Welcome to the MediaGamma blog

The gap is widening between those with a handle on their AI strategy and those only scratching the surface of its potential.

AI in programmatic advertising: a 2020 update

Posted by Victor Malachard

AI Programmatic Advertising

AI in programmatic advertising: a 2020 update

To quote William Gibson: “The future is already here — it's just not evenly distributed”. 2020 is the year that the impact of artificial intelligence (AI) and machine learning in programmatic advertising systems becomes more noticeable. 

2019 was filled with major changes for the advertising industry. Spending on digital ads overtook traditional media like print and broadcast TV as Google, Facebook, and Amazon take an ever-larger share of budgets.

It’s forecast that 69% of all money spent on advertising in digital media in 2020 will be traded programmatically, up from 65% in 2019. There are big numbers involved. 

The total amount spent programmatically will exceed $100bn in 2019, reaching $106bn by the end of 2020, and will continue on to $127bn in 2020 and $147bn in 2021, when 72% of digital media will be programmatic. 

Programmatic advertising started as, and mainly still is, an auction-based system to allow the automated buying and selling of advertising space. Both advertisers (buy side) and publishers (sell side) are now moving to optimise the results they can achieve through programmatic advertising using AI technology.

This year, we will begin to see capabilities not only optimise bids for advertising space, but start to take marketing efforts to the next level by using predictive analytics to target the most receptive customers at the right time and place by leveraging AI and machine learning systems. 


A programmatic transformation of digital advertising

A huge proportion of advertising budget is now spent on programmatic advertising. Agencies and other players in the supply/buying chain have historically used a combination of human intuition and experience with algorithms and programs. With diligence, this type of solution will work — but it will only get you so far. The limitations really start to show with the increasing volumes of data and velocity of ad buys. 

2020 will see more systems capable of not just managing bidding, but altering target audience and content, based on the demographics, situation, interests and buying intent. Data can paint a picture of what the ideal customer looks like, and then spend the least amount of money to reach that customer on a site they are likely to visit.  

For example, a popular fashion brand (Diesel) included programmatic ads as part of a multi-channel campaign. Its ads were specifically designed for a precise targeted setting and outcome. When the website came up with a particular response, the system would automatically serve an ad which empathised with the user.

The more data you have, the better chances you have of homing in on the right audience. Data increases in value at an exponential rate when you are able to add additional, related data sets. To attempt this at scale, you need to build in AI and machine learning capability into your buy-side or sell side systems.


Why AI and Machine Learning take programmatic to new heights

Standard programmatic systems make decisions based on predetermined algorithms. Depending on how you define AI, this could definitionally be considered an “artificial intelligence” — check out our article on the differences between AI and machine learning if you want to learn more. But what these standard systems are unable to do is use data flows and past bid patterns to contextualise their decisions and improve independently of manual updates. This is what ML brings to the picture, and is the bedrock of the programmatic revolution occurring today.   

Machine Learning draws correlations that even an experienced marketeer would not see. However, the key point is the ability to learn from success and failure, adapting actions based on the patterns encountered. An AI-driven programmatic tool will learn about your products, services and market — honing in on a bespoke solution rather than having to work off of generic decisions.  

As a result, an AI can diagnose, predict and plan. Systems can teach themselves to become better in a certain area (media buying, for example) and improve their intelligence over time as they get more exposure to data.

In order to improve performance, AI-based systems monitor and constantly measure any number of campaigns against fundamental KPIs. If a campaign is underperforming, it can be modified, paused or stopped. If it is performing well, more funding can be allocated to maximise performance and optimise cost.

A system that uses AI might conclude, for example, that young women who have a pet and like a particular type of music are more likely to buy DIY tools online. The system will then make sure this audience will be able to see specific ads, reached through optimisations a buyer would never think of — paintbrushes on a pet site? However, when done right, this can allow you to circumvent competition and drive conversion. 


What the future holds

AI and programmatic advertising are the future. Players on both the buy-side and sell-side are making strategic shifts as consumer behaviour, digital regulations, and technical capabilities evolve at record pace.

