Double Verify’s Nachiket Deole on driving campaign performance in the age of AI

The third edition of the DATAMATIXX ASIA 2024 Summit and Awards was held on Tuesday, August 6, 2024, at the Novotel Mumbai. The premier event from Adgully brought together industry leaders to explore the frontiers of data, AI, and marketing evolution.

The event saw Nachiket Deole, Head Of Sales, Double Verify, deliver a keynote address on ‘Driving Campaign Performance in the Age of AI’.

Commencing his address, Nachiket Deole noting, “We have been hearing about AI in the last couple of years; no conference ends and starts without mentioning AI, every session so far has been talking about AI, but I’m going to discuss how we as a company are using AI in our products and how these products are helping us in delivering better performance and how application of AI is made possible. For instance, the headline – ‘What does a Chief AI Officer look like in the marketing world?’ – means the companies are now hiring Chief AI Officers, or on the path to hiring that skillset. In reality, there are specialized resources that are going to be required, which means the business is really serious.”

He pointed out Google is paying publishers to test and unrelease Gen AI platforms, and one of the largest publishers is going to pay to test Gen AI. Synthetic media producer is the latest role addressing Gen AI thing. Everybody wants Gen AI, but eventually, when the usage scales up one needs somebody to monitor that, put rules to that, and really be ethical in their approach when they deploy AI.

“Brands are reassessing roles to build new capabilities, thus everybody is experimenting, everybody knows there is a future, but everybody is scared of what the future holds. So, that’s where AI is and which is creating new challenges in the marketplace surrounding content quality, 95% of the content on digital will be generated by AI in the next 3-5 years. Now the question is with that scale of content, how the companies and publishers are going to monitor and monetize that content, because content can go either way, it’s not about just churning content, but also about maintaining the quality,” Deole maintained.

Continuing further, he said, “Editorially it’s not going to be possible to churn out quality content and maintain the quality of that content if one does not use AI. Massive usage of fraud is being seen, when fraud signatures or fraudsters are getting into the systems, and when the campaigns are being delivered, so how do we combat that? Then there is the scale and complexity – there are so many platforms, formats, and publishers who are going to use this. It is humanly impossible to kind of manage that scale and complexity. There are short-form, reels, long-form video, text, images, and unlimited complex opportunities.”

Speaking about how Double Verify is trying to mitigate all these challenges, Deole said that first is ‘Fraud AI’. “With machine learning models, paired with human oversight, the scale at which we are tracking fraud is now insane, we are probably tracking around 2 million frauds daily, but the challenge here is how AI is mimicking human behaviour. If we look at fraud a couple of years back, probably the behaviour was linear, Now a human accessing an OTT will spend around 15 minutes on one, 30 minutes on another, and 10 minutes on any other OTT. Now with AI, bots are mimicking that behaviour, so we need to have a tool to conquer that behaviour, detect that behavior, and classify it as a bot behaviour. That’s where we are moving as a company,” he added.

Elaborating further, he said, “Our bread and butter is brand safety and suitability solutions for our clients. For that, we need to monitor content, and classify content in the right buckets, saying whether it’s a brand-safe content or not. For that we need to understand the meaning of that content and those keywords, It is about that whole context and article in which the keyword appears. For example, an Apple can be a phone, an apple can be a fruit or a chemist next to you. How we as a company are going to classify Apple in those three different buckets is where AI kicks in. And that needs to happen at a scale that is unimaginable, where millions of articles and content pieces are coming up every day. So, machine learning powers our classification in video as well as other media. Thus, now AI can literally look at video, see different elements of that video, and then look at the risk factors.”

Another factor that Deole mentioned was predictive AI. He said, “Now with all the learnings coming in, how do we deploy it so that advertisers utilize that data and put it into action? That’s where predictive AI comes in. It drives campaign optimization and performance, and optimizes on your KPIs as well as costs. In one of the FMCG brands, 10% of the inventory was going on inappropriate content. Now this 10% was classified because of classification AI, it was identified that out of 100 impressions, 10 impressions were going on non-brand safe content, and 82% had lower block rate which was achieved by using authentic brand suitability solutions that are deployed before a bid is placed. This led to an 11% increase in viewability, which is massive, and an 18% reduction in authentic CPM, which translates to being brand-safe, viewable, fraud-free, and delivered in the desired geo.”

These are edited excerpts. For the complete keynote address, please watch below:

https://www.youtube.com/watch?v=qDZZZxiahKk