The Media Rating Council’s proposed Cross-Media Audience Measurement Standards, especially its plan to implement “duration weighting” in 2021, has produced confusion and some concern among industry stakeholders. The following interview with George Ivie, MRC executive director-CEO, was conducted to set the record straight, clarify the standard, and explain why duration weighting make sense.
Tony Jarvis: Can you provide the current status of the standard (Phase 1 Video) and feedback, and/or issues you’ve encountered in trying to finalize it?
George Ivie: The [standard] sets the foundation for business processes, methods, calculations and disclosures necessary for the measurement of ad and video content across all video platforms and channels -- linear TV, VOD, time-shifted, OTT and streaming video -- on an apples-to-apples basis.
The standard, when implemented by measurement vendors, will enable advertisers, agencies and media organizations to understand unduplicated reach across these platforms and channels, as well as set the processes to manage frequency across these environments -- something that is extremely difficult to do.
The standard addresses measurement granularity, quality of data for audience characteristics, raises the viewable impression pixel threshold to 100%, requires sophisticated invalid traffic filtration for all aspects, measurement of duration (enabling various methods), inclusion of audio, as well as guidance for quality methods of deduplication among various platforms and channels.
MRC is in the process of finalizing the standard after a lengthy public comment period. The standard is substantial, with about 70 pages of content and a significant level of detail that provides value in the form of guidance for sound measurement of cross-media video.
The vast majority of this document is not being debated; essentially having no negative comments from the working group participants or the public. We are now focused on one remaining area requiring attention, where we have introduced a new metric timed to be effective in 2021: "duration weighted viewable impression."
This new metric has received a lot of attention and driven debate among a wide range of industry practitioners. We are working to address concerns and commit ourselves to a process to resolve questions and finalize these requirements by driving as much consensus as possible. Other industry organizations have committed to working with MRC on further research over time, which will help inform the future implementation of duration weighting.
We expect to finalize the Cross-Media Standards in August. It is likely this will contain a provision for “duration weighted viewable impressions,” but this aspect will not be effective until a future date.
We will embark on later phases of the cross-media measurement project, to incorporate display and audio-oriented media into the cross-media guidance in the future. The exact timing is to be determined but definitely only after finalizing the Phase 1 video component.
Jarvis: What does duration weighting cover and what does it not?
Ivie: The duration weighted viewable impression is a counting metric MRC has proposed. It would provide proof-of-performance information on how long the viewable impression conditions persist. Specifically, for the entire ad, or in part thereof, as a ratio to a denominator, either fixed or relative to the ad duration.
“Duration weighted viewable impressions” have nothing to do with whether the user saw the ad, liked the ad, disliked the ad, or any other form of audience or outcome measurement.
The duration weighted viewable impression is being carefully differentiated from outcome metrics in the standards document. We believe duration, including duration weighting, can be used as one of the inputs into outcome metrics. However, this has to be done thoughtfully, since value and duration are not necessarily related on a linear basis. We have seen fairly strong evidence that duration is meaningful to value, but the relationship varies based on a large number of factors: creative quality, placement of brand message, attentiveness, and many others.”
Jarvis: Is duration weighting, as proposed, an unfounded leap based on both Advertising Research Foundation’s 1961 guidelines “Towards Better Media Comparisons” and its 2003 “Making Better Media Decisions,” especially if substantially based on ROI?
Ivie: MRC is not wading into that conceptual framework at the present time. This would be a matter for ARF to consider, rather than MRC.
However, important and relevant to our current cross-media standards draft, a duration weighted viewable impression has nothing to do with ad attentiveness. It’s merely a measure of the extent to which the viewable state -- which is strictly a hardware/browser state -- persisted through the duration of the video advertisement.
Jarvis: Based on the ad industry’s use of the term “impressions” in audience measurement, is the term “viewable impression” the bête noire in these proposed standards?
Ivie: Interesting question. MRC has used the term “impression” in a context that does not include audience or outcome implications, along with IAB, for over a decade. However, you are asking if the user community may be confused simply because that term generally is used in the context of audiences. That may be an issue, and a point we should consider in writing our definitions and making things clear in the written language, although we are not likely to use a different term at this point.
Some on the seller-side continue to consider viewable impressions to be a bête noire, but not MRC. Just a few years ago, the industry was trading on served impressions for digital. We found 40% to 50% of all impressions didn’t even make a single pixel or any amount of time in the viewable port. We believe they should not have been counted, nor serve as an input into outcome/audience metrics.
The viewable impression initiative cleaned that up. Sellers and buyers may still squabble about the parameters (e.g., should it be one pixel for any amount of time, should it be 50% of pixels, 100%, should the time be longer or shorter?), but one cannot ignore the obvious benefits of eliminating advertising waste at the very core.
The use of the term "viewable impression," and the requirement for viewability as an input into cross-media metrics, has not been a point of contention among the vast majority of participants and commenters.
Jarvis: If duration weighting is to be embraced as part of a conditional “proof-of-performance” or “accounting measure” for subsequent audience measurement or payment verification, should a “complete duration” be simply based on the duration specification by the media seller or like the TV industry’s 30-second ad?
Ivie: The draft cross media standard addresses and codifies a metric called a “complete.” However, realistically we know that many -- at times most -- impressions do not have the viewable conditions the persist to completion of the ad. We needed a measure of proof of performance for less than a complete, because those exposures, beyond the base viewable impression threshold, do have meaning and should be counted. We believe they are a viable and important input into downstream outcome metrics, and they have potential value; they should therefore be counted.
The remaining debate around duration weighting, among those who favor it, is whether it should be based on the creative length (relative) or another fixed denominator (absolute). Our public comment draft of the standards proposed an absolute method, based on research that indicated potential bias toward short form may occur with a relative approach. Recent feedback from the working group and the public has emphasized the logic and simplicity of a relative approach. We are considering this when finalizing the overall calculation requirements.
Jarvis: Is the current confusion about duration weighting driven by the “need to remove apparent bias” resulting from very different “viewable impressions” across different screens, in different environments with often different creative treatments?
Ivie: I don’t believe today’s [effectiveness or outcome measures like lift and ROI] are solely reliant on duration as a variable. Most I'm aware of incorporate other types of campaign information beyond duration. Overall, I believe that should be the case -- that duration is only one in a series of inputs that could be used to inform outcome.
However, it is clear that duration and outcome are related in ways difficult to determine and related in ways that depend on a possibly large numbers of factors. Ad agencies have dealt with the relevance, impact, value and contribution of different-sized ads (size in its many dimensions and environments beyond duration!) for an eternity, based on the unique brand marketing requirements, objectives, category position, timing, budgets, efficiencies, reach, etc.
One thing we do believe about outcome metrics. None of them are based on industry standard processes, since no standards exist yet, and just as important, none of them are audited. Don’t forget about that source of “noise.”
For this first Cross-Media Audience Measurement Standards phase strictly for video, I believe we have made a reasonable attempt to create counting metrics that are fair and use consistent viewable impression parameters (now changed to 100% of pixels). For the next phases of cross-media, where we incorporate other media types, such as display and audio, beyond video, I believe your question will be much more relevant and challenging.
Thanks, Tony, for this helpful attempt. I believe that this will be used primarily in digital media---mainly video----buying but I don't see any true comparability with TV. What would be helpful would be several hypothetical examples. Say a 10-second digital video commercial is on-screen for only 5 seconds while a 15-second digital video ad is on-screen for all 15 seconds. Does the longer commercial get credit for triple the audience value?And what about a TV "30", which Nielsen would count only if it was on-screen the full 30 seconds? Does it have double the audience value of the digital "15" and 6 times the value of the 10-second digital ad message?
I'm troubled by the idea of duration weighting for a couple or reasons. Two of them are conceptual.
First, Gross Rating Points have always had two parameters-- GRP is a 2-dimensional metric. One of these dimensions is reach; the other is frequency. If you multiply a campaign's % reach times the average frequency, you get GRPs. (If you reach 70% of the population an averge of 3 times, that's 210 GRPs.) Factoring in another parameter that gets at the quality of the experience of the impression (as duration weighting ends up doing) fundamentlly changes the construct of the GRP. Maybe that's OK and I'm just change-averse. But we've always used GRPs to count, and then other measures to qualify. And within a GRP measure, we've always understood that an impression was reaching one person with one ad one time; are we really prepared to enter into the world of fractional impressions?
Second, as George notes above, there's a non-linear relationship between duration and effectiveness. Should a spot viewed for 15 seconds count as half an mpression as opposed to a spot viewed for 30-seconds? Maybe a disproportionate share of the communication values accrues in the first 5 seconds, and a 15-second view should count 90% as much as a :30; we don't actually know.