Ad agencies like Saatchi & Saatchi and JWT are investing heavily in machine learning, and it’s beginning to emerge in programmatic advertising, audience and budget management, marketing forecasting, social media sentiment, brand safety and natural language e-commerce — to name just a few.

Having completed its testing stage, IBM is now working on collaborations between its iX digital agency division and using Watson for programmatic media buying. IBM is already using Watson for buying media in both the US and the UK, and the highest results show a 71% reduced cost per click (CPC) with the average around a 31% mark.

As artificial intelligence develops, AI and machine learning will scale capabilities beyond human capacity. Potentially, in a few years we will see hyper-personalised ads and even creative videos constructed on-the-fly. However, such advances are still a few years away.


Customer lifetime value

AI also has the number-crunching capabilities to make accurate estimates around CLV (customer lifetime value). This is a very powerful metric able to help retailers (and other businesses) determine the true value of customer acquisition — adding real context to customer acquisition costs. 

CLV is notoriously hard to predict. But with large data feeds, analytics software can (with shocking accuracy) estimate the order in which particular buyers will make purchases, how many purchases they will make, and how long they will remain a brand custormer. This type of technology is already being deployed in a purely analytical capability. Increasingly, however, we are likely to see this type of analysis coupled with programmatic automation — further optimising the types of bids being made using CLV data points.    


Getting the skills

A recent IBM research report reckoned 37% of the executives surveyed are concerned that limited AI expertise or knowledge is hindering successful AI adoption at their businesses. Other barriers cited include increasing data complexities and silos (31%) and lack of tools for developing AI models (26%). Another concerning statistic is that marketeers saw themselves as largely unprepared for digital, particularly with respect to evolving technology. 

Getting systems which are actually designed to leverage data and technology, and run orchestrated campaigns is still an aspiration for many. Most businesses do not have the in-house technical expertise to develop such systems. Increasingly, AI and machine learning expertise is being introduced by way of external consultants. Coupled with Machine Learning as-a-service (MLaaS) and API options, AI consultants are a powerful option at your disposal for upgrading your AI/programmatic capabilities and accelerate past the competition.   

Whether you look into supply-side platforms, demand-side platforms, real-time bidding or other systems, the essential first step is to get AI expertise and fast.

There are challenges involved in doing so. Ideally, it makes sense to have in-house capabilities for strategising programmatic ad approaches along with support from expert partners and consultants for building and implementing holistic and scalable platforms for programmatic campaigns. 

For advertisers and publishers alike, the coming year will present new opportunities driven by a customer-first approach. It will be important for advertisers to actively monitor and understand the big-picture changes happening in the ad-tech landscape as it produces new opportunities to connect brands with consumers. Programmatic AI sits at the heart of this change in 2020.

To learn more about applying AI to programmatic advertising in 2020, get in touch with us here. 

Get in touch

Victor Malachard
Victor Malachard
Executive Chairman

Related Posts

Google Privacy Sandbox: What it means for your programmatic ads

Google’s Privacy Sandbox looks like it might be the end of programmatic advertising as we know it, with third-party cookies largely set to become a thing of the past, and reliable automated data collation at risk along with them. The only way to prevent this from significantly unravelling online marketing efforts is to focus on shifting data landscapes, sources, and handling on a business-wide scale.

Is AI the right solution for your business: a 2020 assessment guide

AI adoption is on the rise at a rate of around 25% year-on-year. That’s an increase of approximately 270% in the last four years alone, and it’s a figure that set to continue rising as big data and tech potential continue to grow hand in hand. Still, despite this popularity, AI is far from a panacea for everything, as is evidenced by the fact that only 23% of businesses report using such solutions regularly. More disturbingly, a reported 85% if AI projects fail to deliver on their intended goals.  

How to train your own NLP model transformers

In recent years, search engines have led the way in natural language processing (NLP) and the benefits it stands to bring across business landscapes. Focused around machine learning capabilities, applications like Google’s BERT work to process natural language without human input. Used right, such systems can remove redundant work processes for faster and more efficient handling, analysis, and translation of written or spoken input from various sources.