Enhance Your Strategy with TV+ Planning Insights!

Get free access to advanced audience insights, benchmark competitors, and uncover new opportunities. Revolutionize your TV advertising now!

Doing TV Like It's Digital

Mary Grace Scully
Mary Grace Scully
Published: Aug. 25, 2020

The advertising world has undergone abrupt evolution throughout the digital age, as have the needs of marketers. This change has led to an intensified focus on performance and growth. In this presentation Making TV Work Like It’s Digital, Simulmedia’s Chief Data Scientist and SVP of Marketing explain how digital has become so successful as a performance driver, and how TV, despite its historically slow-to-adapt nature, can capitalize on these same digital methodologies to deliver better performance, growth, scale, and optimization all through automation.

In summary, Matt & Nico cover:

  • TV Perceptions vs. Reality: There’s a lot of confusion over what TV does and doesn’t do for advertisers. Oftentimes, it is described as too expensive to run, too expensive to create advertisements for, and too difficult to measure. Watch (or read below) to learn how Matt dispels all these rumors.
  • Where do the misinformed perceptions about TV come from? Matt discusses the precedent that old, outdated TV practices have set for current thinking about TV.
  • Next, Matt explains how proven, performance-driven digital tactics can be adapted to the TV environment to boost success. Similarly to Facebook’s optimized CPMs (or oCPM), digitization of TV can make it reliable, cost effective, and most importantly, a growth vehicle for any business.
  • Then, Nico, Chief Data Scientist at Simulmedia, explains the science behind why these tactics work, and how all the evidence suggests that, through tests we’ve run with Simulmedia clients, TV is the best choice out there for many advertisers.
  • In closing, Matt provides marketers with a checklist to consult for driving any growth through a digital approach to TV.

Here is the full conversation. Download the presentation deck.

Full Transcript

Matt Collins: Hi everybody. My name is Matt Collins and I am the senior vice president of marketing here at Simulmedia. And I'm so pleased today to be joined by my colleague, Nico Ricci, chief data scientist for Simulmedia. And today we're going to talk about doing television like it's digital. And it's been such an exciting last couple of months in the world of television advertising, as we've seen a once in a generation phenomenon of viewership spiking in all likelihood as a result of stay at home requirements related to the pandemic, and at the same time, a significant drop in pricing of television advertising. You combine those two things, an increase in viewership, and a drop in pricing, and there has never been a more affordable and accessible time for brands to get on the most powerful advertising vehicle for conversion, and that's television advertising.

Summary of What We’ll Discuss

Matt: So by the time you're done today with our session, you're going to come away with the following. First is a very clear understanding that in fact television advertising can work at the speed and precision and accountability of digital if you are equipped with the right tool set, which we'll get into. I hope you'll also come away with the notion that it's never been more accessible than it is today to be able to get on television, especially when you consider the cost of media and the cost of creative. There is, and has been of course, approaches and solutions available for performance advertisers on television, however, those existing traditional methods for performance on TV have brought some problems that I think experienced marketers are acutely aware of, but there now is a better approach, one that is, I think, familiar in many ways to other tools that they use on Facebook, which we'll talk about.

And then finally, we are going to talk a bit about the math and that's certainly where we're going to rely heavily on Nico and his expertise to talk about why performance marketers should be cautious about simply buying the cheapest possible CPMs, and in fact, optimizing for performance can result in a very different outcome. And then we'll make sure you all know how to get in touch with us afterwards if you have questions.

TV’s Perception vs. Reality

Matt: So let's dig into the perception versus reality of TV. Perceptions have had a long time to build, television advertising is a seven decade old medium, the first ad ran in a New York Giants baseball game in 1941 for Boulevard watches, and in the last several decades, certain perceptions have grown. The first being that, boy is media expensive, and even just to launch a test campaign, it's prohibited. When in fact that's false, it's now possible to get on your first TV campaign and get meaningful signal for as little as $25,000 for that initial campaign, very much in line with the testing that advertisers are accustomed to in other media.

A few TV advertising perceptions and what the actual reality is.

The creative, it's too expensive, it takes too long to make. Well that's false too, there are now a whole breed of agencies that have cropped up to meet a new wave of advertisers that build affordable creative that is built to drive both brand awareness and performance and is ready to test right out of the box. TV can be hard to measure, you either feel it or you don't. That's been a long longtime perception, when in fact TV is now measurable at virtually every stage of the funnel and for every outcome that you can imagine.

Why the Outdated Perceptions?

Matt: Why do we have these outdated perceptions? Well, as I mentioned, TV is a now seven decade old media, but behind that, there are actually several attributes that have contributed to these perceptions. First of all, TV is and was supply constrained. This is so important because it separates television from other media choices. Television almost always sells out and it almost always sells out days, weeks, and even months in advance, of an ad airing, as opposed to digital, which is demand constraint. This is why we have technology to enable near real-time, nanosecond delivery of ads into digital inventory. TV has not had that particular dynamic, and so supply constraint combined with the fact that not long ago, one single prime time spot bought on one of the major networks could reach essentially everyone. And that resulted in a significant price, a very steep price to get into the media and to get into television, that made it prohibited for all but the biggest brands.

And then when it came to measurement, look decades ago, there were no digital options, you were left to measure the impact on retail, which might require getting data back from a retailer, or more commonly, looking really only at changes in brand awareness. So top, top of funnel changes in how people perceive the brand. And of course, creative ran through expensive agencies.

Digitalization is Revolutionizing TV Advertising

Matt: But today digital has transformed television into truly a digital media of its own. It is possible to create custom audiences based on an advertiser's best customers. It is possible to very rapidly measure hard business metrics and their relationship to ad exposure. The barriers to entry have never been lower, there are providers that are automating execution so we don't have to rely on phone calls or emails or spreadsheets, but in fact, we're using software to plug in, and the part where I'm really excited, and actually the first moment here where I'm going to bring in my colleague, Nico, is on AI and the use of machine learning and artificial intelligence to help us with planning.

A collection of advancements in the TV advertising space that is making it more similar to digital advertising.

Nico, at first glance some people may look at AI, television advertising, boy that seems like you're bringing a cannon to a knife fight. Why is it that artificial intelligence is so important for planning and executing TV today?

Nico Ricci: Yes, Matt, that's a great analogy by the way. Yeah, actually I would argue that AI, it's even more important in TV than it would be in other places. And that is because TV fundamentally is a different type of environment than digital, for example. In TV, you're making decisions now to buy media that is going to run at some point in the future, it's a futures market. And so with that, your ability to do two things well is really important. One, you have to be able to predict the worth of this inventory unit that you're purchasing today and is going to air at some point in the future. And predicted both in terms of the performance that it may drive for your digital website or app installs, but also predict how well you're going to be targeting an audience with the campaign that you're buying today. So this ability to predict is really fundamental. And that's where a lot of AI based machine learning techniques are used and can be used to actually predict the worth of units that way.

And secondarily, once you are able to assess the worth of these inventory units that you are buying, it's really important to be able to pick the right set of units. And you do that when you have thousands and thousands of units that you can pick from, you have to employ some optimization algorithm in order to actually decide what's the best, the most optimal set of inventory units for you to go with.

Matt: And certainly the need here has never been greater because unlike 40 years ago, when a single spot on prime time on a network could reach everyone in the country essentially, the average American cable subscriber today gets over 200 channels, and that results in massive audience fragmentation. So Nico, the average advertiser today compared to 40 years ago, how many options are there for them to build a media plan, given the number of channels that are available?

Nico: Oh, a lot, a real lot, also considering buying multiple spots in specific inventory. So if you think about the fact that you have hundreds of networks and seven days of the week, and 24 hours in a day, and you can be really precise in terms of where you actually want your media to run, that is thousands and thousands of possibilities, just in terms of picking the right network, day, hour combination that you want your media to air, and then you want to actually add some frequency to that, and decide how many times you want to air in that specific time. And it really becomes a task impossible for our limited minds to actually carry out.

Matt: Yeah, we have studied this, and when you look at running a two week campaign, given all the options you just mentioned, Nico, that actually the number of permutations of different possible media plans, exceeds the number of atoms in the known universe. So this is why machines are perfectly suited for picking the ideal plan.

