Learn more about ALDO’s in-house generated AI and machine learning strategy

Machine Learning


In the midst of the artificial intelligence hype cycle, shoe and accessories retailer ALDO is working to build its own in-house AI foundation in the hopes that the tool will evolve from a wishful hypothesis to a mechanism that delivers business outcomes.

Last October, the company hosted its first AI-for-retail hackathon in collaboration with Montreal's McGill University and Amazon Web Services. The hackathon led to plans to revamp ALDO's search capabilities and improve product recommendations. The company remains focused on growth in those areas, said Fatih Naebi, ALDO's vice president of data and AI.

“Machine learning and generative AI are things that people are getting serious about,” he says, “but they all require a good foundation from a data perspective to be able to bring it all together.” [the functionalities].”

Today, ALDO uses everything from machine learning to generative AI, including predictive AI, which uses machine learning to forecast future events. Like other companies, generative AI is still in its early stages and is used to generate text, SEO, and product descriptions, while predictive AI (which he says is in development) is used to forecast demand and sales, optimize discounts, and more.

Building and training these machine learning and generative AI capabilities requires data, and Naebi said ALDO's data goes back at least five years to aggregate customer patterns like website clicks, in-store purchases, etc. To keep the data safe, the company has its own data clean room that aggregates insights “that can be applied to anyone,” rather than personally identifiable information, he added.

“We've built all of the foundations to run all of these data and AI products,” he says. “We're already running our retail e-commerce supply chain based on that data, and now we're leveraging that same data to make recommendations around demand forecasting and other insights.”

Simply put, data from ALDO's in-store interactions and e-commerce business is collected, anonymized, integrated and used to power AI and machine learning models for product recommendations, demand forecasting, sales projections and more, Nayebi said. The company declined to share specific numbers on how AI is impacting its business, but Nayebi said it expects to save on marketing costs as things progress.

No matter how slowly Google’s third-party cookies die as data privacy efforts continue, ALDO aims to “future-proof” its datasets, relying on its own first-party data to feed the so-called AI beast.

“We know that tracking and things like that aren't going to be something we have to rely on going forward,” he said, “so what we're looking to do is bring insights based on aggregate levels of customer patterns.”

Amid the boom in generative AI, some companies are looking to use AI to gain immediate traction, while others are looking to use it to solve efficiency problems where AI has potential, but it remains to be seen whether it will become a mainstream product that delivers the desired results.

“Every company is feeling the pressure to figure out how to incorporate generative AI and machine learning into their processes,” said Brian Yamada, chief innovation officer at VML. “The question is how much they're going to focus on building in-house systems for an AI future, or working with external partners like OpenAI or Microsoft AI,” he added.

“It's going to be a mix of build and buy. The speed at which the market is moving is an issue. It's hard to keep up, so it's a big challenge,” he said. “It feels like we're spending our time drinking from an AI firehose and trying not to drown.”

Freddy Dabagui, managing director of activation at advertising agency Crispin, echoed similar comments.

“Brands are leveraging their own tools and systems to gain more control over data privacy. These in-house teams are also building AI task forces, but they are not moving as quickly as marketers are,” Dabagi said in an email.

Yamada added that it's important to remain flexible as AI changes itself and its most efficient use cases become clear as the hype cycle continues. Building an internal system to collect data, train models and create internal processes, as ALDO is doing, can be a step in the right direction even as the situation expands and changes, he added.

“The key is to be flexible and not be too fixed or rigid,” he said of the future of generative AI. “We're moving to a much more fluid environment.”



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *