E.l.f. Beauty’s chief digital officer shares her strategy for the AI era
As every brand and retailer grapples with the implications of artificial intelligence, they’re having to ask themselves tough questions about where they will and won’t use AI, and what their guiding principles are as they experiment with seemingly endless use cases.
E.l.f. Beauty’s guiding principle is simple, according to its chief digital officer, Ekta Chopra. People are its purpose and, therefore, even as it uses AI for more and more tasks, it will always keep humans in the loop.
That means E.l.f. is more hesitant to use AI for creative generation. Chopra said E.l.f.’s customers will “call us out” if they feel like E.l.f. Is putting out content that isn’t authentic. So, the company is more focused on how it can use AI to alleviate the workload of its existing workforce.
At Shoptalk Spring, Modern Retail caught up with Chopra to discuss E.l.f.’s AI philosophy, its strategy for training employees on AI usage and more. This interview has been edited for clarity and length.
To start, can you give me an overview of how E.l.f. is using AI today?
“I’ll start with a statement: At E.l.f, everything starts with a purpose, and out purpose drives the people. …So, [within]that framework, we want to do AI with a purpose, and the purpose is that the human will always be the conductor of what we do.
We think about AI as a tool, and you still need the human to, sort of, give it the context and ask it the right question, and so forth. So, that is the No. 1 overarching thing — that everything is done from that framework.
Within that, we have four [areas of focus], or pillars. One is: How can I make the human more productive? [The point is] to make their job easy, so they’re not having to do certain tasks. Right when they come in on Monday morning, everything is ready for them, and they’re actually using the time to build connections with the team and being part of the culture that is so special at E.l.f. The second pillar is to really reimagine every single process in the organization: How can you think about it with AI? So, for example, we did a listening tour with our creative team. We were like, ‘Tell us your pain points.’ We got the team together, and we invited the Google team because we’ve selected them as our LLM. [We got] 100 use cases [out of it], and it was all just simple things, like, ‘I want to do a trademark search.’
The third component, which is very important as a retailer, is the whole agentic commerce piece. So: How do you show up in the LLMs as the brand authority? That’s another pillar of work. And lastly: How can we really think about end-to-end autonomy within finance, when you’re closing the books and stuff? So, it’s a very operational focus. Those are the ways we think about it.”
You mentioned it’s very important to you to always keep a human in the loop. Can you walk me through that? Maybe a specific process where you’re using AI, but you’re still able to keep a human in the loop — how do you do so?
“We have a lot of retailer [product description pages], and then we have a PDP for our site, right? We can do whatever we want on our own product detail pages, because we control them. But Walmart has a specific spec, and Target has a specific spec, for example.
So, something that used to take our retailer dot-com team weeks — which is to take information and make sure it’s ready for Walmart.com and Ulta.com — [is now partially automated]. Now, every time a page needs a refresh, the [AI] agent sends the refresh on a schedule, and our retailer dot-com team is just refining it. It’s saving them weeks’ worth of work. So that’s like an example of something that used to be done manually. We’re now having the agent assist and get it to a certain point, and then the human refines.”
How are you working with your teams to ensure they’re using AI in the best and most efficient ways?
“I think it’s two-fold. One is the enablement for the organization. We have a platform called E.l.f. U, which is our learning platform. We partnered with Section AI [a workforce transformation company] on [things like], ‘Hey, I’m an accounting person. How can I use this?’ — the basics of prompting, and things like that. That’s company-wide.
We have also launched an AI policy — tools that you can or cannot use. Then, we have these power people; we have a council that comes up with new tools. They’ll be like, ‘Can you use this?’ And then we go through and embed that into our organization.
And that’s all to say that everybody is expected to use AI.”
Do you have any insight into how E.l.f.’s customers feel about how you’re using AI?
“Our community calls us out; they do not want us to use AI for customer-facing. …They’ll say, ‘E.l.f. This is not you; you’re not [being] authentic.’
So, that’s the first thing, and we take that very seriously.
An example [of where we use AI] is in a product we built called E.l.f.-fluencer. And the whole precipice of that product was testing and learning. … Our community is always asking questions, but we would only answer 40% of them, because there’s only so much time in the day and so many people who can do it. So, we’ve been training our AI, and now, 90% of the answers the AI sends to the community manager, and then they review it and post it. So, that’s made their job easy, and now, we’re answering the community faster. That’s how we serve the community, versus launching, like, a sexy campaign that’s built in AI — that’s not us.”
And I’m curious: In this age of AI we’re in, it feels like there have been a lot of advancements very quickly, maybe faster than ever, compared to past technological innovations. Have you found that you’ve had to think differently, or challenge your team to think differently, to adapt to this new age of AI?
“Everybody’s trying to, like, stay young, and they don’t know [how to separate] hype from reality. … When you think of agentic AI, if you don’t have the data [set up] in the right way and the connection set up properly, [you should] forget about [AI] agents, right? The conversation needs to shift from, ‘I want to jump onto this bandwagon, and do this integration,’ to ‘Hey, let’s make sure we’re actually showing up on those LLMs, and we have brand visibility, and we’re scoring higher.’ We should be asking, ‘How are our KPIs changing?’ I don’t see a lot of that discussion. I see more of, ‘Hey, what’s the cool thing that I need to know about?’”