Parham Aarabi, chief executive officer and founding father of Pre, took the stage on the Apparel & Retail CEO Summit to share the impetus behind launching his company, which helps retailers optimize their web sites based on simulating consumer behavior.

Aarabi began his presentation, titled “Optimizing Web sites for Maximum Impact: Predicting and Adapting to User Behavior,” by noting he founded ModiFace (now owned by L’Oréal) and has seen try-on technology evolve over time.

“There are recent technologies and buzzwords in the sweetness and fashion industries, but the important thing query is: Do they work?” Aarabi said. “And maybe a more relevant query is: Do they work in your unique brand and your unique audience?”

Aarabi said the impetus behind founding his recent company, Pre, was based on observations of try-on technology that usually performed “very otherwise on different sites, and sometimes different sites would have which have the identical demographic and similar audiences but would have very different conversions.”

The AI-powered Pre predicts conversion performance “based only on mockups and screenshots,” Aarabi said, adding that there’s no must implement anything. It’s done through the platform and mimics consumer behavior. There’s no need for live A/B testing or installing APIs.

“First, we start with screenshots of a site — you really take every page of the positioning,” he explained. “Then we decompose the positioning of each element and each image and each button, and so forth, for every of those. The AI then analyzes how likely users will interact with an item — including placement and size, but additionally user intent.”

The technology can run simulations using virtual AI shoppers to disclose behavior and preferences for various products and searches. Aarabi then demonstrated the technology by animating a screenshot of the product display page. The AI’s cursor moved about as an actual online shopper would navigate a page.

Aarabi said the “fun starts whenever you make site changes across the user experience and redo the AI simulations. And inside seconds, it tells you ways the changes impacted conversions. And through the use of mockups, you don’t should implement any code or anything. The AI tells how changes helped or hurt conversions.”

“All of that is contingent on the AI simulations being accurate — mimicking the actual human experience on the positioning,” Aarabi said. “So, accuracy is kind of key.”

He said when the corporate’s technology launched a couple of yr ago, his team and retail partners were on the early stages of truly mimicking human behavior. “So, accuracy, initially, was low,” Aarabi said. “But with every partnership, we kept asking, ‘What can we do higher and what mistakes will we make that we don’t make again?’ So, we learn from those mistakes, and we truly train the AI. And today, we discover that we are able to predict with 90 percent accuracy.”