Puma wants to improve its brand. AI might help.

Puma said it has already deployed Google Cloud’s Imagen 2 model to help it craft personalized website images corresponding to the locations of some online shoppers. For example, shoppers in Japan may see an image of trail running shoes on the foothills of Mount Fuji, the company said.

Google Cloud Chief Executive Thomas Kurian said the tool is able to create a relevant image in several minutes, whereas the traditional process of conducting a photo shoot would require several days.

“It is truly more than an order of magnitude reduction," Kurian said. Puma declined to share specific numbers, but Kurian said, “On average what we see, it’s more than a 10 times speedup."

Puma CEO Arne Freundt said he sees these generative AI-driven personalization capabilities as less about cost saving and more as key to building up Puma’s brand in areas like the U.S. and China, where it historically hasn’t been as strong a competitor. Right now Imagen 2 is simply personalizing based on geography, but in the future it could also do so based on age and individual sport preferences, he said.

Since deploying the tool, Puma has seen improved click-through rates by 10% in India, but it declined to share results of other markets.

“We need to be more clearly anchored as a true high-performance, innovative sports brand in the consumer’s mindset," Freundt said

Puma’s AI push comes as the company and competitors such as Nike and Under Armour face increased competition from upstarts like Deckers-owned HOKA and On Holding, which have been able to build strong brands and show they have pricing power, said John Kernan, a research analyst covering the retail and consumer brands sector at TD Cowen.

Puma also has a smaller direct to consumer business than its peers—76% of its sales are wholesale—meaning it has not only less pricing power, but less of a direct relationship with its end customers, and less data on them, since retailers rarely share significant information, Kernan said.

Like other companies looking to collect and analyze valuable customer information, Puma has leaned heavily on its digital business. In March the company announced its initial partnership with Google, which included moving Puma’s e-commerce site onto Google Cloud’s infrastructure and leveraging some of Google’s data and analytics tools.

Puma in August trimmed its earnings outlook toward the lower end of its guidance range. In addition to higher freight costs and duties, Puma said subdued consumer sentiment, particularly in China, prompted the change in outlook. The new guidance came with results for the second quarter, when sales fell 0.2% in reported terms to €2.12 billion, equivalent to roughly $2.36 billion. Net profit dropped by nearly 24% to €41.9 million.

Brands that sell primarily through third-party retailers have struggled for years to learn about their customers, said Brendan Witcher, vice president, principal analyst at research firm Forrester Research. By offering consumers better, more personalized online experiences, they are also motivating customers to share information like an email address with them and even make a purchase.

Even small changes, like the geographically personalized website imagery on Puma’s website, can make a difference in terms of starting a dialogue with a customer, Witcher said.

“We are triggered very quickly by very small things," he said. He added that using generative AI to build more personalized experiences at scale is a long journey, and we’re only in the early innings. “Nobody is doing full-on gen AI. We’re not there yet."