Join the Waitlist for Skybeam: Be First to Access Our Self-Service CTV Tool!

Win big this holiday season: Cutting edge TV ad strategies, AI power, and pro tipsPreview Skybeam's features and sign up to be among the first to use it when it launches.

Take Your Ads to New Heights with Skybeam

Launch TV campaigns that drive results with Skybeam's intuitive, self-serve platform.

Simulmedia Patent Helps Target, Reach, and Create Customers

Kyle Hubert
Kyle Hubert  |  Chief Technology Officer
Published: Feb. 19, 2020

One of the highlights of any career in science or technology - mine included - is the opportunity to work on an innovation that earns a patent. That's because patents uniquely validate and memorialize a project’s value and uniqueness. Today, I’m honored to announce the latest patent we here at Simulmedia have earned and to explain why it’s a very good thing for our clients.

First, I want to establish some context. Buying media that maximizes audience reach at scale is really hard. While these vehicles, e.g. many offline channels, homepage takeovers, etc., can deliver audience breadth, they often lack precision targeting. This type of classic media placement tends to be indiscriminate instead. For this reason, it requires brands and their agencies to negotiate rates across all users, which tends to make the media more expensive. The media buyer will attempt to negotiate access to opportunities where their potential user base is well represented, perhaps with the goal of driving customer volume to the brand’s website or app at the lowest possible cost per visit or download, or maybe just to boost brand awareness. Estimating the impact and value of even one of these activations is difficult. Comparing two or more such campaign opportunities gets even harder because it isn't always clear how to evaluate the relative quality and quantity of their respective users.

Estimating the impact and value of even one of these activations is difficult. Comparing two or more such campaign opportunities gets even harder because it isn't always clear how to evaluate the relative quality and quantity of their respective users.

In addition, these deals are negotiated well in advance of the campaign actually running. The high touch process of getting these campaigns live means that instructions must be finalized and the operational work must be allocated before the campaign starts, sometimes by months. This means that the media buyer must part with most, if not all, of her budget far in advance of potential customers actually interacting with the media, much less determining whether it’s working or not. That’s a big risk.

Mass media works very differently compared to addressable media, which is sometimes referred to as people-based advertising. Addressable depends on tracking, which uses consumer behavior to effectively allocate media budgets while executing campaigns. Ad servers and exchanges make decisions on which brand and creative is served to which individual based on continuous feedback from previous ad delivery and up-to-date consumer behavior. This ad decisioning occurs within a very short time window. For all these reasons, addressable media gives the buyer a level of precision that’s often lacking in broad-based buys.

Ad-supported TV media is representative of the big, broad-based, reach-maximizing buy that creates risk and work for buyers. All viewers of a network will be exposed to brand messaging, the media must be purchased in advance of the campaign airing, and the network needs to finalize the mix of ads and content before the broadcast begins. Some people in the industry have argued that if only the networks overhauled all their operations and modernized, surely this process could work more like addressable media.

[Simulmedia's patent] solves selecting media over a range of inventory with similar or disparate audiences. It allows these selections to be made in advance of the media executing, thereby predicting future behavior.

While that’s an understandable scenario to ponder, the investment this would entail is so large that it isn't clear when or even if the networks would break even. It also overlooks another key distinction: while mass media is supply constrained, addressable media is demand constrained. While the latter relies on technology to remove friction from the buying process in an effort to spur demand, the former nearly always sells out, sometimes months in advance. Applying the same technology that makes addressable media tick to the world of mass media is a solution in search of a problem.

This doesn't mean, however, that the networks can't use technology to make substantial improvements for TV media buyers that would preserve the benefits of mass media without incurring large infrastructure costs.

This is precisely where Simulmedia’s patent steps in. It solves selecting media over a range of inventory with similar or disparate audiences. It allows these selections to be made in advance of the media executing, thereby predicting future behavior. Our invention runs today, without forcing costly upgrades for the networks. Most importantly, by placing it into a single, end-to-end platform, we offer clients the ability to act on the insights the invention provides, allowing them to buy and measure the media our platform recommends. The result: more cost-efficient audience reach, which can deliver the web traffic and app downloads today’s digitally-oriented brand marketers require.

Here’s what the experience is like. When buying media from Simulmedia, first you will be asked for your target audience. The attributes your audience possesses will describe a subset of all viewers watching TV using their individual characteristics. For instance, perhaps adults 18 to 49 years old who own a cat - which describes about 12 percent of all TV viewers in the U.S. - would be interested in your product. Our platform then selects the individual viewers with these attributes from all of the TV universe (though our innovation can work off of any data source). Simulmedia is the only company that uses this audience definition in a disaggregated fashion. By “disaggregated,” I mean that when the platform builds the brand’s target audience each individual person is targeted.

If we have the choice between network/time slot A and network/time slot B, and we have already decided that network/time slot C is best for us to buy, which is the best second choice: A or B?

Using historical viewing patterns from these audience members, our platform determines the probability of every individual viewing each of the television inventory units listed for sale. This is forecasted into the future, using machine learning to predict what TV programs and networks the individual will watch. Because the viewing predictions are based on the Nielsen national universe of programming, it provides a rigorous data-set upon which to build these predictions. It isn’t just about network viewing, either. We can predict how many minutes an audience member will watch for each program. This allows networks to list inventory for sale by program or by time period in our platform, making it easier for advertisers to more precisely select the inventory that will offer the highest concentration of target audience viewership.

The next part gets really interesting. At this point, our platform will carry a list of each network’s inventory, and the probability associated of every target audience member watching that network. Using your brand’s budget restrictions, our platform can calculate all feasible combinations of inventory and pick the media plan that maximizes the most cost-efficient reach against your audience. This is where our invention really shines.

If we have the choice between network/time slot A and network/time slot B, and we have already decided that network/time slot C is best for us to buy, which is the best second choice: A or B? Thanks to our rigor in disaggregating all data, we can check the individual probabilities and detect what happens to the audience reach curve. A way to visualize this is to ponder a deck of playing cards. When playing a game and trying to get a straight, being given a face card the player already has in her hand doesn’t increase her chances of having a straight. If she can acquire a few new unique cards, she’ll have a better chance of getting a straight.

Carrying this analogy forward, Simulmedia can calculate the best chance of a straight against all possible combinations of the cards in the deck. When talking about U.S. based television, that’s more combinations than there are atoms in the known universe. This is serious computational power being brought to bear for your media buying needs! Additionally, this is done in an iterative loop, with the user of the platform being able to evaluate and change her buy while observing the impact on her campaign goals. We believe this unlocks value and is yet another the reason why no one beats Simulmedia when it comes to cost efficient reach.

Using historical viewing patterns from these audience members, our platform determines the probability of every individual viewing each of the television inventory units listed for sale. This is forecasted into the future, using machine learning to predict what TV programs and networks the individual will watch.

After all this work is done, Simulmedia’s invention has now enabled your brand budget to select your strategic target audience, find them in broad-based, non-addressable media, iterate until you’re satisfied, and purchase the plan that maximizes reach of your audience, which is a critical input to creating more customer volume for your site or app. Your campaign will go live, and you can rest easy knowing that you have secured the best media possible at scale.

All of this was made possible by the hard work here at Simulmedia, now publicly recognized by the US Patent system. Read more about Patent US1056782B2 here.