Ipsy has built a business around using AI to match products with consumers.
The company, co-founded by beauty YouTuber Michelle Phan in 2011, is constantly looking to offer more personalized subscription boxes for its members as it learns more about what they like and don’t like. In order to achieve that goal, the company leverages its proprietary AI tool, dubbed Ipsy Match, which has been around in some capacity since the early days of Ipsy. The tool pulls data from feedback on past boxes, as well as a quiz that customers fill out as part of the sign-up process, in order to develop personalized algorithms. More recently, the company said it sees an opportunity to use AI further when it comes to helping people easily find beauty tutorials and advice they need on its platform.
Ipsy’s AI tool helps the company compile consumer sentiment data and inform product launches for its brand partners and in-house team. Sarah Rose, Ipsy’s chief product officer, said that the company’s ability to match shoppers with products has been instrumental in attracting brand partners and scaling the business. At any given time, she said the company has “tens of thousands” of SKUs.
In 2020, Ipsy acquired BoxyCharm for $500 million. At that time, the two companies combined had over 4.3 million subscribers and had generated $1 billion in revenue that year. As of 2022, Ipsy has raised more than $230 million in funding since its launch.
Rose spoke to Modern Retail about how the company uses AI to scale and the additional use cases it sees for the tool. This interview has been edited for length and clarity.
Apart from product curation, what other ways is Ipsy using AI in its operations?
AI has truly been a game changer for many brands, including Ipsy, on factors like internal productivity. We are leveraging AI throughout the organization to work better and work smarter.
So for things like understanding consumer sentiment, we have processes that are really dialed into our customers. Things that would have taken us hours and hours, now literally take us seconds because we have AI at our fingertips to help us understand sentiment over massive swaths of data.
We have, at any given moment, thousands and thousands of customer touchpoints, and it’s really hard for a human to go through and read them and pull out the most important points of that.
What are some of the steps the company has taken to improve its AI tools over the years?
In the early days, we were starting from more of a linear algorithm or rules-based algorithm. Then in 2016, really building onto that, [Ipsy introduced] more machine learning technologies that start to learn from itself and see patterns that humans cannot and bringing those technologies into Ipsy Match as well as into other areas of the business. We use [machine learning technologies] a lot in demand planning on our merchandising team to help us figure out what are the products that we should be sourcing and offering to our members.
Then, most recently in the last year, I think what’s so exciting is generative AI. Generative AI is really helping us, especially on the internal productivity front — everything from developer productivity, like software engineers, or even thinking about its application in our supply chain.
Your tool allows you to match products to customers who will most likely want them. Do you see that as a major draw for your brand partners?
Yes, absolutely. When we think about Ipsy as a company, we actually think about it as a beauty platform that connects consumers, creators and brands.
Those brand partnerships are incredibly important to us, and essentially, are a key part of the formula. What we’ve been able to do over time, is generate awareness for those brands of their products as well as provide them with incredible amounts of feedback and data so that they can understand how those products are being received and experienced by members.
For example, Ipsy has almost 300 million product reviews… and many of those reviews are sent back to our brand partners so they have an opportunity to learn a great deal about how their products are performing in the market.
The retail industry has grown more interested in applying AI in their business. How do you think this trend will continue to impact the beauty industry?
I think there are so many interesting applications for AI. The whole industry is in an experimental phase right now.
Some of the interesting developments that I have noticed is in the personalization of the product itself. [For example] in beauty, the formulation of the product, so leveraging a quiz or even biometric information to understand exactly what’s going on with your skin type.