What a 19-Year-Old High School Dropout Is Teaching Big eCommerce Brands About AI-Powered Product Recommendations

Tech has its share of dropouts who made it look like the stuff of champions — Larry Ellison, Steve Jobs, Bill Gates — it’s an impressive list. They grew up to reshape the world. As adults.

Add Depict.ai Founder and CEO Oliver Edholm to that elite group. Now 19 years old, he’s going his own way, too. At 15 his summer internship at Klarna turned into a software development job while finishing high school at night. At 16 he was off to Singapore to start a new chapter.

The Y Combinator alum completed a $17 million Series A in February and is wasting no time.

People on passionate business missions are often like this. In Edholm’s case, his breakthrough was the knowledge that 35% of Amazon’s sales result from product recommendations, but few (if any) competing eCommerce players have recommendation technology to rival them.

Speaking with PYMNTS’ Karen Webster, Edholm explained that the “related products” widgets served up by Amazon “use some form of artificial intelligence, and AI today needs a lot of data to function properly. Most eCommerce merchants don’t have as much data as Amazon,” he said.

“The ‘related product’ widgets you see on Amazon we democratize, [providing the ability for] great recommendations for all eCommerce stores.”

Depict.ai’s disruptive innovation is creating a recommendation engine with which any eCommerce site can integrate “to provide recommendations without as much data, and through that to create really high-quality recommendations and provide much better value.”

Absent a pond of data on a given shopper or group persona, recommendations are using AI to gain an understanding of consumers as they shop based on real-time observations.

Combining data on what a shopper browses and buys along with the right product information enables online users to understand “the eCommerce products on the site almost in the same way as if you went to the physical retail store and asked a person there for advice,” he said.

“That’s the level of understanding we try to translate into what we’re doing, and through that, though it’s very technical, you can require much less data and provide a better service.”

See also: 100% of Top-Performing Online Merchants Offer Product Ratings and Live Site Help

Recommendations Without IT Pain

Even after two years of digital shifting, some merchants still need persuading that technologies like AI can have a transformative effect on their eCommerce business.

To get around it, Depict.ai lets eCommerce sites use its recommendations-as-a-service platform for free, running a straight-up A/B test showing sales with and without recommendations.

Edholm said, “When we measure how many purchases are generated through recommendations, we’ve seen a 2x to 4x lift, which is quite a lot.”

Ongoing conversion measurement tells the retailer which recommendations are underperforming and adjust, constantly gauging the revenue recommendations are creating.

This not only does the front-end job of generating additional sales, but takes pressure off IT.

“eCommerce merchants have a lot of problems with IT resources,” he said. “What we figured out how to do is build technology and an onboarding experience where we can integrate without any IT resources at all from the e-commerce merchant’s perspective. This results in an experience where you can get this live on your site very fast and see that it works.”

If the merchant doesn’t see lift from recommendations within two months, Depict.ai steps off.

“There are a lot of learnings you accumulate once you work with merchants, and there are a lot of quick wins you can have. If you can’t properly show the recommendations in a user intuitive way in the right place of the website, then there’s not too much value. We come with both.”

With a client roster currently including “regional parts of Staples and Office Depot” as well as major Nordic region brands like Björn Borg Group and Ideal of Sweden, Depict.ai works with as much data as partners can provide, then let’s its AI solution watch and make suggestions.

As a for instance, he said, “We can get margin data for all your SKUs [stock-keeping units, bar codes] and help mix the recommendations to push, let’s say, higher margin or lower return products.” This would be especially useful for apparel sites where returns are a major issue needing a solution.

“You can figure out which items have the lowest returns and then push those,” he said.

See also: 90% of Acquiring Banks Say Using AI to Monitor Merchant Transactions Boosts Efficiency

Next up: Search

Strategic use of an AI-powered recommendation engine can help companies understand which sales approaches are working, which aren’t, and how to remedy the situation quickly.

Edholm said “a lot of [merchants] think they need to push more stuff. We help with [that by identifying] what gives the most revenue by helping the customer find the product they want.”

Specializing in businesses with high SKU counts, Depict.ai is operating primarily in the office supplies, fashion, furniture, home decoration and jewelry verticals, he told Webster.

In 2022, priorities involve moving beyond product recommendations to search.

Edholm said the company’s search engine is in beta with several clients now.

Expanding into search takes a wider look at the individual searcher, and feeds “that back into the recommendations and vice versa so it becomes a much smoother experience. And from the eCommerce merchants perspective, you can work with [fewer] providers.”

With the sheer volume of size, color, style, and two dozen other considerations going into consumer calculus, product recommendations are highly complex, and Edholm finds that eCommerce merchants are clamoring for a simple, effective platform solution.

“Our customers are very explicit about wanting this thing from us,” he said, “and we are very focused on ensuring we can provide it, since it’s a complex product.”