The Rise of AI in Programmatic Advertising

Elon Musk is all for it, Mark Zuckerberg warns of its dangers – and, as if to prove the point, Facebook recently had to shut down parts of their AI experiment after chatbots had become too clever and invented their own way of communicating. But science-fiction-fuelled horror scenarios of a future ruled by artificial intelligence aside, how does AI benefit programmatic advertising? Ignasi Prat (pictured below), CMO, Tappx, discusses what AI is contributing to the ad tech industry – and why the use of AI should be welcomed.

Artificial Intelligence (AI) is powering the ad industry’s transformation into the ad tech industry. This demonstrates why the rise of machines buying and selling ads in real time is more efficient than human ad operations management. Humans are unable to manage the vast amount of data provided through programmatic media for complex processing and atomisation.

The term ‘AI’ often evokes the world of science fiction with sentient robots taking over the world. In reality, AI is fast becoming an everyday term in the vocabulary of leading tech companies. Up until now, humans have done much of the thinking whilst machines have been responsible for mechanical tasks. Today, machines are the ones now doing much of the thinking. This isn’t a futurologist statement, but based on present day facts. Consider the incoming wave of autonomous vehicles or smart algorithms that can predict what products we will desire, even before we realise we actually need them.

Ignasi Prat, CMO, Tappx

The development of AI algorithms must be at the heart of any ad tech company that wishes to deliver the best solutions for publishers, brands, and agencies. By supercharging data and targeting, intelligent algorithms can empower advertisers and publishers to exponentially improve the effectiveness of their campaigns. Intelligent algorithms are constantly learning; they’re fully capable of modifying or customising their actions via their learned ‘experience’. Based on a particular context, we understand that intelligent algorithms take actions (or make decisions) to maximise a defined objective.

Different types of classifications and regression algorithms combine together to answer complex questions, without having all variable elements/data available. One of the commonly used intelligent procedures focuses on neural networks. These learn to recognise patterns and associate them with other patterns by forming and changing the connections amongst nodes. The significance for mobile advertising lies in the fact that virtually any pattern may be represented by the network. Also, the network can make these kinds of matches mathematically reasoned, even if the input data is incomplete, noisy, or otherwise imperfect.

People usually rely on experience and intuition, which of course isn’t too useful for the ad tech industry, which is normally characterised by changing audience needs, versus the best ads to serve, according to the ROI and the highest revenue for publishers.

AI can solve numerous challenges facing developers and brands including:

– Discovering the price ad networks are willing to pay for an impression

– In-app promotion, finding the probability of an impression converting into an actual app install

– Identifying the optimum times of day to serve an ad for target consumers

– Discovering the probability score for a user to engage with an ad unit if it disrupts their reading experience on a news site

– Calculating how likely a user who has similar apps on their phone is motivated to interact with certain products

Up until now, these challenges could only be answered ‘posteriori’. But to assume that future predictions will be similar to past events is a fallacy. Intelligent algorithms are now able to provide advertisers with invaluable predictive knowledge for a set of attributes that represent the interests and needs of the user to impact.

These can include the history of the latest ads that have impacted a user, how a user has interacted with them, the context of the impact, the web pages history, the list of applications that users have installed on their smart devices, and lists of items which have been purchased. All of these inputs influence the AI algorithm’s decision-making process, so one of the biggest challenges that ad tech companies face when trying to implement AI into their core technology (along with of having the right team of professionals) is to select and add new variables each time to create new patterns, and to train algorithms correctly to generate the best outcomes.

The implementation of AI technology is positive at all levels. All of the headaches associated with campaign management, such as assigning inventory and creatives, defining price levels, analysing, and reporting takes on a new dimension in terms of efficiency and effectiveness. These achievements can only be reached with the help of machines. The future belongs to them, so choosing the right partner will make a significant difference for the success of your campaigns, helping your brand make better decisions and support to maximise advertising revenue.