Pushing CTV measurement forward in a fragmented ecosystem

With OTT advertiser spend expected to hit 5 billion by 2020* and connected TV (CTV) consumer adoption skyrocketing, the digital industry as a whole is struggling to keep up with the demands of this burgeoning market. The explosive growth has created several hurdles for advertisers, a big one being standardized measurement. From an October 2019 eMarketer article titled, “Connected TV Ad Spend Is Growing, but There Are Still Plenty of Challenges,” researchers explained the difficulties advertisers are facing with this new medium – while an exciting area for innovation, some advertisers are hesitant to funnel their ad spend into a platform that can not be fully measured. 

CTV measurement is difficult for a variety of reasons, beginning with the fact that the market is flooded with so many new devices and services that it has become a highly fragmented ecosystem. Then there is the extensive use of Server-Side Ad Insertion (SSAI), the process by which ads are stitched directly into a video stream before ever reaching a user’s device. SSAI is preferred over traditional ad serving methods as it greatly reduces the risk of latency; the consumer wins with a seamless viewing experience, but measurement is challenged. By nature of SSAI, technology vendors have no direct access to the actual device, which traditionally has been a requirement for measurement. Finally, core measurement technology, like javascript, is not universally supported across CTV devices. 

IAS’s first-to-market CTV solution

In an effort to address advertisers’ concerns about the lack of transparency on connected TV devices, IAS began partnering with the largest video publishers in 2018 to develop a CTV solution that validates video ads play to completion, free from invalid traffic (IVT), on a TV screen. When developing our solution we quickly realized the importance of working directly with publishers to ensure that we are receiving and fully understanding all relevant signals in order to correctly identify invalid traffic. It is our belief that anyone not working directly with publishers runs the risk of incorrectly classifying new and unknown signals as IVT.

Measurement across CTV has posed new challenges for us to solve and has required new technology to accurately measure at scale. Our solution is the first step in creating scalable IVT measurement across CTV, but ultimately standardization will be required to unlock comprehensive measurement. We’ve learned a lot from our work thus far and have compiled key learnings and recommendations to help the industry advance towards the necessary standardization. 

Findings 

Device manufacturers are eager to capitalize on this opportunity and new versions of streaming units are popping up everywhere, all over the world. While this gives consumers endless choices and ensures companies are offering the most competitive products and prices, it also creates a challenge for measurement as the current checks and balances supported by the industry for detecting invalid traffic are not known to all the players involved. For example, we had identified a legitimate new streaming device out of Australia, but since the device wasn’t registered with the IAB we were required to flag it as invalid traffic. Upon contacting the company we were able to help them understand the requirements for registering their product and resolved the issue. 

On the software side of things, as new CTV devices and versions are being released, publishers are quickly working to develop accompanying apps. This can result in unstandardized user agents and app IDs across publisher inventory, which can be misclassified as invalid traffic if not properly standardized. 

We also found that across the ecosystem there are variations in how SSAI is implemented, even across different properties within the same company. It is not guaranteed that the minimum device signals needed for invalid traffic detection are accessible, and there is no reliable way of knowing what signals are being made available when.  

Recommendations

1. CTV device manufacturers must make it a priority to have new device user-agents correctly added to the IAB list to avoid being identified as unknown and automatically classified as invalid traffic. This will help alleviate any reporting issues and ensure we are set up to successfully discern real fraud from legitimate devices. 
2. Publishers should review that existing user agents and app IDs are properly standardized across their properties and employ diligence in maintaining standardization in future development. What’s more, well-defined app IDs will become increasingly important in order to deploy scalable brand safety solutions across the CTV ecosystem. The IAB is currently working to define a CTV app ID standard, but in the meantime, publishers should ensure their naming convention is consistent and as descriptive as possible.
3. It is imperative that there is swift adoption of the elements of VAST 4.1+ that standardize how SSAI is communicated throughout the ad delivery chain. Since SSAI inserts an intermediary (an ad stitching service) between the device and measurement parties, it has been highlighted as a source that can be exploited for fraud. Until we reliably get standardized transmission of required data points across the whole industry, it is difficult if not impossible to separate fraud from the false positives caused by a lack of data. VAST 4.1+ also incorporates additional supported data signals that provide value in CTV measurement and we’ve outline our suggestions here.