I have a very simple proposal: Let's rate the impression.
Wait, I know what you’re about to say. “But impressions are rated now, thanks to the MRC ratings that were announced earlier this year!” Okay, those were baby steps. But they aren’t really “ratings.” It was a great start and I appreciate it, but we still aren’t assessing our “impressions” nearly enough.
We all know that not every impression is the same — that’s a no-brainer — so let's weight them based on math. Don’t just say an ad was “viewable,” but additionally rate long the ad could be seen, how much of the screen the ad took up, and more.
Easy. Right? Shouldn’t everyone want this?
Nope. Impression farms, the term I use for the people churning out these volumes of near-worthless impressions, would be very much against it, and even quality content publishers would need to be paid a fair rate in order to want these standards. We’ve been surviving as an industry while either turning a blind eye or outright complying in the packaging of meaningless impressions into volumes so huge that when they all add up they look good. This is what I’ve referred to as “the subprime ad crisis,” because a similar mass packaging of something —else — junk mortgages — is exactly what led to the subprime mortgage crisis. This situation is really no different. And that is exactly why we have to move to rate impressions better — or watch true advertising as we know it disappear in a digital world.
advertisement
advertisement
I think everybody, inside and outside of the advertising industry, would agree that two seconds of half an ad on a page (who knows if the sound is on, or if a person even looked at it) is near worthless. But since we as an industry intentionally set the lowest common denominator for qualifying as an "impression," we are left with this binary definition. There are some companies, like MOAT, that are working hard and beginning to shed light on what advertisers are really buying and what publishers are really delivering. But without expanding the definition of "impression" to recognize that not all impressions (or potential impressions, really) are equal, it won't fix the bigger problem.
So like I said, a simple proposal: Let's rate every impression on how much of a consumer’s screen it takes up and for how long it does so. It wouldn't have to complicate anything at all. Instead of buying binary impressions — was it viewable or was it not? — an advertiser could buy units of time in view. Now, there is still the problem that 30 one-second ads can't tell a story, whereas one 30-second ad can sell a car, and both would technically have the same “time-in-view” score. But let's start with baby steps.
When will it be that when someone tries to solve viewability they also address that dig ads are not worthy marketing messages that deserve viewability? Maybe the problem is these ads are seen. Why don't we address this first problem that has hampered the digital business since day one? Rocket science: crappy ads = crappy results.
(Slow clap). Especially love the "subprime ad crisis" phrase--and your thinking here. Great post, Joe.
I like where your head is at. Not an ultimate solution but it, at the very least, calls to the mat every publisher and every ad we sell.
It might be simpler if brand advertisers simply went back to dealing more or less directly (or private exchange/programmatic) with big brand publishers who restrict their inventory to quality/viewable units...and pay a fair price. Stop playing games with thousands of third-tier sites the huge 3rd party ecosystem that drives them (and siphons off all the margin from those who would like to do better). There's more than enough reach available via the top 500 or so publishers, who could be vetted (although they're generally trustworthy so if the contract for a certain size/above the fold, that's what they'll deliver). Quit monkeying around trying to score tonnage, and all those reams of "meaningless impressions" will go away, and it doesn't require another layer of tech/3rd party toll-keeper to execute, or any grand industry consensus. If there's a market for quality impressions, big publishers will deliver them.
Nice one Joe. I'd add that if quality measures can be used for bid decisioning, the entire programmatic marketplace rights itself. All auctions depend on knowing the value of what is being bought of course. Right now, advertisers and buy side platforms that get this are doing very well. This might be the most important opportunity in the marketplace at the moment. There is progress on creating indices of quality, as you suggest (e.g. Moat creates a composite index including time in view etc.). Maybe the industry will turn the corner when advertisers, instead of throwing up their hands, realize that exchanges are the key to market liquidity, and figure out that measurement pays for itself as multiples of the cost.
When you have a product like advertising that you have to force or trick people (with rare exceptions) into viewing, you possibly have an unethical business model. The folks who really like advertising are those who peddle it.
In 2003 I wrote a piece of software for Foote Cone & Belding that basically boiled everything down to impressions, then used the impressions to mix media reach. It was basic, but, hey, I was EVP International Media, so I could put some muscle behind it. It was readily adopted by clients... they understood, all impressions come from essentially a single bucket, people. So, if you reach X people with radio, Y with TV, Z with web... it would make sense to treat everything at the molecular level: Impressions.
I used the Sainsbury formula to guesstimate duplications, triplications, etc., and arrive at a net net reach, then transformed the impressions into TRP's and got average frequency.
It is a simple, thoroughly workable idea: everyone comes from the same bucket.
As to the issues of "exposure" or real viewing, we have always faced this on television, radio, newspapers, magazines, out of home... it is worth continuing to learn along those lines, but not worth it to be completely paralized if you don't have the data