This is the first story in a trilogy on how artificial intelligence is affecting the work agencies do. The next installments will publish on Wednesday and Friday.
Xaxis, a partner of dynamic creative platform Jivox, is leveraging the latter company’s newest product, released on Monday: an AI-based dynamic creative content recommendation engine.
Xaxis will layer on its custom audience segments to make real-time recommendations more personalized for consumers.
“Jivox allows us to move from audience targeting into audience engagement,” said Tim Bagwell, SVP at Xaxis Ad Labs, the company’s dynamic creative unit.
Jivox’s core technology collects contextual and site-level data about how people engage with product ads online. It uses that data to dynamically assemble creative for lookalike audiences.
The new product, dubbed Neuron, makes that process more personal by allowing marketers to layer on their first-party data to tailor product recommendations to custom audience segments. A makeup retailer, for example, might recommend a lipstick rather than a blush for a certain audience based on its characteristics, even if the blush is a better-selling product.
“You can make that more intelligent decision about the product not simply based on pure counting,” Bagwell said.
Today, Xaxis is using Jivox’s product recommendation engine for three ecommerce clients, which Bagwell calls an “easy use case.” But Xaxis will leverage the AI tool for other forms of display.
“We might collect data on what colors an individual likes or how they respond to animation,” he said. “We can leverage Jivox to understand on an individual level what type of layout to recommend.”
Neuron will allow Xaxis to serve dynamic creative much faster because it stores data in its own memory as opposed to in a database. Bagwell compares it to a RAM drive in the computer, which stores information on its hard drive.
“It’s much faster than using data storage,” Bagwell said. “That’s definitely helped us on performance.”
Bagwell’s team likes Jivox because it’s willing to build for market needs. Xaxis was influential in putting the content recommendation engine on its road map.
“They’ve grown their technology with us, listened really well to our marketplace needs,” he said. “[It] sounds like we’ll be in a place where Jivox is providing a framework where we can bring our own AI to the platform to use for our own purposes.”
But Xaxis doesn’t discount other vendors, like IBM’s Watson, which can also execute dynamic creative.
“They’re not mutually exclusive,” Bagwell said.
Dynamic creative is just one use case for AI at Xaxis. Its Co-Pilot initiative was launched in July to improve campaign performance and operational efficiency using advanced modelling and automation to value media impressions, the company wrote in a blog post.
Today Co-Pilot is Xaxis’ internal AI development team, which builds, buys or borrows to develop AI applications. In programmatic, machine learning allows marketers to take in real-time data to determine the value of a particular impression in the given moment and environment it’s being bid on.
“Xaxis is already heavily invested in the AI research space,” he said. “We’re using audience data, which is information we believe we know about you, to decide whether or not we want to address you, and then layering on custom algorithms.”
With platforms like Jivox, Xaxis Ad Labs is trying to bring a similar real-time decisioning function to creative assembly. The team develops models around how different audiences interact with pieces of creative. For example, some users might like to engage with an interactive ad while others prefer static, Bagwell said.
“If I have 15 different audience profiles around you,” he said, “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?”