It's worth noting that in print planning systems, there's long been the notion of "noticing weights," where planners are free to assign differential factors based on size and placement of the ad. Presumably, conceptually duration weighting is getting at a similar idea, but for time a opposed to space.
I wonder though, if the standard shouldn't simply call for (a) exact-second measurement of ads; and (b) segmentation of reporting by creative duration. That way planners, buyers and sellers would have the pieces necessary to take into account spot length and duration of spot viewed, with the flexibility to determine how best to take these things into account.On this platform, this 30-second spot reached a million people, who viewed an average of 18 seconds of the spot.
(And of course we should remember that all this depends on the ability of the measurement providers to account for viewing time to the second...)
Good points, Josh. I also feel that this type of metric may be confusing media planning with buying. A media planner can and should consider the judged effectiveness of particular ad units ---say "15"s versus "30"s in TV for a given brand's campaign and, based on research, the planner may determine that the shorter message has 80% of the impact value of the longer one----not half, as one might assume by their comparative lengths. This can lead the planner to use "30s" more heavily at the outset of a campaign, then employ "15s" later, as a way to maximize efficiency. However, another planner, working on a different brand may value a "15" at barely half the value of a "30"---again, based on specific research or the intent of the creatives---in which case, the longer messages might account for most of the TV time buys---with "15s" being used only on low CPM cable channels. In other words, a single time based metric, has no particular relevance at the planning level. There are too many non linear variables to consider.
When we get to the buying function, a single metric might make sense providing it represented a generally valid finding---like how many seconds does it take for the basic message to be gotten across. Based on lots of research, it might be determined that most "15s" get their core messqage across in 8 seconds, while most "30s" do the same in 12 seconds. Accordingly, such adjustments might be adopted by some advertisers as modifications of their reach and frequency evaluations. Here, they would be estimating the effective communications reach and average frequency of their ad schedules. The question arises whether the sellers would go along with such generalized "value" assessments? Perhaps, but only if they were accepted by the vast majority of the advertisers and their time buyers.
The basic problem is that we are trying to find some measure of comparability between digital media and "TV" commercials when neither---and especially digital media---has anything approaching a measurement of ad viewing. Perhaps, that issue needs addressing before we worry about making comparisons across platforms.
The observations of Ed and Joshua resonate for me on this matter,
especially after I read this quote from the MRC:
“'Duration weighted viewable impressions' have nothing to do with whether the user saw the ad, liked the ad, disliked the ad, or any other form of audience or outcome measurement."
I have always appreciated and applauded the principled manner in which the MRC,
especially George Ivie, has sought to bring order to chaos.
Nonetheles, has anyone visited the home of a "live" Nielsen Household Panel Member and observed behavior during the viewing and hearing of TV?
It may cause you to ask yourself, "Why are we concerned about duration weighting?'
Do we really know how the measurement works?
Grateful thanks to Ed & Josh and of course George Ivie. I am planning a folllow-up piece and hope to include other POVs on this tricky and still confusing area.
Based on George's answers, I do not believe that MRC intended to change the construct of audience metrics - GRPs, reach or frequency - all audience measurement based terms but apparently it is believed that MRC is proposing to do just that. My undersatdning is that duration weights would apply to "viewable impressions", a dreadful confusing term, that are solely a proof of performance metric - was the ad/content rendered on a given screen per the media vendor specification and if not complete over time by how much. As such, the proposed "duration weighted viewable impressions" have nothing to do with audience measurement or campaign effect per se only as a qualifyer that an ad/content was rendered and that consequently an audience impression of a specified ad/content could be legitimately counted.
I tend to agree with Josh that as proposed duration weighting appears to introduce an adjustment based on a single video dimension - duration (of very very many - what about screen size!!!) concerning the quality of the potential experience of the rendered ad/content to a subsequently exposed audience.
Surely reporting that an ad/content was rendered on specific screens for the specified duration (or partial therof) should be separated as an independent POP reporting metric for the audience measurement company, media agency and media vendor. These POP dimensions can simply be considered along with many other media & campaign attributes when evaluating traditional audience metrics (GRPs, Reach & Frequency) and the relative "value" of those metrics across media platforms when presented to advertisers as is current practice?