Human vs machine: when to use AI to guide creative decisions

Automated tools can now predict a digital ad’s in-market performance in as few as 15 minutes. Are traditional survey tools day’s numbered?

Where should AI solutions sit alongside survey-based testing? What are the strengths and advantages of each? And when should you use one or the other? 

Jane Ostler, EVP of Media and Creative solutions at Kantar, explains why there will always be a role for traditional ad testing even as AI becomes ever smarter and ubiquitous.

The success of ad campaigns is dependent on a whole number of factors – most notably the right creative approach coupled with the right execution. Without a good driving insight or idea, ads will struggle to engage or cut through. The quality of creative is in fact, the second most important factor in profitable advertising, and it also drives both brand equity and sales.

From and kicking off the strategy and ideation process, through to launching your campaign and optimising in-flight, getting creative right is a journey. How long it might take depends on whether you are launching a new landmark campaign or changing assets in a digital ad.  But ultimately the process is the same.

At Kantar, we recommend an iterative testing process to maximise your chance of hitting the right creative formula. So, for example, to maximise ROI, a TV execution should be tested three times: first of all, at an early stage, then again after the edits based on the findings, and finally at near-final stage, to refine. But what’s the best approach for digital? And what role should AI play?

The rise and rise of AI testing

Artificial intelligence (AI) ad testing solutions are attracting more attention, and budget, than ever before; 49% of global marketers plan to spend more on AI creative testing next year. And with good reason, given it promises quick answers and lower costs than survey-based approaches. It might seem like AI and creative strategy are strange bedfellows – after all what can AI tell us about something as subjective as how humans will respond to an ad? And how can a machine understand the merits of a creative approach or which version will have most impact?

AI-powered creative testing has a crucial role to play as a predictive tool for marketers. It can give clear indications on a creative approach and as it costs less than alternative creative testing methods, it opens the possibility of testing at high volumes, as well as testing competitors’ advertising – something that has largely been cost-prohibitive before now. But crucial to the output of any AI tool is the data which feeds it, and the assumptions going into generating those speedy indicative results.

You get out what you put in

So, when it comes to testing ads, it’s not all about AI. The advances in AI underline why and when marketers should use survey-based testing. That will always be needed to form the foundational data which the AI tool can draw. This foundational data is crucial, because an AI tool is only as good as the data sitting behind it. Here’s where scale matters, and quality. The data fed into Kantar’s AI tools for example, comes from a database of over 230,000 real-world ad tests.

For marketers, being able to identify the right tool for a job is key. For an indicative result at speed, and for testing high volumes of assets, AI is a great option. For a full campaign or new product launch, where you want to really unpick the nuances of how viewers react to your creative and why, a survey-based approach would be wise. Most clients we work with are adopting a framework for their different campaigns, depending on the job to be done, and are using a hybrid approach.

Keeping up with the demands of digital advertising

With the advertising landscape constantly changing, and as digital ads commandeer an ever greater share of media budgets, it is essential to use solutions which allow you to respond to the specific challenges of the fastest-growing, highest volume, most used marketing channels in the world. But hope is not a strategy. This is as true for digital advertising as it is for any other business decision. Digital advertising isn’t throw-away, it’s worth giving it care and attention to get it right. 

In response to the specific demands and pace of digital advertising, Kantar recently launched Link AI for Digital on Kantar Marketplace, which offers creative effectiveness predictions for digital video ads in as few as 15 minutes, assessing against the behavioural and creative metrics that drive ad performance. It gives marketers the ability to predict the performance of digital advertising before it goes to into the market, evaluate different versions of an ad, test competitors’ creative and test high volumes of ads to identify trends and build creative benchmarks. 

So, is AI the future of ad testing? Yes, but the future that we see is not one where AI supersedes survey-based testing. We envisage a hybrid approach working best for marketers looking to maximise the return on their creative.

By Jane Ostler

EVP, Creative and Media solutions