January 25, 2023
Commentary
Which 'Impressions' Are You Counting?' Were 'Media' Measurement Opportunities Missed?
- by Tony Jarvis , Op-Ed Contributor, May 12, 2021
Did today’s Nielsen-sponsored Virtual Summit on streaming -- “What’s Next for CTV?” -- miss the real opportunity for industry leaders and Nielsen to drive meaningful cross-"media" measurement that is begging for a solution?
Examining ad-campaign outcomes and their drivers is a fundamental requirement for brands, which Marketing Mix Models …
2 comments about "Which 'Impressions' Are You Counting?' Were 'Media' Measurement Opportunities Missed?".
Read more comments >
Next story loading 
Agree, Tony. It's simply unrealistic---and unfair---to saddle the media with the responsibility for convincing consumers that the advertiser's ad message is valid and they should buy or use the product. From time to time a media seller will go this route---but only after getting a firm read on likely outcomes based on lots of past research from the same source that will conduct the future monitoring of "outcomes"--usually a share of market lift. The resulting "guarantees" almost always fall well within the likely range of outcomes and, as a result, are virtually meaningless---except as an ad sales promotion ploy. If, by chance, there is a shortfall in "delivery" this is handled via make good ads ---which are readily available at no cost to the seller as a rule. Most sellers---sensibly---don't venture into this--to them---very dubious area.
As for the campaign to convert everything from GRPs to "impressions"---as if doing things the "digital way" represents a major improvement--- this is simply an absurdity as the two mean the same thing. And what exactly is an "impression, anyway? Is it an inflated projection---actually an assumption---of "average commercial minute viewers" when the commercials were on-screen as we get for national TV? Or is it average quarter hour total audience, including times when the ads are zapped--for local TV? And what about out-of-home TV? If the PPM device "hears" a TV set does that mean that the person wearing or carrying the PPM was "watching? And digital? If a 5- or 15-second video commercial was on-screen for at least two seconds is that a valid "impresssion"? Did the user---even if present---pay any attention to the ad?As for print media or display ads in digital media we simply have no audience measurement of specific ad page audiences.
Bravo Tony.
Totally concur on the 'impressions' comments.
Though one thing I found about MMM is that the 'good' models can be data agnostic. For example, if you had (say) linear TV 'impressions', CTV 'impressions' and digital video 'impressions' and good MMM will analyse and basically be non-comparative of those three types of 'impression' and (if they could speak) would say ... well I don't know what a (say) CTV impressions is, but I know that whewn we have truckloads of them we sell more product.
But one thing I found in MMMs is that the model will still have 20%-30% 'unknown' variation. That is, it can explain/forecast around 70% of the variation in sales.
In many cases that is because the MMM focuses too much on the media component at the expense of 'real world' factors. A good MMM will have things like unemployment levels, average income data, consumere sentiment, season, weather etc etc. We did a model for Macca's Happy Meals. Rainy afternoons were GREAT for sales as the mums and dads were picking up the kids after school, stuck in rainy traffic, didn't want to get home and cook dinner so ... anyone want a Happy Meal? The Happy Meal ads (and live reads) on drive-time traffic reports on the radio worked a treat on rainy days.