Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More
According to Google, Meta and a number of other platforms, generative AI tools are the basis of the next era in creative testing and performance. Meta bills its Advantage+ campaigns as a way to “use AI to eliminate the manual steps of ad creation.”
Provide a platform with all of your assets, from website to logos, product images to colors, and they can make new creatives, test them and dramatically improve results.
For a small business with few design resources, this is a fantastic advancement. Imagine being able to develop brand-appropriate creatives almost instantly that follow social media platforms’ design guidelines and formats perfectly. It will make a huge difference for millions of small advertisers.
For large brands, it’s likely to be a very different story — and the reason is the “why.” AI can ingest information and spit out new assets. AI can also test creatives and optimize toward the creatives that are performing. But when it comes to knowing why a creative performs better than another, AI falls short. For any enterprise that highly values its brand, AI will play a different role.
Join us in San Francisco on July 11-12, where top executives will share how they have integrated and optimized AI investments for success and avoided common pitfalls.
Asking questions is a good thing
Give a media buyer the results of an A/B creative test, and I hope that the first thing they’ll want to know is why one performed better than another. Getting to the “why” is important in nearly every other aspect of a well-run business; why would creative be any different?
Few good media buyers can get away with blindly following test results without having a good answer for their client as to why one strategy, design or approach worked over another. And most CMOs are in the process of accumulating as much data-driven knowledge as they can to justify every dollar they spend.
The why is often very specific and very important. Take one example of two banners developed for a quick service restaurant with different variations of product and design. To an AI-driven creative testing algorithm, “burnt orange” stood out as a color associated with the higher performing creative.
This insight could lead to an optimization of banners to be predominantly burnt orange, which may or may not work because the orange color was actually a cup of coffee with cream. While not clear to the AI, it appears obvious to a person that the most performant banners have cream in the coffee vs. plain black.
Brand images are complicated
Not only do global brands have high quality and design standards, but few want to leave their brand strategy or reputation up to AI. Feeding assets into a machine and letting it rip can set the stage for a variety of issues.
Take, for example, the conundrum that advertisers have grappled with for ages: Whether to use “real” looking models in advertising versus overly polished, idealized versions of consumers. For a long time, studies showed that people reacted better to the overly polished types, so researchers assumed most people tended to be aspirational when it came time to pick brands and products.
But recently, a big movement has pushed advertising closer to reality. More and more brands are featuring models that more fairly represent their consumer base. Add to that the desire that many marketers have to more fairly represent the diversity of their customer base, which is not about testing for performance, but about correcting an inherent problem with the old norms.
Social context, implications
Would AI be able to weigh the pros and cons of which direction to take from a brand equity perspective? Certainly, the AI could create a variety of banners and test them, but the social context and the implications for the brand long term would be MIA.
There’s also the case of long-term vs. short-term campaign goals and the research that goes into making smart strategic decisions. Humans are still best suited to make these decisions and need to be part of the data-driven process, even if AI plays a significant role.
Deloitte finds that 57% of consumers are more loyal to brands that commit to diversity, for example. This finding may not be available to an AI performance algorithm at the moment they are testing creatives, nor may an AI algorithm have the ability to weigh the various inputs that determine the right balance of representation.
Helping AI get better
This isn’t to say AI isn’t helpful and, frankly, exciting. In fact, AI is revolutionizing creativity today for major brands and their agencies. Today AI can help with many manual tasks, inspire new ideas and directions, and deliver insights. Tomorrow it has the potential to be part of the creative process at an even deeper level.
AI may not understand the “why” right away, but we can get more out of AI the more we train it and interact with it. Telling an AI algorithm that the driver of performance is not actually “burnt orange” but is, in fact, “coffee with cream” is but one example.
Another is to input findings from larger studies about brand perception, sales and loyalty so that AI-driven outputs can be tuned to the metrics that matter to large enterprise brands. Finding ways to deepen an algorithm enhances that algorithm’s ability to be useful. The power of insights is not to notice a difference but to understand the “why” behind that difference and apply that back into the system to create a positive upward cycle.
For any business that cares deeply about its brand, AI will really come to the fore when it can work hand-in-hand with creative professionals, data analysts, brand managers, media teams and other experts that have the expertise and are empowered with context to understand the “why.”
Scott Hannan is SVP of corporate development at VidMob.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!
Read More From DataDecisionMakers