Traditional Approaches to Performance on TV Fall Short

Matt: So let's talk about performance. The first direct response, what we think of as a performance TV ad aired in 1984, I believe for Herbalife. It was the first time that consumers were able to directly contact a company to make a purchase. And ever since then, performance advertisers have often gone to this type of television inventory to anchor their plans, to really build out their performance plans. And it's worked very well for so many advertisers to get on TV and to start experimenting and learning. However, as advertisers get accustomed to these traditional methods of performance on TV, they often encounter obstacles to plans that are really anchored and deeply founded on this particular inventory type.

And they include limited targeting. The networks are really in charge of deciding when a spot will air. You could say, "We want to air in the afternoons." They could come back and say, "Well, it'll be Friday, or it could be Tuesday." And you really have very limited control. Similarly, the networks are in a position to cancel at anytime. Now, the advertiser oftentimes holds that option as well, but with the cancelable rights by the networks, what means is that it's hard to predict, week to week, month to month, what's going to actually clear and what won't, which makes it much, much more difficult to optimize.

It doesn't scale into every inventory type. So there are certain slots, prime time is one, sports premium programming is another, where it's just hard to find these spots that traditionally performance marketers and vendors, traditional performance vendors, have provided. And the pricing, it's actually not always the case that this inventory is less expensive than inventory that's guaranteed to air.

Better Way to Drive Growth on TV - Similar to oCPM on Facebook

Matt: So the good news is that there are now methods that address this. That take on some of these problems head on and give performance marketers an option to the traditional methods, either to augment those methods, or to replace them altogether. And as I was thinking this through and talking it over with Nico, we came to this conclusion that what we're prescribing here is very similar to optimize CPM bidding, or oCPM bidding on Facebook, which many of you will be familiar with. And that's primarily because it is possible now to optimize a plan based on predictions for what ad units are actually going to convert the best. And because of the nature of the way that they're bought, they can be bought with reliability, meaning that the advertiser knows that the ad will air, and that makes the targeting so much more precise, you can really zero in on the best performing ad units and have some comfort in knowing that they actually will air as planned.

This depends on AI, as Nico just walked us through, cannot be canceled, it can often be more expensive than just a traditional straight up CPM buy, but just as with oCPM, you typically see that premium come through on the other side.

The Math Behind Measurement and Optimization

Matt: So look, there's a lot behind the surface of doing television like it's digital. And there is really some key milestones that involve, as we mentioned here, some math, some math that's accessible, but nevertheless, important for people to understand the quantitative aspects to this. And so, Nico, why don't you walk us through these next slides. You point out really that there are three key stages for an advertiser that's running TV like it's digital to get the very best out of the medium. Why don't you walk us through them, starting with measurement.

Nico: Yeah, and these three key stages are really involved in any decision about a future course of action, you can really generalize them. There are about one, measuring the worth of something, and in our case, it's measuring the worth of the spots that have aired in your previous campaigns. Two, is the ability to predict based on those measurements, how likely a specific thing that you're going to buy in the future, in this case an inventory unit on TV, is going to deliver value to you. And three, is the decisioning layer, the optimization layer, where you were going to basically pick the right set of units in this case to actually maximize the client's objectives.

Line chart showing how lift in upper funnel digital activity from TV advertising is measured

So, let's start from measurement where there are definitely differences between TV and digital when it comes to measurement, it cannot be otherwise. As we know, digital, a lot of it is deterministic, you can track people along, on TV, you cannot necessarily do that, not at the scale that digital can, but there are definitely techniques to actually get similar signal from the data that you have at your disposal that is going to eventually inform your predictions and then optimizations. So for example here what you see in the screen is a typical technique that is used in TV for measuring the impact of a spot that is airing, for example, in your campaign, in terms of some digital activity. So in this spot, we're seeing the black line, which is the responses, which is the visits to a client's website, and overlaid underneath them are the spots, those circles, those are spots that air.

And you can see that these spots do generate. When you calculate it, you have to make sure you calculate at a dynamic baseline locally to make sure that we account for baseline behavior, but then you can actually see that these spots do generate spikes in activity to your website, in this case, we're looking at visits to a client's website. And you can then, of course, attribute what is in this case this yellow shaded area that you can see from the spot running on ION at 2:47 AM, that generated 66 attributed visits to a client's website. And so you can see how if you run a first campaign with quite a few spots, you can start gathering information about the ones that worked the best and the ones that didn't.

So if we go to the next slide, then the second step of course, is that one of leveraging this information, and predicting the worth of a specific inventory unit that you can buy in the future. So when, as I was saying earlier, TV is a futures market. You're making decisions today for spots that are going to air at some point in the future. So you want to be able to predict the worth of these spots, and you can do that, leveraging the data that is coming from the measurement in a year. And so for a given unit that is available, so let's say that you have 10,000 units available next week, and you have to pick from, you can actually assess the worth of all of them in terms of the amount of visits they're actually going to potentially drive to your website, cost effectively. And you can actually compare that also against how effective that unit is in terms of its CPM efficiency when it comes to actually reaching an audience that you might be interested in.

Visuals showing the process of predicting performance lift down to individual spot unit level and optimizing inventory based on campaign goal.

So you can imagine how now you can have two different objectives as an advertiser, when you're putting together a media plan. One is reaching an audience cost effectively and minimizing the CPM against the audience, and another one, on the other end of the spectrum, is that one of driving as many visits as possible to my website, or as many app installs as possible. And so there's no one media plan that actually will accomplish both things simultaneously, and you can see here in this graph to the right, how as you start changing your preferences, over how much you care about driving digital performance, your efficient media plan can change. So you have this set of efficient media plans, which is this blue dots here, that are all optimal, depending on how much you care about CPV, or how much you care about CPM. So this is the CPV versus CPM trade off that actually starts to arise.

And if you're an advertiser that only cares about brand awareness, then you're likely to end up with the plan that's in the upper left part of this graph in blue, the one that actually minimizes the CPM against the audience that you want to reach, but then you also think that as you start caring more about performance, you can start traversing this efficient frontier and find hybrid plans are probably the best trade off in terms of CPV and CPM, to finally end up on a performance only plan, one that does not care about CPM, and in fact, you're going to lose a lot of CPM efficiency in exchange for CPV efficiency, which is driving visits to your website as efficiently as possible.

How Plans Vary By Objective: Performance vs. Brand

Matt: We have some examples of that actually in the next slide here where we've blinded these results from an actual set of campaigns that we run, but help us understand what these show and specifically the distinction Nico, between the orange and the blue in each chart.

Nico: Yeah. So here we're seeing an actual efficient frontier materializing with real data. This is from, again, data that comes from one of our advertisers and it's of course anonymized and blinded, but you can see how, as you change your preferences, as you go from the awareness plan to the performance plan, how the anatomy of the plan changes, and you can see that in those plots, those three plots on the right, and the blue dots represent all the units that are available for you to choose from in the market, in this specific example, it was for a week long period of time, and the orange units are the ones that are selected, and you can clearly see how the mechanics of the optimization changes, the selection changes, as you go from awareness to performance.

Visuals showing how TV media plans vary based on campaign objective - whether TV advertising goal is awareness, performance, or a hybrid.

And you can see that in the awareness case, there's a horizontal line pushing down to get you the best units in terms of CPM for specific audience that you have in mind, as you go to the hybrid, you see this curve that pushes you to the lower left corner of this CPM/CPV trade off and gets you units that are the best of both worlds, basically. And then if you don't care about CPM at all, and you just go with performance, you get this vertical line that pushes you to the left and gets you units that are the most efficient in terms of CPV.

Matt: It's super interesting to see this all illustrated Nico. And I think it comes to though, to a point, and we see this all the time come up in conversations with advertisers who are accustomed to these dynamics, is this notion of understanding what are you optimizing for. What do you hope to get out of your campaign? Are you going for brand awareness where reach and efficient CPM matters, or performance and efficient CPVs?

A CPM vs. CPV Trade-off

Matt: And here we have, again, coming from an existing client data set that we've blinded, something that really clearly encapsulates what these trade offs mean in terms of CPM versus CPV, why don't you walk us through what this is telling us.

