Self-Driving Campaigns And Generative AI: Are We There Yet?

In 2018, I was invited to join WPP STREAM and their clients in Yogyakarta Indonesia. We were all asked to come up with one big idea that we then had to present at the event to an audience who wanted to hear more about it. We essentially had to compete to get attention by coming up with something interesting, exciting and yet believable. This did not sound easy in a large group with some of the best marketers and agency folks in attendance. It was also a little stressful as I wondered – what if no one was interested in what I proposed as an idea?

I walked up to the whiteboard and put my name down and wrote next to it: “The rise of the self-driving campaign.” Given the event had a lot of marketers and agency folks I figured it would evoke some kind of response given the implication that in the future campaigns would essentially run themselves. It did and I had a very well attended session with some very curious people!

My premise at that point was that increased automation via AI would significantly reduce the need for the extremely manual and time consuming processes that are involved in launching marketing campaigns. This, I said, was especially true as brands strive to deliver personalized experiences across many channels.

At that time marketers were experiencing an explosion of new social channels, formats, data strategies, cookie deprecation and COVID hadn’t even happened yet! With COVID of course came another set of challenges: budget cuts, fewer personnel, and attrition of tech capable folks as overfunded silicon valley companies were on a hiring binge paying whatever they needed to spend to fuel their growth.

All of this meant more work and less people available to do it. Clearly something had to change. Meanwhile, CMOs were demanding more precise, personalized and relevant campaigns to improve consumer engagement and media performance.

The time was ripe for automation and indeed the self-driving campaign. A lot of automation has gone into media buying over the years. DSPs have for many years now really leveraged data and algorithms to better target and reach the right audience for each campaign. Compared to a decade ago, today, media buying is a significantly automated process requiring very little human effort with the DSP platforms (and media platforms like Meta and Google doing a lot of the heavy lifting). The same is however not true for creative and content.

In a recent analyst briefing, I was told that the volume of inquiries related to Dynamic Creative Optimization (DCO) and Creative Automation has increased significantly due to the fear of “signal loss” as media platforms lose their ability to precisely pinpoint and target users using 3rd party cookies. This means brands have to sharpen and increase the precision of their content to make it more relevant and precise or deal with significantly lower media performance as cookies go away.

Most automation around creative optimization has centered around creative automation (i.e. creating ads at scale – multiple versions, languages, offers etc.) and on precision of targeting (i.e. decisioning using 1st party data) and optimization (A/B testing and multivariate testing). Going forward AI will have a critical role to play in realizing the vision of self-driving campaigns.

Generative AI is a game changer here as it enables automated content, creative and even idea generation. This automates one of the most manual parts of DCO (i.e., building creative assets from scratch. Generative AI has the potential to significantly reduce the dependence on custom photoshoots, video shoots, graphics design, copywriting etc).Today, this is largely done by humans. A large part of the industry already relies on stock photography and video so in some sense outside of very high value branding campaigns many marketers and agencies are already repurposing stock content. Generative AI takes it a step further to actually automate the creation of such content. There are, of course, issues of brand safety and copyright protection that the nascent Generative AI industry is dealing with. But, like any new breakthrough technology from streaming to social media, the early challenges eventually got resolved simply due to the demand in the market and the imperative for the producers of such technologies to address those issues.

AI also has played and will continue to play a key role in creative optimization. Dynamic and personalized ads are very data rich. Unlike TV ads or print ads which are one version for all, a single dynamic ad campaign could contain thousands and even millions of variations of ads. The amount of data generated by these ads is staggering and excellent fuel for machine learning and AI algorithms.

Here at Jivox we invested in AI and machine learning way before it was cool, introducing Neuron in 2017. Since then our team has continued to improve optimization algorithms that ensure the right creative and content is delivered to the right person at the right time. This increases relevance at scale and it is not unusual for Jivox customers to see anywhere from 30-300% improvement in ad performance as a result.

What’s new and exciting about Generative AI is its ability to be “prompted” by humans and ability to “converse” – this creates powerful new models that can help automate DCO even further.

The vision of self-driving campaigns, much like self-driving cars, is not very far off and brands would do well to work with platforms that are investing heavily in AI as the campaign of the future is going to be highly automated, require fewer people to operate and if not self-driving, at least self-optimizing.