Global brands stick to AI’s testing basics… for now

AI Basics

Why it matters

AI is often over-hyped, so it is vital marketers can balance future promise with current limitations. They will also need to balance AI-driven content with human oversight.


  • At Heineken, AI is being used to speed insights generation as inputs for the creative development process.
  • Relatively simple applications, such as applying ‘fit for platform’ rules to ensure a campaign is suitable for each platform, have generated additional effectiveness.
  • At Mars, AI is being used to test creative ideas and potential content against what its years of neuroscience research suggests makes a good ad.
  • Mars believes the biggest opportunity of AI is speed – helping push out the volume of ad content that is needed for multiple platforms and devices.

AI promises revolutionary breakthroughs in helping marketers to better understand customer behaviour and to use new insights to engage audiences more effectively. Part of this potential centres on Generative AI being able to crunch though massive data sets to give creatives a steer on how to frame their messaging in future campaigns. Adverts storyboarded, created, tested and optimised by AI are mooted as just a matter of time.

While this full service of autonomous advertising may be a little way off, marketers agree there are early signs it is a realistic direction of travel. In the course of a recent Kantar webinar (September 2023), Heineken’s Global Head of Brand and Communication Insights, Sander Bosch, revealed the beverage brand is already experimenting with AI in ideation. The company is using the technology to experiment with how it converts customer data and campaign goals into creative ideas, and has so far been impressed.

 “AI has the potential to increase the speed of how we generate learnings as inputs for creative but also the whole creative process from idea generation to execution to action or testing,” he said.

“For now, though, (we’re using AI’s) speed in the form of insight generation, to look at vast amounts of data to generate insights as inputs for the creative development process, for idea generation, and even early-stage creative executions. It enables you to test and start experimenting before you even spend on any media.”

The appliance of science

At Mars, the focus is to use AI to test creative ideas and potential content against what its years of neuroscience research has revealed will typically make a good advert. Its Senior Director of Marketing Effectiveness, Sorin Patilinet, credits this deep understanding of how customers interact with adverts as giving its many brands an edge in advertising. By programming AI to understand the importance of how a message is being received, through inputs such as eye tracking, the company is edging towards being able to automate the process of integrating its customer data insights into content creation.

“We’ve been known to be explorers of neuroscience research in advertising,” Patlinet said. “It’s how we understood the importance of attention and emotion and various rules, such as branding in the first seconds, that we implement as much as possible in our creative processes.

“AI is helping us track that, it’s helping us measure compliance to our best practices that are derived from research. AI is helping us recognise emotions from facial coding, figuring out engagement and tracking of content.”

It is this process of using AI to ensure adverts comply with its deep understanding of neuroscience that he believes will be the most useful in the short term. Although there is a possibility AI will be able to create, test and optimise an advert autonomously, it is the testing stage where it currently offers the greatest help because it solves a very immediate issue around digital channels – there are now so many more adverts that need to be checked before being released. 

“The biggest opportunity of AI is speed,” Patlinet averred. “It’s about solving the madness we’ve all created in the advertising space in the last 10 to 20 years.”

It’s not so very long ago that an ad would be developed, tested and then put on TV, where it could run for the next six months. “In today’s world, you have about three hundred ads starting tomorrow, and then you need answers for all of them. It’s not humanly possible to do it without AI,” he said. “That’s why I think AI is needed – to get the breadth of content that we push out there, to get knowledge on that content and to help us make better, more automated decisions.”

The human-AI hybrid

Whether, or how soon, this pre-launch testing extends into optimising live adverts on the fly is up for debate. Heineken’s Sander Bosch has never been persuaded that the time and effort required to optimise adverts in real-time has even been worth it. He suspects many marketers share his experiences of optimisation taking up a lot of resources and often not being worth the effort, particularly as the feedback and necessary adjustments take so long; a campaign can be nearly over by the time it is improved.

“I’ve been a bit sceptical of the early-stage dynamic creative optimization, for us as a company that is not very focused on direct response where you have a very clear signal that you’re optimising against,” he explained.

However, if brands get a better idea of what they need to optimise for, in terms of brand attributes and not just direct response, he believes there could be a future in AI optimisation. “As AI gets smarter I expect that the doubts will go away and we can become more comfortable with using AI to optimise creative on the go.”

Heineken has so far got the best results from applying ‘fit for platform’ rules to ensure a campaign is suitable for each platform, such as YouTube or Meta. It is a “relatively simple” use of AI, Bosch admitted, but it is highly effective and embedded across the business. Now, he suggests where marketers are likely headed is finding a ‘hybrid’ testing and optimisation.

Relatively simple applications have generated a lot of additional effectiveness in our advertising.

“We’re still at a very early stage, but even these relatively simple applications have generated a lot of additional effectiveness in our advertising,” he reported.

“AI has helped us greatly in raising the floor on our creative performance. So we’re currently also looking at how we integrate that into our protocols to find a hybrid between AI and actual consumer retesting.”

This hybrid approach is the most likely next direction of testing and optimising campaigns according to Steve Silvers, Executive Vice President, Creative & Media Solutions at Kantar. He forecast that AI’s adoption will continue at pace because budgets are increasing for creatives, but so too is the pressure on teams to deliver ROI in their spending.

AI can currently deliver a great deal in terms of time savings but is not fully trusted to optimise on the fly. To understand how adverts are being received, human panels and human oversight are likely to be a key component of a hybrid approach, he predicted. ROI will speed up the process and deliver savings but people will likely still be relied on for feedback.

About the author

Sean Hargrave is a freelance journalist covering digital marketing, technology and business.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *