Vidora’s AI Spotlight Series is a series of conversations with leaders, academics and pioneers in the world of artificial intelligence about where AI is today, and how it will change the way we do business.


Adelphic is a leading cross-channel platform, providing an enterprise-ready self-service software solution for agencies, brands and other large media buyers. Adelphic does this in order to to make meaningful engagements with consumers across all devices and formats. Time Inc. recently acquired them, because their technology overcomes the limitations of user identification across digital devices. As a result, this yields rich, nuanced portraits of real people. This therefore gives marketers the power to deliver precise communications that result in meaningful interaction, dialog and transactions.

Yael Avidan is Adelphic’s Head of Product. Yael has led the charge with developing Adelphic’s powerful platform solutions. We sat down with her to chat about the state of the mobile world today. We also spoke about Machine Learning and Artificial Intelligence technology, and the future of AI. Finally, we discussed how publishers like Time Inc. are leveraging that technology in a changing industry. Here is our conversation:

On Artificial Intelligence and its distinction from Machine Learning

How would you broadly describe the state of Artificial Intelligence (AI) in your industry?

YA: I think it’s important first to be clear on what AI is. Only a handful of companies engage in “true” AI. I refer to companies like Google, Facebook and maybe a few others. Everybody else is doing machine learning (ML). In ML, algorithms are sifting through big data and coming up with predictive outcomes from it, depending on what one is trying to accomplish. Humans still guide those outcomes and algorithms. As a result, to me, this separates it from what true AI is.

On the state of ML in the mobile space today, and its potential for marketers

Given that distinction, what kind of impact do you think ML has made in the advertising space to date?

YA: ML has created a lot of change in the automation of mobile advertising. Advertising, and mobile advertising in particular, experiences something very similar to what happened in finance. In finance, we’re now able to collect metadata on trillions of transactions, all at once. As a result, we can use this information to get to automated and optimal decisions. Auctions run two-thirds of advertising today, and those transactions are now being automated with ML.

ML in Automating Mobile Advertising

So how does ML help automate that process, and what is the role that Adelphic plays in this?

YA: Adelphic sits of the heart of the mobile advertising transaction, and sees 40-50 billion transactions a day. Imagine every one of those transactions as a vector that has close to 50 different dimensions to it. There’s obviously a lot of analysis and decisions that need to be made in terms of optimizing those transactions. In the past, this optimization was done with simple rules-based approaches, but its scalability was very limited. ML has been important because it can handle high-dimensional data at scale. At Adelphic, we use ML in 2 places. First, the multi-layered decisioning required to sift through the trillions of requests that come to us and determine if we want to bid, and what price we want to bid for. ML is also linking user IDs across devices.

Interesting – let’s discuss that second area in a bit more depth. How do you link user IDs across devices using ML?

YA: Linking a user who is not logged in to a service across devices gives us a huge boost in terms of the information we have on each user. It’s worth noting that Adelphic takes user privacy very seriously and that none of the information we collect enables us to personally identify users – in other words none of it is personally identifiable information. Every time we get an impression from a user, we receive a lot of information about them – what site they were on, what device they were using, what time of day, geo, and so on. The ML then learns which users are the same across devices by reasoning: “the behavior of this mobile user is very similar to that desktop user, and beyond a certain confidence threshold, in all likelihood, those two users are the same person.”

ML in Targeted Advertising Moving Forwards

How do you see ML impacting the targeted advertising space? What do you think the path ahead is for those two applications of ML?

YA: We’re progressing through first phases of how this information can be used to target advertising. For instance, imagine that I am a company, and I’m running a campaign that’s trying to identify potential customers and target them with an offer. At the moment, ML is telling the company “these are the people you should reach” – but that’s where things stop. We manage the rest of that journey separately, often manually. Moving forward, I think more of the customer journey will be driven by ML. I think that this is what makes Vidora interesting, since these are the kinds of problems you guys are working on. I think we’ll see a more profound merging of adtech and martech.

Can you give me an explicit example of that?

YA: Sure. Adelphic is baking the location of a user into the optimization. One of the ways we do that is by tracking location through the app on people’s phones. To track users we use deterministic and probabilistic methods to link physical location to a person’s device – again respecting user privacy. For instance, Adelphic can run a campaign for a car company and then tell them whether or not someone who’s been targeted by an ad has since physically been to their dealership. With that, we can think about ways to leverage ML and AI to automatically change the messaging that person sees across a different set of devices.Next, the algorithm gest better at predicting outcomes and at managing that customer journey in a very powerful way – that would be a marketers dream.

On the importance of AI to the media industry moving forward

Time Inc acquired Adelphic, so naturally there would be a lot of thought being given to the implications for ML in the media space. What potential do you think publishers see in ML?

YA: I think that ML can be a real boon for publishers. We also all know the challenge to survive in the media space today. Google and Facebook are winning the data war, and know far more about the publisher’s users than the publishers do themselves. This is making it difficult for publishers to compete. I don’t know that publishers necessarily need to develop AI tools in house, but I think that’s where companies like Vidora come in – to help them make sense of their data and their users. That’s ultimately going to help them survive.

You can find out more about Adelphic here

Read more from our AI Spotlight Series:


Want to Learn More?

Schedule a demo and talk to a product specialist about how Vidora’s machine learning pipelines can speed up your ML deployment and ultimately save you money.