Connected TV advertisers know more about who sees their ads than where those ads run. The change marks a sharp contrast from the days of linear TV buying against program guides.
While buyers’ calls for transparency are becoming louder and more urgent, programmers are saying their hands are tied over a video-rental-era law from 1988, the Video Privacy Protection Act (VPPA).
The VPPA prohibits video content providers from disclosing information about video consumption that’s linked to an individual.
The VPPA takes effect when viewership data is linked to personal identifiable information (PII) without proper consent.
Fast-forward a few decades, and programmers are interpreting the law to mean they can’t include both a user ID and show-level information in a programmatic bid request.
The law from the ’80s has “come back to life” for video programmers in a “mini resurgence” of litigation and caution, said Gary Kibel, a partner with the law firm Davis+Gilbert.
In the digital age, the VPPA affects how broadcasters choose to sell connected TV media and what they report back to buyers, said Ryan McConville, EVP of ad platforms and operations at NBCUniversal.
Privacy, please
Unlike streaming, linear TV ads are historically bought and measured against pre- and post-roll program logs. Those logs confirm what played and an aggregate count of who watched. That log-based reporting isn’t tied to identity and can’t be connected to specific households, McConville said.
But because more data is generated by connected TV streaming, publishers have to be more careful about disaggregating viewership data from granular ad call data that contains IP addresses and mobile ad IDs, he said. Those identifiers can be attributed back to a specific user.
NBCU, for example, tells advertisers the top 10 performing shows for their CTV campaigns based on audience viewership at an aggregate level, but not at the level of a deterministic device ID.
McConville says NBCU does break out show-level reporting, but not on an impression-by-impression basis.
“We do not provide show-level metadata in the bidstream with identity data,” he said. Under VPPA protections, “you just can’t have identity data next to show-level data.”
There’s no media industry standard on how exactly to apply this decades-old law to the business of modern streaming services. But the message sent by VPPA is clear: Viewership data can’t be linked to an individual.
In many cases, CTV buyers struggle to get any episode-level information in their campaign reporting at all.
Like NBCU, Tubi also gives advertisers aggregate reporting on its campaign’s top performing shows, but it doesn’t share program-level viewing data in the bidstream, said Mark Rotblat, chief revenue officer of Tubi.
NBCU and Tubi aren’t alone, either. Another programming distributor that didn’t want to comment on the record about legal matters told AdExchanger that it didn’t include show-level data with IDs because it would violate the VPPA.
“I’m not aware of other companies sharing show-level information in a bidstream,” Tubi’s Rotblat said. “I don’t think there’s much progress there, and that probably won’t change anytime soon.”
Matching what someone is watching to an identifier is very problematic, he added, and the onus is on publishers to make sure they process their user data in compliance with the law.
Agency GroupM sees the effect of VPPA on its CTV data visibility.
“A lot of connected TV publishers have been using the VPPA as a reason for not rolling out show-level transparency with bidstream data because it can be linked to specific households,” said Mike Fisher, VP of advanced TV and audio at GroupM agency Essence.
Risky business
Streaming services that leak subscriber viewership data tied to an individual without obtaining that individual’s proper consent have been subject to VPPA litigation.
A lawsuit in 2015 accused Hulu of sharing viewership data to Facebook with a “Like” button on its site, but a judge ruled that Hulu didn’t intend to transmit video title history that was interwoven with personal identifiable information (PII). VIZIO also settled a class-action lawsuit for violating the VPPA in 2018.
And earlier this year, HBO was also hit with a class-action suit for sharing identifiable video history with Facebook. This time, the suit argues that HBO knew Facebook would be able to link viewership data with Facebook profiles because HBO is a major advertiser on the platform.
Given the threat of legal action, the way to avoid a video sharing lawsuit seems straightforward: Don’t link viewership to a user ID.
Meting out measurement
Since getting show-level data through the bidstream just isn’t happening because of the VPPA, buyers are looking elsewhere.
Device-level data, namely automatic content recognition (ACR) data, is becoming more popular among buyers to fill in the gaps between viewers and actual viewership, which explains why measurement providers are currently busy getting mitts on as much ACR data as humanly possible.
GroupM, for example, uses ACR data to determine when and where an audience cohort was exposed to an ad and whether or not that exposure led to an outcome that’s meaningful for a brand’s KPIs, said JiYoung Kim, chief products and services officer of GroupM’s North American business.
Cohort-based targeting isn’t as deterministic as tying viewership to a deterministic device ID, but it’s a way to incorporate more granular content signals into media buys than just content genre or category, Tubi’s Rotblat said.
Tubi, for one, is trying to meet CTV advertisers’ performance marketing goals using what it calls “content clusters,” which it uses to measure aggregate audience viewership against specific types of content. NBCU also shares campaign reporting at the aggregate level against content types.
And because audience cohorts and look-alike modeling are becoming more popular, expect data clean rooms to become even more important in the CTV measurement and data transparency discussion in 2023, Rotblat said.
Buyers can also use blurrier audience signals, like cohorts and look-alike modeling, to connect their audiences to viewership data. An advertiser working with GroupM, for example, can match viewership signals with aggregate data sets to determine whether a target audience segment expresses interest in certain types of content, Kim said.
The goal is to figure out what shows resonate with a brand’s target audience and be able to make more informed buys in the future, even without user-level matching. Given where general privacy trends are headed for media and marketing, it makes sense that CTV advertisers will also have to figure out a way to make do without a guarantee of user-level signal.
