Here’s today’s AdExchanger.com news round-up… Want it by email? Sign up here.
By Unpopular Demand
Last year, Google launched a product it calls Demand Gen. It’s cast from the same mold as Performance Max, in that Demand Gen is a black box controlled by Google’s machine learning algorithms.
In terms of functionality, however, it more so mimics Meta. You know, taking a list, as Facebook is very good at, and creating endless lookalike audiences.
But Google ecommerce and SEO operators are sounding off about a new Demand Gen option that attributes all conversions to Demand Gen campaigns rather than to multiple touch points. The new conversion tracking is mentioned in support files, but apparently hasn’t rolled out to accounts yet.
If all this sounds familiar, it’s because it mirrors what happens on social platforms, which often over-attribute to themselves by claiming full credit for conversions tied to ads served within broad attribution windows.
Demand Gen uses fractional credit – partly because it prospects and then hands customers off to PMax. But Google also owns so many properties, including Search, YouTube and Gmail, that usually fit in at least somewhere for advertisers.
Fractional attribution is better for advertisers, but it makes Demand Gen’s ROAS look bad compared to Meta Advantage+ Shopping Campaigns – which is likely why we’re seeing this new attribution mode from Google.
If It’s Not Broker, Don’t Fix It
Consumer data brokers attract a lot of scrutiny for their tracking practices. But Ad Age reports that brokers like Acxiom, Experian and Epsilon have majorly misguided audience segments and profiles.
For one thing, broker data isn’t refreshed consistently. So information from years ago is still knocking around. For example, people could be assigned as “auto purchase intenders” because they browsed a story about electric vehicles multiple years ago.
Brokers also may have one individual classified as both a man and a woman, a high and low earner, a resident of multiple states and pegged as a native language speaker hailing from a country that person has never visited – all at once.
It’s like the third-party ecosystem actually doesn’t track people effectively, at all. Amazon, Google and Apple, by contrast, maintain detailed, intimate portraits of all their users.
Meanwhile, consumer data brokers owned by agency holding companies – Publicis owns Epsilon, and IPG has Acxiom – are another complication. There are clear conflicts of interest when an agency, ad tech vendor and data supplier are owned by one entity.
TTD’s TV OS
The Trade Desk has been secretly building a TV operating system to be implemented by manufacturers, reports streaming TV news outlet Lowpass. The first device to carry The Trade Desk TV’s OS could be launched as early as next year.
Developing a TV OS puts The Trade Desk in more direct competition with Roku, Google TV and Amazon Fire TV, in addition to other device manufacturers with their own advertising businesses, such as LG and Samsung.
One ad exec theorizes that TTD wants to raise the number of addressable devices it can reach through targeted advertising – the same reason why TV manufacturers build ad businesses and vice versa.
The Trade Desk is “betting its future on its ability to become the primary buying system for the future of TV,” writes media and ad tech veteran Ari Paparo in his Marketecture newsletter.
But perhaps the most important aspect here, Paparo writes, is the ability to generate automatic content recognition data, the device-level viewing data buyers crave as a way to add some consistency to cross-platform measurement.
But as for The Trade Desk’s new OS, the most pressing question out of the gate will be scale and reach, since it starts at zero.
But Wait, There’s More!
Reddit doesn’t have a formalized publisher program (yet!), but that hasn’t stopped publishers from experimenting with the platform. [Digiday]
Here’s how much value investors that supported Elon Musk’s Twitter acquisition have lost since he took over the company. [WaPo]
National Novel Writing Month (NaNoWriMo) angers writers with its confusing new stance on AI. [The Verge]
Meanwhile, a new report from the Australian government suggests that generative AI is consistently worse than humans at summarizing information. [Crikey]
You’re Hired!
MiQ appoints Zuzanna Gierlinska as UK chief commercial officer. [release]
This newsletter has been updated.