can't identify a user with too much, too much of a small query set.

Um being able to keep data where it is not copying data. So this is an important concept that really is using AWS clean rooms to power that this is a concept in how we share data with our partners.

So how does this fit into one platform? So you see the two squiggy boxes and there's two pillars where it comes in one is data, identity and identity and measurement. Um so we have this product as a part of NBC unified, we call NBC unified access. And so this is how we share and enable data from NBC unified um to our clients and partners in order to do profiling to do um kind of further insight analysis, do measurement.

Um this is how we share data that way. Uh but what we did was build an integration with Amazon publisher cloud in order to interact with Amazon ad signals and that's how we connect in to do um what I'll talk about later on.

So uh we first approached this as a POC. So what, what specifically was this POC? What we were doing was combining Amazon ad signals and our first party Peacock data via the Amazon publisher cloud. Um in order to look at an index, um what what viewership um signals do we have with Amazon's that really found the the sort of what users were watching against what we are really looking for.

So as an example, you can imagine a signal of, you know, Amazon shoppers that are interested in home appliances and if we can overlay and index that with our viewership behavior, and maybe we find that they over index on action genres, then we can then target those action genre viewers to give them a better chance of finding those users.

So take a step back. Um what is Amazon publisher cloud? So Amazon publisher cloud is a capability um where we as publishers can go interact and query and and collaborate with Amazon signals using our data. So you see on the left side, those are bring in the the Amazon data and the publisher data, bring it into um a clean room. And so Amazon publisher cloud is powered by AWS clean rooms.

And so in that clean room environment, being able to do that data collaboration in order to do things like planning activation and measurement. And so the the clean rooms itself um are really what you know, provide you that security keeping data where it is making it easy. If you have data already in S3, you can make that available um having the privacy controls in there as well.

So how did this workflow work for us? So we start on the left side. So we have our data with an NBC five. We have that now in NBC to provide access and how we make that available. So imagine we query segments on our side, we can query those segments of, let's say action genre watchers, we query those segments, put them in that, get the hash identifiers, load them into S3.

Um once they're in S3, we'd register them in a Glue catalog and then it can be made available in AWS clean room. So it's all very quick and easy, especially if you have your data already in S3. Um and then you get into the Amazon publisher cloud. So now our data is there, Amazon's data is on the other side in the publisher cloud, then be able to do this querying and analysis to get to that indexing.

So in our case for this POC, we had um the Amazon ad signals on on who are really the Amazon shoppers that were chasing for this um this campaign and then finding that they over indexed on certain genres. So you see what the output was, we find out what those audiences were. And then on our side, in Freewheel, we can target those audiences to give us a better shot at, at those users.

And then from Freewheel, we can share that impression data back into the Amazon publisher cloud so that we can do measurement and see how that campaign performed.

So how did that do? Well? Uh we did that match using hashed emails um with Amazon. Uh we did that indexing and found that action fans and documentary fans actually indexed highest um for the the audiences that we're looking for. And then we went ahead and targeted those on Peacock.

Well, so we split the budget in two. So one just being regular runner network. On the second side, we actually targeted those action fans and documentary fans as targeted segments. And when we compare the results, we saw that the targeted users yielded a 3.6 times more reach as well as a 67% lower target CPM. So it was, it was a success and something that now we can take forward with this integration.

That's all. Thank you.

Right. Thank you NBC. Um I, I think Jess and Eddie were also being quite humble to themselves. Both of these innovations that were shared today are industry leading and we're super grateful for the partnership, as mentioned earlier, monetization is only one of the pillars that us at AWS are invested in in the media and entertainment space here.

You'll see recommended, recommended sessions, sorry through the course of the rest of the week. Hopefully you can enjoy the next 30 minutes walking somewhere across Vegas to your next session. Um we really appreciate you taking the time today if you could also kindly fill in the survey Eddie dresses and I will also be hanging around for the next 15 to 20 minutes before our next meeting. Uh if there's any additional questions, we certainly welcome them as well. Thank you again."