LAS VEGAS – Future Experience Cloud releases could use artificial intelligence (AI) to guide businesses to their dream customers, or automatically optimize videos based on their target audience, if the latest Adobe Sneaks are anything to go by.
During the San Jose, Calif.-based software maker’s annual marketing conference, Adobe Summit, researcher Vishy Swaminathan presented a feature he called “Video Ad AI,” which predicts a video ad’s success based on the channel use and audience sought, and automatically adjusts its length to optimize its chances of success.
“Wouldn’t it be cool to know how successful your campaign will be before running your video ad?” Swaminathan asked the audience during Sneaks, a popular fixture at Adobe conferences in which researchers demonstrate proposed features still in development.
Like many Sneaks presented Wednesday, Video Ad AI is powered by Adobe’s artificial intelligence platform, Sensei, which Swaminathan programmed to compare two types of content: a 60-second TV spot-turned-social media ad, and video ad content exclusively produced for social media. Each was assessed based on its past performance and awarded an effectiveness score.
In the case of Swaminathan’s sample, a travel ad that started with kayaking, followed by pans over beaches and closeups of physical activity before ending with a sunset, Video Ad AI concluded the ad would be effective on television and Facebook, but not Instagram, where its “watchability” was estimated at just 20 per cent.
Fortunately, Video Ad AI also suggested recommendations for improving the TV-turned-social media ad’s effectiveness score, such as changing the video’s length and flow, and automatically sent its own attempt to Premiere Pro for editing. Advertisers would then be invited to adjust the ad as they please, Swaminathan said, newly aware of how their efforts could affect its likely performance.
“Several factors that could affect [its watchability score], but in this case I think it’s just too long for my Instagram audience,” Swaminathan said. “So I can choose any duration I want, and… [say it’s] 15 seconds. When I tell it to summarize, Sensei uses a third-generation neural network to create a 15-second summary for me completely automatically… with absolutely no human intervention.”
We can’t reproduce it here, but the 15-second version of Swaminathan’s sample ran smoothly, complete with a call to action that cut out just before it showed the website he wanted viewers to visit.
But that was easily corrected, he noted – and still saved him time.
“I still want my creative team to have final say on the summary,” Swaminathan said. “I’m still happy because it has saved a tonne of work for me.”
More importantly, according to Video Ad AI, its effectiveness score on Instagram was now 95 per cent now, with 90 per cent confidence.
Build the ideal customer journey
Another Sneak worth mentioning from Wednesday was researcher Pankhri Singhai’s “Perfect Path,” a proposed Adobe Experience Cloud feature that uses Sensei to automatically generate a personalized journey optimized for every customer targeted by a given business.
“I can send everybody an email, but that may not be the right strategy,” Singhai explained. “I may prefer email. You may prefer push notifications. And Leslie” – here she addressed Adobe Sneaks co-host and Saturday Night Live castmember Leslie Jones – “you may prefer push and email both, depending on the device that you are using.”
To identify which method a user might prefer, Singhai’s program used Sensei to first analyze her target audience, then mapped a series of potential customer journeys onto an “experience graph” before the audience’s eyes, before recommending the path deemed most likely to lead to engagement and conversion.
“What you’ll see [is], Sensei went through lots of user data, saw where you went on the client’s website, what products you clicked on… what emails you clicked on, what kind of communications you received, and finally came up with a journey,” Singhai said while leading the audience. “The journey that you see up here is a specific journey for a particular user segment, and it’s one of hundreds of journeys that Sensei is suggesting… this particular path is just the one that is contributing [most] to the conversion.”
Changing her target audience – to, say, students, or yound professionals – produced a different set of potential customer journeys, and identified the content they would most be interested in as well.
“As a marketer I can understand, and see, and feel confident about [my decision],” Singhai said. “I can update the workflow, and go back to the same canvas, and save my journey, and I am done.”
Each of the seven Sneaks showcased on Wednesday, including an automated tagging system and a feature that could extract commercial opportunities from photographs, was evaluated by the audience. Singhai’s proposal landed in fourth place; Swaminathan’s came in third.