Michael Treon On How PMG Is Driving Innovation In CTV With Eye Towards AI And Video-Level Contextual Advertising

As Programmatic Lead at PMG, which recently acquired Camelot Strategic Marketing & Media, Michael Treon is always looking to maximize marketers’ return in their advertising investment. As streaming becomes the preferred medium for TV viewers, the challenge is not only reaching the right people, but in the right moments at scale. He spoke with TVREV about how Camelot is leaning into innovation to with video-level contextual and how the IRIS_ID is making it easier for marketers to invest in streaming TV.

ALAN WOLK (AW): Can you describe PMG’s approach to video advertising and how it is unique from other strategic marketing and media companies?

MIKE TREON (MT): First it’s worth noting that in December, Camelot was acquired by PMG, so now we’re part of an even larger ecosystem focused on data-informed, tech-enabled, integrated solutions, which of course, includes video advertising. A big part of our approach has been not settling on just one path. Our lineage has always been creating strategies that drive outcomes for our clients and lead with investment in “first, best and only” opportunities. It's an ongoing and ever-iterating process in which we don’t settle and are constantly evaluating and evolving our planning, buying and measurement playbooks. Those strategies are particularly important in the converged video space, where the fragmentation and constant evolution means there is no easy button as we keep pace to drive those outcomes. We're focused now on scaling those playbooks and strategies, and at the same time, being able to pivot, test, and take advantage of market shifts, new opportunities and respond to changes in addressability, technology and privacy.

AW: Linear TV has been known as a branding medium, whereas search, social, and display  are proven channels for performance markers. Does CTV, with its mix of TV-like content and digital buying capabilities have the necessary conditions to be a desirable place for performance marketers? If not, what needs to be done?

MT: For sure. I think CTV will always evolve with aspects that are appealing to both brand and performance. We try to approach it with a lens on both the top down and bottoms up effect of the medium, and employ tactics from both a linear brand and content focused strategy, as well as the data-driven programmatic approach. I don't think there will ever be a point where we’ve solved for every outcome or metric, but constantly evolving the approach gets us further down that line of employing and developing measurement that helps us refine our focus on signals across that spectrum and ultimately help us to make the medium desirable.

AW: Contextual signals have emerged as an alternative to traditional people-based signals as privacy regulations and technology limitations have made it harder to reach the right viewers in CTV. Does Camelot employ contextual data in your campaigns? If so, can you describe your approach and some of the results?

MT: Yes, overall contextual signals are a key part of solving for the widening addressability gap across the board. Contextual targeting is proven to deliver results in formats like display, and now in CTV, contextual targeting strategies require access to video-level data and in light of the Google cookie deprecation announcement on January 4, we're making that a priority in our discussions with streaming video publishers. With CTV in particular, it's proving to be an important tactic in aligning to the content focus of linear and extending that throughout the various inventory categories we evaluate from the direct to consumer apps, to the multi content distributors and OEMs, to the FAST landscape. With a solution like IRIS.TV, and the contextual engines enabled by their content data platform, we’re able to combine the scale and targetability of FAST with a nod to content, sentiment, and brand safety. These solutions are proving to be just as important as the audience in delivering value, particularly in the FAST landscape where curation at scale is difficult. It definitely draws an interesting parallel to the audience-focused programmatic buying strategy we’re familiar with in the display, mobile, and online video landscape; however, when paired together, we think it's a great approach to solving for the use case and landscape of CTV.

AW: While AI solutions can scan and analyze video frame-by-frame to create targetable contextual and GARM brand suitability segments, publishers are also sharing more content metadata in the bid stream. Is publisher-declared metadata sufficient for you to shift more investment into CTV? Or are you seeing greater value from video-level contextual segments?  

MT: More signals will always help as long as we are aligned with the supply side. We’ve taken efforts to have that communication with publishers and be more transparent on our side about how we’re using signals, our strategy, the technology solutions we need to accommodate for, and how it helps drive demand. The different categories of CTV supply -  and their inherent nature - dictate a different approach to transparency. Given the scale of FAST, the addition of contextual AI really helps us scale that programmatically, more so than relying on curating publisher-declared signals.

AW: How is Camelot solving the transparency problem in CTV, and why should publishers care? Or said another way, what specifically can publishers do to attract more investment from buyers like Camelot?

MT: Solutions for transparency really depend on what signals work, align to the technology and human infrastructure, and can be scaled and replicated. Our approach has been specific to the nature of the content being delivered, whether it's owned and operated, distributed, or long-tail curated. Overall, we have an eye on safety, a focus on alignment, and a mindset to find what works and is scalable. Publishers will need to consider their role, their content, and the objections that are hindering adoption and be flexible on supporting those in mutually beneficial ways. There are a lot of effective solutions that solve for safety and quality content that are not just show level reporting.

AW: Is the lack of consistency and interoperability of the publisher-declared data a challenge? Do new data signals like the IRIS_ID address any of these challenges?

MT: Absolutely. We’ve done our best to try and align on some of the more standard signals like network and channel, but it's an exercise in big data when you go to leverage, plan, or report on the metadata itself. The meaning of genre, title, and other content data vary significantly because publisher declared metadata is subjective and less uniform, which hampers signal adoption. The IRIS_ID sets the stage for a unified view of content combined with technology to scale activation and is proving to be a valuable tool for how we standardize content targeting across supply sources at scale. What sets it apart from other content IDs and registries is its interoperability with data solutions and ad platforms making it easier for buyers to invest in streaming TV.