CTV: $20B Faith-based marketplace
The billions spent with little measurement to prove it.
The argument
Perhaps never have we seen an advertising marketplace spring up with so little measurability. Yet there is a line of brands down the block desperate to spend. Consequently, investing in CTV is a faith-based exercise where we trust the marketplace to deliver on the promise without verifiable proof of delivery or quality.
At the top, I firmly believe that 100% of television advertising will be addressable and delivered to viewers via internet pipes within ten years. There is zero chance that the current linear business model survives into the next decade.
However.
Today, it is a mess.
Measurement of campaign delivery and performance, in line with modern internet protocol-based activation systems, simply does not exist. Yet.
Our industry faces a confluence of challenges, including technology interoperability and poor [non-existent] incentives to change.
Hang on just a second
You: “Travis, today I’m able to measure my CTV delivery and performance.”
Me: “Mmm…no you’re not. You’ve just been led to believe that’s true.”
You: [Nervous look on face]
Me: “Let’s follow your ads through the pipe.”
How you purchased the CTV ad is irrelevant. Programmatic and direct buys arrive at the publisher’s yield management and ad insertion platform. This is where your direct measurement and transparency end. Full stop.
Server-side ad insertion (SSAI) is a remarkable technology that changed the game in streaming user experiences. Since the ads are inserted into the content stream (during commercial breaks) in the cloud, users are no longer faced with buffering and freezing due to ads. The experience is designed to be as smooth as traditional television.
However, this came at the expense of true client-side measurement. In other words, your measurement vendor tags (ad server, verification, etc.) never actually reach the streaming device or app. They stop in the cloud at the SSAI point.
Your vendors are not actually measuring inside of the end user’s streaming experience. Your vendors are stuck much further upstream.
Any information they do receive is passed to them from the publisher’s ad tech stack.
This is like being “told” what happened versus actually “witnessing” what happened.
In my book, that’s not measurement. That is passthrough reporting.
See also: our two-part series on CTV, including important industry terms:
What does this mean?
It means a lot of unanswered questions.
Where did the ad run? You might be able to find out what app the ad ran in, but you are much less likely to know what program it ran in. None of your ad tech vendors will be able to tell you.
Was it a legitimate impression? Tough to say. Your verification vendor did not get to measure inside of the app or device, so they have very little information available to them to determine if the impression was a real person.
Was it viewable? Hahaha. Of course it was, chief.
What was the frequency? Lolz. You’ll take it with no cap and ask for more. However, even if you want to measure it, SSAI makes that a challenge.
The alternatives? Panels.
For the ultra nerdy among us, that’s a play on words because there are panel measurement vendors and panel manufacturers that can both provide interesting solutions.
Since riding the rails of the traditional ad server + verification vendor provide limited data and transparency. Instead, we can turn outward.
Panel-based measurement vendors
Instead of measuring within the ad delivery stream, panel measurement vendors set up shop in the homes of their panelists. Volunteers (often with nominal financial compensation) agree to have themselves “watched” for their TV viewing behaviors.
A panel can consist of a few hundred or tens of thousands of participants, all of whom agreed to have their second-by-second TV consumption monitored in real-time. This allows the measurement vendor to “see” what shows are being watched and what ads are appearing in those shows.
The results are extrapolated from the sample to give a reasonable picture of what occurred in any given campaign.
However, the results are only accurate to a point and often have challenges knowing “how” an ad ended up in a particular program. For example, understanding if a specific car commercial was placed there by a DSP, direct buy, or network can often be impossible for a panel-based measurement firm to detect.
TV Manufactures (panel makers)
Television OEM’s have technology in them called ACR that enable them to detect what is being shown on the screen at any given point in time. This is powerful.
The TV makers themselves have millions of devices in the field capable of sending back second-by-second consumption data.
Unfortunately, they too suffer from some of the sourcing challenges where it is difficult for them to understand how the ad actually got placed in the program. Still, millions of devices are far more scalable than the traditional panel measurement vendors.
The solution
Spoiler: There isn’t one…not in the short term.
I mentioned at the beginning that we suffer from two challenges: Tech and incentives.
The incentives are the part that we as marketers can influence. We need to incentivize transparency and genuine, open measurement.
Unfortunately, we may have to wait for some of the Top 25 CTV advertisers to throw their weight around and demand that publishers and platforms allow granular, device-level measurement by independent third parties. Today, there’s not an ounce of incentive for content owners to allow this because demand outstrips supply.
Yes, I know many open measurement initiatives and committees are sprinkled around the ad tech ecosystem. They are noble endeavors, but they too require massive advertiser weight to drive adoption.
In the meantime, the CTV market will continue to froth until even the sleeping giants have had enough.
For now, stay optimistically skeptical of everything you’re told. It isn’t wrong…it just isn’t completely true.
Support Independence
Getting the FREE AdLingo newsletter supports independent commentary, education, and insight into digital advertising.