Nico: Of course, yes. I mean, and this is just another example, a tabular example of what we saw in the previous slide, but this is again, we had a very successful first campaign for one of our clients, and in the second campaign, we're able to get enough data to start looking at this trade-off really materialize. And here you can see how when you go from a brand awareness to hybrid to performance campaign, of course your CPM becomes less efficient, and you go from a $19 CPM all the way up to a $29 CPM for the performance campaign. But on the other hand, you can see how the estimated performance, the CPV, the estimated CPV also moves in the other direction, and that's where the trade-off really comes from. And you can see how the performance plan is many times as efficient as a brand awareness plan when it comes to performance and delivering a low CPV.

Table showing how impressions, CPM, reach, and estimated performance changes for a TV advertising campaign based on different campaign goals.

Matt: And for me, the key take out of this is, be cautious as a marketer of anyone who would simply be selling you primarily on a cheap CPM. It's possible that a low cost CPM or a CPM that meets your cost tolerances can also deliver a strong cost per site visit or cost per install. But, and quite often the case, as I think Nico you've demonstrated here very persuasively, optimizing for low cost inventory, low cost impressions, oftentimes means making sacrifices on the cost per the conversion event that you are trying to go for. And I think we've seen that in the data here.

Nico: And actually, interestingly, one more point that you could take home from this is that, and to your point here is that we have seen that if you compare the performance campaign to the brand awareness campaign in this specific case, you'll see that, for example, late night is not as present in the performance campaign, and weekends are more present in the performance campaign. And that is coming from the fact that in the previous campaigns folks reacted to these specific day parts and day weeks much more when it comes to actually digital performance. And so that actually has been included in the campaign itself. So, yeah, that is to your point.

Doing TV Like It’s Digital - A Checklist

Matt: So we're just about to wrap up here. We wanted to leave folks with a notion of a checklist. So as you may be considering television advertising, perhaps for the first time, or perhaps you've been in television advertising, and considering what else might be available, we prepared this, just this brief notion of, hey, what are the things to look out for?

The first is can they offer up a solution that works like this oCPM, optimized CPM approach on Facebook? Is their automated buying that is integrated deeply into the networks? Why does that matter? Well, boy, does it accelerate the purposes of planning and then buying when you don't have to rely on a phone call or email to get that done. Is there artificial intelligence involved to sort through all the myriad permutations of possible plans that are themselves delivering different levels of performance, depending on your campaign objective, but are the right tools in place to select the best possible plan?

A checklist showing how you can make TV advertising work with the speed and precision of digital.

Can the vendor measure appropriately my campaign objectives and link that convincingly back to the ad exposure? I like to talk about it in terms of, can you get measurement that's strong enough to persuade your CFO? Because if you do, then you're probably finding the right solution. Can you build a custom audience based on your best customers, and not rely on age, gender demos in television? And then finally, does that provider have an exploitation and an exploration strategy? I want to give you Nico the last word on this, because it seems to me that even performance marketers have to be mindful of the long term. Why are both of these things so important?

Nico: Absolutely. Longterm is really important. What we've gone through here is really the mechanics of how to make a decision about a media plan, but thinking about how to get your brand to be successful longterm, how to have a successful series of campaigns, one after the other, that will get you to the best place possible is really important. It's an important conversation to have, because it varies, it's not a one size fits all, it varies from client to client, it's important to have that conversation.

And for example, the exploration, exploitation strategy here is, usually what we do for clients that are always on and are performance driven, it's important we use reinforcement learning to actually dictate the amount of exploration and amount of expectation you want to do. And when I say exploration, I mean, units that we want to try and find and figure out whether or not they are performing for you, continuing to find new units that are performing for you. And exploitation is basically continuing to actually buy units that have been performing for you in the past. And so having these meta strategies that will get you to the most successful place in the long term, and having that conversation early on is really important.

Matt: So we've thrown a lot at you today, we really appreciate you watching. If you have any questions, we've included our emails and contact information here, please get in touch, we'd love to hear from you. We hope you're holding up well, and again, thanks so much for joining myself and my colleague, Nico, for this presentation. We are Simulmedia and we hope to see you soon.