The drums are rumbling legislation warnings. Last week three states proposed bills that would illegalize the best deterministic targeting the advertising industry has ever had the chance to use. The exception being advertiser first-party IDs, which tend to be existing and fallen customers not suitable for acquisition, which is the key to brand growth as Byron Sharp and Peter Drucker have taught. The closest one would be re-acquisition of lost users.
In terms of being able to analyze results, the proposed legislation also prohibits matching the first-party data with any other data. This would block or water down most ROI attribution methods in use today, although advertisers who can read the brand website visits and conversions themselves would not be blocked from doing so. The CPG, QSR and car advertisers for example who depend upon third-party data to measure sales effects would be thwarted, except for random control "Intent to Treat" trials wherein the ID buying the product can be traced back to the media mix cell it was randomly assigned to.
ARF RCT-21 uses addressable TV as a proxy for linear in order to be able to do random control trials in linear. Those who feel uneasy about such proxying can default to matched market trials as in the old days.
The bigger problem is the quality of the targeting. It's ironic that as more and more TV becomes addressable, the ability to get maximum value out of the addressability appears about to go away.
Charlie Fiordalis was the first guy who said to me, "We need to come up with great new ways to do context targeting, because privacy legislation is coming," and that’s when his agency Mediacom began to extensively analyze RMT Ad-Context Resonance.
Context Targeting Can Pack a Wallop
I was excited when, at TRA, focusing on aggregate level analysis of matched deterministic third-party data sets when using context targeting of heavy swing purchasers, we were able to average +28% increases in ROI.
I was even more excited when, at RMT, using context targeting to select contexts psychologically resonant with the ad, we were able to get +36% increases in ROI and +62% increases in first-brand mention using context targeting (Nielsen NCS*and605 respectively)
Context targeting is not to be sneezed at. It isn't an old-fashioned thing, not the way we do it. It's just as cutting edge as any form of deterministic ID targeting (which we also do, using the ad's resonance with the person, Neustar finding +95% increase in ROI in the most recent campaign).
The Reality Is, Even Today One Must Combine ID and Context Targeting
Semasio went from 268 million to 168 million IDs available in the U.S. as result of Apple's final third-party cookie deprecation. Swiftly adapting, RMT and Semasio use both ID and context (website) targeting so that if the programmatic bidder sees a match with the ad in either the ID list or the URL list it knows to make a bid.
Note that the bidder should not require double qualification that both the ID and the URL resonate with the ad, as that would be too limiting to reach. If either qualifies the impression will be high quality.
Data Driven Linear (DDL) On the Rise
Advertiser Perceptions' latest survey finds that 58% of advertisers are planning in 2022 to increase their spending in Data Driven Linear (DDL). That was the highest score of any video category. Even CTV/OTT, which was the belle of the ball in 2021, comes in second to DDL this year. Many brands have found that both media types can get great ROI and branding results, and DDL may have been bypassed as not as shiny an object as before.
DDL, in addition to many wonderful specialist media such as A4 and Freewheel, also includes all the MVPD selling organizations and practically all of the television networks. The way most of DDL works is with advanced audiences, often through OpenAP, most frequently with lookalike audiences supplied by Nielsen or Comscore. Sometimes these audiences are even deterministic. Although DDL is not addressable, it is in fact contextual targeting. IDs can be pre-analyzed to find best context matches with the ID list.
The deterministic ID lists tend to have double-digit lifts of KPIs, whereas our experience with lookalikes is single-digit lifts. The deterministic IDs cost more but more than pay back that incremental cost.
I am an advertiser myself (selling my books), and I get the best results by combining targetingof known category purchasers and RMT resonance of IDs and contexts with my ads. More brand-new readers of my books come in through the resonance IDs/contexts than from the known purchaser IDs, as might be expected.
The legislation might not come, or it might be rationally adjusted in the political process. One way or the other, the cookie deprecation is sufficient reason to shift to a hybrid ID/context approach as soon as practical.
*Nielsen NCS read the post-evaluation results weighting by GRPs which skewed to low resonance, so their report accurately said +23% not +36%. Also, +36% is what we calculated by using straight averaging to find out what it would have been had the GRPs not been skewed to low resonance.
Click the social buttons to share this content with your friends and colleagues.
The opinions and points of view expressed in this content are exclusively the views of the author and/or subject(s) and do not necessarily represent the views of MediaVillage.com/MyersBizNet, Inc. management or associated writers.