Ex.Co Gets Into the AI Content Recommendation Business: A Chat with CEO Tom Pachys

AI and machine learning seem to be all the hype in the advertising ecosystem now that generative AI has stepped on the scene.

But many players throughout the advertising ecosystem have used AI and machine learning to increase productivity, create more meaningful dashboards, and serve more relevant content and advertising to audiences.

So it comes as no surprise, the EX.CO, a leading publisher-first video tech platform would boost its video offering for publishers, especially after acquiring Bibblio, an AI and machine learning company. The ad tech company’s new AI-driven content recommendations and premium video libraries for publishers will identify the most contextually relevant video for each article and the highest-yielding feed for a page on a publisher’s site.

I spoke with Tom Pachys, CEO and Co-Founder, EX.CO, to learn more about the launch, as well as how EX.CO increases value for publishers by streamlining the supply path, and how AI can help fix ad tech’s sustainability issue.

Lynne d Johnson: AI is all the hype right now because of the popularity of Generative AI. You recently launched AI-driven content recommendations, after acquiring Bibblio, an AI and machine learning company. You guys already provided content recommendations, how do these enhancements improve upon what you already do?

Tom Pachys: EX.CO as a technology company always believed machine learning should be used to solve problems that are impossible to solve. As part of our online video platform, we’ve invested years in creating AI-based algorithms to improve monetization yield. While the buy side had AI-based technology since the early days, publishers never had access to such advantages without the likes of EX.CO.

After our acquisition of Bibblio, we were able to further enhance our offering using machine learning to select the most relevant video content as well. By using such technology, we enable publishers to always show the most relevant or highest-yielding video on each page. This is one of a few enhancements to our video content capabilities, another one being a premium video library with over 50,000 videos in different verticals that will now be available to our customers.

LdJ: Why should publishers care about this new feature in your video technology platform? What are the benefits for them?

TP: Our new content recommendations have proven to increase revenue up to as much as 17% for publishers, so this is a meaningful boost. With this new enhancement to our player technology, the video shown on each page of a publisher’s site will be the one most contextually relevant to the content of the article that video is placed in, which leads to a more personalized and engaging experience. In addition to revenue growth, our contextually relevant videos have improved visitor dwell time by nearly 50% on average. We’re very excited to bring that type of innovation to the publishing industry. It’s really a win-win all around for publishers and their audiences.

LdJ: And for buyers, how does this change things for them in terms of campaign performance or campaign management?

TP: We’re equally excited about how these new capabilities will impact advertisers as we are publishers. First, we pride ourselves on representing the shortest path to the inventory; we believe in reducing mediators and increasing value to the publisher. When brands and agencies buy inventory from EX.CO or the publishers that are working with EX.CO, their video ads will be seen by the most engaged audiences and within the most relevant video content. Users that watch more personalized videos really lean in more to the content they are consuming which can only improve brand recognition, brand recall, and purchase intent. That’s really what everyone has been waiting for, isn’t it?

LdJ: As much as people are excited about AI, AI also gets a bad rap for either going off the rails, as we’ve seen with some ChatGPT examples. But also bias has been a big issue that AI has been called out for. I mean the technology is only as smart as the data it’s being fed. Can you tell us about the human involvement that goes into your AI-based algorithms to ensure that publishers are actually serving the most contextually relevant video for users without any of these issues that have been concerning about AI’s future?

TP: It’s true; the algorithms are only as smart as the data they are being fed and the investment in building them. While we could look at engagement data for a user like dwell time, frequency of visits, etc. this tells us very little about what the content (and in extension, the algorithm) is doing to a person’s emotional response.

We understand a multidisciplinary approach is needed to achieve great results. It’s not only context that is important but also, recency of the article, popular interest, semantic analysis, safety filtering,  and other metadata such as tags and categories. This makes it a very pure, contextual algorithm with very little potential to cause negative impact. The challenge is achieving the human “wow factor” of generative AI while balancing the different publisher’s needs.

Many of the issues you are referring to tend to be more prominent when the technology is working on huge data sets including UGC where the fundamental “nutritional” value of the content is obviously less controlled. Our challenge is often the opposite: how to create a good recommendation on a much smaller sized data set (library of videos) while making the recommendation both relevant and accepted by journalists and editors working under a defined set of moral and ethical standards.

LdJ: And what about sustainability? That’s a big issue in the advertising ecosystem right now as everyone is thinking about being more responsible and hitting their carbon emissions targets. Well, as AI grows, so does its carbon footprint. Are you guys thinking about doing anything to make sure your technology is not becoming part of the larger problem? 

TP: When looking at the publishing industry, one of its biggest polluting components is the inefficient use of on-page ads. Most of the industry works with a “spray and pray” approach—more vendors, more requests—which leads to more pollution. We believe AI is the solution to that. Our biggest focus is supply path optimization, making the value chain more efficient which boosts our customers’ bottom-line revenues as well as supports the environment. EX.CO is pushing that agenda in all force and I believe we’ll see big positive changes in the industry in the coming years.