Today, we have measures of ROAS that not everyone believes. I have recently completed a study with the Coalition for Innovative Media Measurement (CIMM) on Market Mix Modelling (MMM), which found that brands place greatest trust in MMM, and least trust in value-add survey-based brand lift studies.
This doesn’t mean that brand executives uniformly take MMM to be totally accurate, but they need some compass for broad media budget allocation decisions, and they feel that MMM is the best they’ve got. Their suspicion of value-add brand lift studies is caused by the perception that they always show the media platform paying for the study in a positive light.
This situation is a big improvement over where things stood for most of the 20th Century, where direct response advertising aside, most practitioners believed that measuring the sales effect of advertising was impossible.
Singlesource arose as a method in the 70s-80s, and Mars, for one, used it to great advantage, publishing the fact that they had tripled their ROAS by using it in England, France and Germany. These were small panels, typically under 5000 homes, such that only big brands could see statistically significant results, and therefore the method was not going to last. P&G, Unilever and several other advertisers made a valiant attempt with Project Apollo which lasted only a year.
In 2005, my colleagues and I – a multinational team – introduced TRA in the U.S., which used naturally occurring data (set top box, frequent shopper card, auto registration, etc.) to do singlesource on millions of homes. This method took hold and digital emulated it and called it MultiTouch Attribution (MTA). TRA added digital to TV starting in 2007, and from the beginning had licensed IRI price and promotion data so that as many marketing stimuli as possible would be included. Today, many companies are using big data singlesource.
Most of the leading brands use MMM for broad media budget allocations and big data singlesource for tactical decision making including allocations among specific media platforms, pulling dud creative off the air asap, targeting specific types of purchasers, maximizing reach and minimizing excess frequency, and so on.
A third ROAS method in use is in-market experiments. The most rigorous type are Random Control Trials (RCTs) in which a single variable is isolated as the only difference between a Treatment group and a Control group. This is true science. This can prove causality, not just correlation. RCTs have always been fiendishly hard to execute and the traditional media have not done anything to make them easier. One of the virtues of big digital has been the facilitation of using this most credible technique.
One of the conclusions of my new CIMM study is that the time has come for the media that have reason to believe that they have the greater sales and branding effect per dollar, to make it easier to do RCTs or some type of experimental design using their media platforms. This is most achievable using addressable TV (MVPD type or streaming type or ATSC 3.0 type). Linear can improve the use of network cut-ins by making them easier to set up, more foolproof, automated, with clean room capabilities for matching with sales data, creating matched market tests which are programmatic. Cable with its zones can do this in spades. Addressable can bring it down to zips or homes.
This is not just wishful thinking. Some brands are already doing matched market trials without help from the media. The companies which brands trust to do MMM generally also have the expertise to do in-market tests of all kinds. And Rick Bruner operates a company called Central Control which has built much of what is needed and he should be sought out by the media who believe they deserve more credit than they get for sales and branding effect. Measurement companies included in my new CIMM study that have the most experience with field experiments include Nielsen, Circana/IRI, 605, Marketing Attribution, but this leaves out many others with such skills.
In my presentation at the CIMM Summit 2023 on October 11, I recommended that the television networks and MVPDs band together to offer marketers a programmatically easy system to use which enables RCTs and other scientific experiments. These would ideally include all possible marketing stimuli as measured controls including price, promotion, all forms of digital, direct mail, cinema – TRA managed to achieve that which proves it can be done.
However the type of RCT called Intent To Treat (ITT) does not need to measure all that, it simply looks at the media dollars put into a given media type and the incremental sales dollars that come out of that cell. ITT ignores the interim measures and clears away a lot of fog and debate. You can tell the CEO and CFO what they want to know without making them sit through stuff they don’t care about: simply which media type has the higher ROAS.
This new CIMM report contains a total of ten recommendations of which this is only one.
A second important recommendation (which is all that we will have space to cover in this article) is to use all three ROAS methods in an integrated manner: MMM, singlesource, RCT (or other experiment). I think of this as Triangulated ROAS. MMM for strategic decisions, singlesource for tactical decisions, and experiments for verification and reality-adjustment of MMM and singlesource models.
The Marketing Science Institute (MSI), which like CIMM is now part of ARF, has invited me to help set standards for the future of ROAS measurement within their new MMM Initiative. Many brands are already participating in this initiative and MSI invites wide industry participation. If you’re interested in learning more please let MSI know at research@msi.org.
Who benefits from having measures of ROAS that both the buy side and sell side feel are strong enough to regard as actionable currency? Everybody.
Who benefits the most? Hard to say. Could be the brands, who will be able to make decisions that pay off bigtime, probably leading to renewed brand growth and brand love as the school of hard knocks teaches its lessons. Could be the individuals within companies associated with these innovations. Could be the media types that have always proven the strongest in delivering real sales and branding effects. Could be the agencies that step in and build the tech ahead of the others.
The full CIMM report is available to ARF, CIMM and/or MSI members here.
The deck I presented at the CIMM Summit is available to the same folks here.
Posted at MediaVillage through the Thought Leadership self-publishing platform.
Click the social buttons to share this story with colleagues and friends.
The opinions expressed here are the author's views and do not necessarily represent the views of MediaVillage.org/MyersBizNet.