Modern Retail+ Research: How 3 retail executives evaluate what AI tools make sense for their brand

• Paul Michaux, co-founder and vp of digital product at customizable hair- and skin-care brand Prose. Prose’s AI-powered algorithms help develop custom products for shoppers based on their beauty needs.
• Kyle Simon, co-founder and COO of The Clear Cut, a direct-to-consumer jewelry brand. The Clear Cut recently developed Eugenie, a proprietary AI engine built on years of internal sales and customer behavior data. Eugenie helps the GIA Graduate Gemologists that The Clear Cut works with to analyze inventory movement, regional preferences and pricing patterns, among other trends.
• Sarah Davis, founder and president of Fashionphile, a luxury resale platform. The company has been around for 26 years and has developed its own predictive pricing algorithms.

How these brands use AI today 

Michaux: “I would say we probably [use AI] the usual way people use AI, which is CX – 50% of our tickets are now done through AI – engineering, marketing – all of these functions, from creating ads to [creating] better code. 

But what I think makes us unique is three areas we love to talk more about. The first one is around: How do we use AI to better understand our customer? We created more than 200 algorithms to approach part of that — to understanding their needs and their desires. [Then]: How do we use AI to create formulas and routines?  We’re creating from scratch this formula with a human in the loop to make sure that we have the best quality and safety, … and coming up with 100% personalized formulas. [And then]: How do we use AI to recommend the right products for users to use and to choose from, and then [build a] feedback loop? And then the final one is business strategies. We have more than 1 million formulas and more than 2 million [data points] of customer feedback. So, we use AI, of course, for customer clustering, insights, predictive modeling and all that.”

Simon: “We built, without AI, a propriety platform [where] all of our customers get matched with a GIA graduate gemologist and have a bespoke one-on-one experience where each diamond is specifically curated for them. And we looked up and realized we were sitting on a tremendous amount of data. … We operate in a very fragmented [space with] old school supply chains. We have built a lot of tools in the back end to make that more efficient, and now we have hundreds of millions of dollars worth of data on diamonds that have flowed through the system. And so we’ve built models to train them to, first of all, think like a gemologist — because there are just so many different data points on diamonds that are relevant to some and completely irrelevant to others. And, second, to make sense of that from a pricing perspective, from an optimization of sourcing perspective and from a customer experience perspective. And so we made the [AI] tool Eunice, named for the first female gemologist. We have been using it a lot on the back end, and we are starting to use it on the front end as well.” 

Davis: “We’ve been in the business for so long — 26 years. … When I started selling online, you could not process a credit card online. … We hired a company in 2017 to help us with our ‘big vision’ AI, to help us develop visual recognition technology and to develop neural networks starting in 2017. That’s not that long ago, in the grand scheme of things. But the way things have progressed over time is just amazing. 

It’s actually super interesting, because some of the issues we’ve always had are still here. How do you put together a model that’s going to help us identify an item, or how do you put together a model that’s going to help us authenticate an item? For some of the authentication tools we developed, you have to have data. [But] how much data are you getting for a bag that was released this season? It’s the first one we’ve got, so we have no data.

Something at Fashionphile we always focus on is HI and AI, meaning: How does human intelligence coupled with AI really get us to that next level?”

How they evaluate vendor pitches and what external tools to use 

Simon: “I really value when people [in my organization] come to me with a cost-saving idea. It’s like, ‘Hey, we have this really expensive photo shoot, but I think we can actually just like get the product shots and do a lot of stuff in-house. … Frankly, the amount of B-to-B software products — forget AI — that we’re pitched is kind of overwhelming. It takes a lot to pay attention to that. But when internally people are like, ‘Hey, I think I can save this group in the company money,’ it gets our attention.”  

Davis: “To my team, I’m like, ‘Please don’t go [to industry conferences]. Pick one, and then don’t go [to more]. Because all it is is people saying, “Here are free AirPods, … if you listen to our pitch. Honestly, one of the reasons I believe that we’re profitable is because we’ve been disciplined to say, ‘We don’t need all this.’ … Because a lot of times, what [happens is] we’re just adding an additional feature or an adjustment that actually doesn’t save us money. It doesn’t make us a sale. 

And at the end of day — let’s just look at the financials — what’s it really doing for us? And sometimes, it is totally worth it — in post-production, video editing is very expensive. We [look at this very granularly] and know we paid however many hundreds of thousands of dollars last year in post-production editing outsourced company help. Now, we just use this tool, and we’re saving all [this money] — it’s so easy. But you have to just look at [the financials] really granularly, because every SAAS company has got a new angle. … They’re now ‘AI.'”

Michaux: “I’ve been doing product management my whole life. … ‘Build versus buy’ is a pretty old challenge, right? Of course we [look at]: Can we save money with that tool? How easy is it to implement? How much are we locked into that? How much data are we sharing back with this tool, with [anything] proprietary or IP? We go through all of this, but at the end of the day, the good challenge for us is: Should we build this in-house? Are we the best to build that tool in-house? Is that building toward our moat or our uniqueness? Is that building toward better experience, better product? We build our personalization engines from scratch, of course, because this is really core [to Prose]. I’m really happy partnering with outside companies, when it comes to customer service and using these tools they built, because they raised millions and millions of dollars to do that, and I just don’t have to do it on my end.”