Creative is an inherently human endeavor. The imagery and psychology behind advertising is all predicated on how it will relate to consumers on an emotional level. But that doesn't mean it can't benefit from a little AI ingenuity. An integral part of creative strategies is personalization, but there's no need to manually produce multiple versions of the same campaign to test on varying audiences.
When we launched IBM Watson Advertising Accelerator ("Accelerator") in January 2020, it was designed for display advertising, but in the past year we realized that this technology can help bring media and marketing creative to scale. We have recently expanded it across platforms, including video and over-the-top (OTT) internet-based video channels, which is significant considering that streaming consumption has now overtaken traditional TV viewing. Because of this, in a survey conducted by the Interactive Advertising Bureau, media buyers said they are allotting a solid fifth of their overall budgets to digital video.
Of course, video and OTT assets are much more difficult and time-consuming to create variations for, but it is possible to take the manual work out of dynamic creative setup with the help of AI. A brand or agency provides its video assets—video sequences, voice-overs, music, end cards, call-to-actions, etc.—and IBM organizes them into themes to feed into Accelerator. Using these themes, IBM Watson machine learning analyzes hundreds of data signals such as weather conditions, type of device, time of day, day of the week or geographic location that surround the ad call. Then, in near real-time, Accelerator serves unique creative spots to each audience based on the likelihood of them taking the desired brand action, such as watching a video spot in full or secondary actions like visiting the brand website or making an online product purchase.
AI can make this all possible, allowing brands to quickly create and test up to 50 versions of a dynamic video ad, as opposed to the two or three ads brands often have available for linear television or OTT spots. AI can help marketers predict and generate impression-level ad variations, recognize what works best against the brand's goals, and use those learnings to predict the creative that will resonate with the next consumer, all in near real-time. It is campaign optimization at the highest level.
Keep up with industry upheaval
Marketers have a lot to contend with at the moment. With changes to identifiers and the upcoming deprecation of third-party cookies upending digital marketing, it is going to become increasingly difficult to personalize creative advertising. Fortunately, streaming and OTT do not rely on cookies today, but these changes have spurred conversation around the need for privacy-forward solutions across platforms.
We know businesses are working on developing their own solutions, or searching for products that will help meet both advertisers and consumers' needs. Accelerator is already a cookie-free solution—it leverages a variety of data signals at the time the ad is served to optimize the creative without relying on potentially problematic personal identifiers.
How? Clustering consumers into cohorts based on who reacts similarly to various creative variations. AI is able to identify trends in the data in seconds to predict which audience attributes resonate with certain creative variables. This is the type of advanced machine learning that allows brands to understand the complexities of their audience in a more privacy-forward manner. Brands can then be more flexible with their creative approaches and be far more granular in the individual creative elements that are served to an audience.
Derive creative insights to help improve performance
Accelerator's ability to optimize creative for each audience results in creative insights for brands that are often not available for video. A brand can then understand how each of their creative assets resonates with the different audience cohorts identified by the AI, often going a level deeper than any preset targeting on the media plan. This reveals insights such as which brand and product positioning resonates best with different demographics, and which creative variables work best for unique combinations of data signals such as weather conditions, type of device, time of day, day of the week or geographic location.
The majority of Accelerator campaigns actually discover that there is no “low performing” creative in the mix, but instead sheds light on the importance of delivering the right creative variation to each audience. Our initial Accelerator for video test campaign on The Weather Channel digital platforms (an IBM owned-and-operated digital property) exceeded video completion rate (VCR) benchmarks by almost 20%, showing that AI can have a real data-driven impact on creative and help brands drive improved performance.*
Using AI for good
Last summer, we announced “Call for Creative,” an initiative that offered complimentary access to Accelerator for select brands to use for their purpose-driven campaigns. Among the participating brands were the Ad Council, which used Accelerator for its Love Has No Labels campaign's “Fight for Freedom” initiative. The display campaign generated 81 creative ad variations and saw a 113% lift in click through rate from start to end of the campaign, along with a 69% increase in conversions on the Ad Council's website. Based on this success, the Ad Council is currently leveraging the technology for its “It's Up to You” COVID-19 vaccination education campaign.
Overall, the five participating 2020 Call for Creative brands saw an average increase of 120% in creative performance from campaign start to finish, which is why we've announced year two of the program in 2021, awarding up to five brands use of Accelerator to run "creative for good” campaigns that address pressing world issues.*
Yes, predictive technology and the ability to home in on what's going on can help drive a particular campaign. But we also believe in the creative, human benefits that AI can foster. Why work within smaller, more frequently trod boxes when you could expand your ideas in a more innovative, prolific way?
Perhaps those unproven concepts that might have fallen on the cutting room floor were just the sort of inspired takes that consumers would respond to, if only there were the time and resources to test dozens of options. With new video and OTT technologies, we believe marketers can cover more ground than ever. And you can too.
*The client examples cited are presented for illustrative purposes only. Actual performance results may vary depending on specific configurations and operating conditions.