The marketing discipline known as dynamic creative optimization (DCO) offers the capability to instantly assemble and deliver an ad crafted to be the most effective possible for a given individual.
To execute DCO, marketers must create multiple versions of each key component of an ad, each designed to target a different demographic. For example, there might be various fashion models to represent different types of people, different versions of a product, varied scenes and settings, or even a number of color schemes for text, backgrounds or other elements. Let’s say that a browser cookie helps detect a user who is a college student. The system could then choose to insert images of younger-looking models into an ad, models with whom those students should identify with more than they would for models of an older demographic group.
At its best, DCO can also filter for even more fine detail, such as for region or ethnicity. The right creative elements can add measurably to an ad’s positive effect for hundreds of pre-defined segments of potential customers — for example, for schools based in more northern U.S. climates and those based in southern beach locations.
“The creative discipline has shifted from attempting to embody the brand’s essence in a single expression to creatively leveraging data signals and available content to craft a near one-to-one conversation with an audience,” writes Matt Sweeney, CEO, North America, of Xaxis, an advertising technology company that has gone heavily into DCO.
A limiting factor for DCO has been that it requires people involved with creating the campaign to not only conceive the possible segments and variations of the advertising’s creative elements, but also to understand how to execute all the possible permutations and combinations.
Unilever, for example, created a video campaign for Axe body spray which divided its target audience into four segments, each of which could be given variations on six of the 11 scenes in in the video, according to Ad Age.
Enter Artificial Intelligence
Imagine, though, if artificial intelligence could be used to help glean not only who a given person is being shown a message — as well as their location, previous behavior, known preferences, type of device and other relevant factors — but also search for and assemble the ad from elements that had not been pre-conceived.
The AI would be able to not only assess what components might be most effective for that individual, but also find the pieces — or the pieces within pieces — that could be chosen from the open Web and assembled into that seamless whole.
The AI would even be able to understand usage rights and make many of the decisions a photo editor might, but much faster and at a tremendously higher scale.
Suppose a college student was detected consuming media on her phone in Rio de Janeiro, and she was Portuguese speaker. Today, an ad for, say, a summer beverage might show her an image of college students who look American, on a beach in Miami, because of limited creative elements available.
Adding in AI, the machines could search the web for images of a beach in Brazil and Brazilian-looking models, creating a new sub-segment just for this user and assembling an ad that would resonate more strongly for her.
Tag lines she’d be shown in the ad — in Portuguese, of course — could be based on what’s trending in social media that’s seen as relevant, which the machine also understands and integrates.
Much of this kind of image-finding and decision-making capability is “only inches away today” for artificial intelligence, says Derek Wise, CTO of ad-tech company Grapeshot. Wise says he expects to see this kind of AI at least tested before the end of next year.
AI is already being deployed for everything from facial recognition to identifying specific brands of handbags in images for eCommerce purposes, as Microsoft’s Bing does, to choosing scenes for a horror movie trailer — with creepy results, Wired notes.
“We might collect data on what colors an individual likes or how they respond to animation,” Xaxis Ad Labs’ Tim Bagwell tells AdExchanger. “If I have 15 different audience profiles around you … which one is most meaningful to this impression, and how can I lay out creative for you in a way that is more engaging and relevant?”
AI Now is ‘Supervised’
Today, machine intelligence may be applied to helping choose the elements that go into an ad, but executions of AI in DCO are still what’s known as supervised, as requiring people to instruct the machines on what is and isn’t relevant so they can start to learn how to make decisions.
Increasingly, though, scientists and the companies they work with are applying layered neural network technologies which have the capability to learn on their own without human input. If you’ve used the object recognition function in Google Photos, you’ve employed this kind of technology.
That use of intelligence combined with increased processing power and network nodes means we’ll soon see the intelligence from machines operate in ways that better enhance what the capabilities available today.
Through the application of AI and DCO, marketers may soon be able to reach, engage and inspire college students — and many other target groups — with more efficiency, scale and effectiveness.
Why Offer a Student Discount?
A student discount can nurture lifelong brand loyalty in a rising consumer group. Download this ebook to learn more.Download