The science of media optimization began to flower in the early 1960s. It had a major impact on my career. My first work on optimizers was done for the agencies I worked at, including Grey, K&E and Interpublic. I later helped build optimizers for TV-radio-magazines at Brand Rating Index, for radio (for Telmar), for addressable commercials at Next Century Media, for spot TV (for McCann), and the first optimizers for Nielsen Respondent Level Data (RLD) for BBDO, Turner, Discovery, USA Network and The Weather Channel.
For decades media optimization was stuck at mere reach optimization. Once TRA introduced the ability to link ROI to specific types of media environments it became possible to optimize ROI and at TRA we built an optimizer for that use case. "My" current technology company RMT improves upon that using the psychological resonance with the specific creative by media context and by audience ID for addressable media. In 2018 I wrote an ARF paper outlining the many variables which can be included in cross platform media and creative ROI optimization. Every few years I update a short history of media optimization.
With all that, I'm always very interested when someone makes serious forward strides in media optimization, especially when it involves outcomes metrics not just audience metrics. Media companies have tried to move into the space of guaranteeing outcomes, but that is a challenging premise because media placement has a small impact on the overall success of a brand. Also, the market has been challenged to apply the same analytic rigor to forecasting outcomes as forecasting audiences. My interest was piqued when I uncovered the work that TelevisaUnivision is doing with a new forecasting and optimization company, datafuelX. This work is led by Brian Lin, Senior Vice President of Product Management & Advertising, Kent Tseng, Vice President - Product Management & Advanced Advertising, and Matt Weinman, Senior Director of Product Management.
TelevisaUnivision obviously plays a very important role in helping advertisers drive successful campaigns with the Hispanic audience. As a way of enhancing that position and responsibility, they brought a new advanced advertising product to market during the 2021/2022 upfront, a product that optimized against outcomes for the targeted brand. Conceptually, the product was structured in a way that controls for the impact that media placement has on the success of a campaign- TelevisiaUnivision focused on the incremental consumer behavior driven by the optimized campaign, compared to a baseline schedule. So TelevisaUnivision therefore was in direct control of what they guaranteed.
In its initial version, TelevisaUnivision leveraged EDO search behavior as the outcome metric to focus on. Search behavior is a solid mid funnel metric, reflecting consideration or engagement. They called this advanced advertising product Search Lift.
TelevisaUnivision then turned to datafuelX to build the analytic structure for the Search Lift product. Howard Shimmel and Dan Aversano, who created one of the best network television media optimizers, joined by Michael Strober and Spencer Lambert of Warner Media, Jay Amato one of the founders and CEO of OpenAP, and other industry heavyweights, have further expanded the optimization envelope in a new company datafuelX. datafuelX provides advanced analytical models to forecast and optimize media plans across platforms, channels, and currencies.
In order to support Search Lift, datafuelX builds forward looking forecasts of search activity(Search Engagement Rate/SER) at a selling title i.e., program level across all Univision networks. SER reflects the responsiveness of a sales rotation across TelevisaUnivision networks, in driving search activity for a targeted brand. Reflected as an index - the average of all TU networks is 100. These forecasts are built using historical EDO data for the brand or category or show, correlating them to Nielsen delivery by demographic for the ads, ingesting Univision's forward looking ratings estimates. Having costs, demographic estimates and SER estimates allows datafuelX to build optimized plans that control for the brand's CPM guarantee but optimize Search Lift. Once the media plans starts airing, datafuelX produces a mid-flight read to guide any in-flight optimization, then provides a post flight read.
TelevisaUnivision has run Search Lift campaigns since 4Q 2021, has campaigns in flight now, and have sold in this capability in this year's Upfront.
TelevisaUnivision has found that Search Engagement Rate (SER) can be increased across verticals with the average increase in SER being over 20% in CPG, and around 10% in cosmetics, insurance and telecom.
datafuelX has a contractual relationship with TU to help them bring new advanced TV/digital products to market. The two organizations share a joint interest in building capabilities to allow media companies to drive more impact for advertisers, from top of funnel metrics to bottom of funnel metrics, using other data sets like website visitation, location and sales, leveraging single source data sets.
In an industry which has not paid enough attention to the importance of developing outcome optimizers for too long, and looking to improve the value that linear TV brings to advertisers, forward-looking networks such as TelevisaUnivision are stepping up to change all that. We applaud their filling the gap in the marketplace and look forward to following the flowering of their work across data sets, media companies and other client types, and funnel levels in the years to come.
"We at TelevisaUnivision strive to be the most comprehensive Spanish language marketing platform, and to do so we understand that it's imperative to drive business outcomes that truly matter to our clients," says Brian Lin. "datafuelX's optimizer product OutcomeX helps us leverage data and analytics to help achieve those outcomes."
Outcomes vs. Demos: When Demos Do Harm
Howard Shimmel of datafuelX mentioned to me that they would be able to get even higher lifts if the client gave up on the sex/age demo. This rang a bell in me because I've been there. First time it was at TRA. The clients wanted to have the optimizer solve for max ROI and do it without increasing the CPM against the sex/age demo. I explained that the ROI increase would be >5X the cash cost of any incremental sex/age CPM. Many minds would not go there. Or possibly they were totally there but had no interest in the tradeoff between ROI (generally treated as feelgood data) and the sex/age CPM (agency contracts involve guarantees involving the agency's own money based on that parameter).
To the advertiser, ROI is much more important than CPM, and is certainly not looked at as feelgood data, it's real money and can be accounted, and while modeling data is soft data, when it lines up with the bank account it earns respect. TRA earned respect the hard way and would have it no other way.
You know when you are in a plane and you're in a big cloudbank, totally surrounded by white, you will be waiting for emergence and waiting, then there will be a little break in the clouds and you'll see a most beautiful blue for less than a second? You know that blue is going to be the coming condition at some point.
In much the same way, I know that the conditions created with 77 big brands when TRA hit its peak, and that confidence in ROI data has kind of slid back a bit today, the day is coming when all the solid ROI data suppliers will be trusted partners, and this time it will be forever.
Which is why it's right for datafuelX to have these frank discussions with practitioners, letting them know that the CPM tradeoff can be included in the optimization and you will see the proof in your own bank account (with the help of outcomes measurement that will be trusted to the degree that it does not defy the bank account).
Heck, at TRA we were able to quantify the degree to which the sex/age target was predictive of ROI. It was eye-opening. Here's the slide from 2011: