How execs from Ulta Beauty, Tarte and Beekman 1802 are implementing AI into workflows

Speaker A: Hello and welcome to the Glossy Beauty Podcast. A, ah, show about the business of beauty and wellness. I'm Lexi Lebsack.

Speaker B: And I'm Farid Frickfiner.

Speaker A: Today we are sharing a panel that I hosted last week during Glossy's annual E Commerce Summit in Miami Beach. It features three top industry leaders discussing one of the biggest theaters themes of last week's AI, but more specifically how companies are actually implementing AI into their workflows today. But first, Sarah has some news to share with our listeners.

Speaker B: So a little bit of a bittersweet announcement. This is my last time joining the intro with Lexi. Uh, next week will be my last week after over five years at, uh, Glossy launching, uh, the pop vertical. Running the pop vertical and has been a wonderful five years. I'm taking a little bit of time off, but in July I will be starting a new job where I will continue to write about the business of beauty and you will be able to still find my work online. So I'm excited to keep up with our listeners from, uh, a. From a new landing spot soon.

Speaker A: Definitely. Bittersweet, Sarah, you've done so much. You. I mean, growing the pop vertical has been such a huge feat and it's been so nice to watch the last two or three years that I've been here. Um, we've already tried to tap Sarah for where she's going. She's keeping it under wraps for right now, but, um, yeah, we'll definitely be following along.

Speaker B: Yeah, it's for sure bittersweet and more to come. But yeah, tell me, tell me some of what you learned. I mean, AI just is such a hot topic right now. I feel like it's in my inbox every day. What was like, your biggest takeaway? Most insightful AI fact?

Speaker A: Sure. Um, great question. Um, you know, I think that the tactics that a lot of brands shared around implementing AI, how executives are actually using it at their desk day in and day out, how they're training staffers in AI and also how they're holding staffers accountable to AI trainings was very, very interesting. And we're going to get into a little bit of that in the panel. Um, I also noticed a trend amongst some larger companies that I feel like I have to call out, which is that some are really empowering their executives and staffers through dynamic trainings and access to the latest AI tools. And then there are others on the opposite end of the spectrum that are really, we're hearing, hindering their employees and workflows by limiting access to AI tools and trainings. And it kind of feels like maybe some of those companies or some of those employees might get lost a little bit as we continue to experience more innovation in AI. Um, we do have. I do have one example though. Outside of the panel we're going to hear from today, we heard from Julie Bailey Blanche. She's the VP of Global Marketing at, uh, Shark. And they have one of the most dynamic strategies. They basically shut down their offices for an entire week to host an AI crash course intensive where they effectively stopped all of their work. And they brought in AI educators, they brought in people to teach the employees about AI. They hosted, like, big, you know, know, hackathon type things in their offices. Um, yeah, Julie basically said that they did a year's worth of training in one week. And she knows, you know, ah, she has all this confidence now to code, to be doing all of this different stuff on AI platforms. And I think that it's really a glimpse into one of the more dynamic things that are happening right now. Companies that are like, we're going to join the revolution. It's not how. It's not going anywhere. Um, and that's how they're doing it.

Speaker B: It's so interesting that it's like one extreme or the other, and it's like either all in or all out. And I mean, I've had some conversations where I've been shocked at some of the things people were using AI for. Just in terms of like, oh, I thought you would have known that. But in a way where it was kind of reassuring to me, I was like, wow, everybody is kind of just flying by the seat of their pants in this life. Um, and like, people that you think have it all figured out don't always. And there is something like, kind of humanizing about that. I mean, it's separate of AI, honestly. But yeah, specific to this panel, um, who are we hearing from today?

Speaker A: Yeah, I completely agree with you, too. I think that this summit was me and Danny and Jill kind of looked at each other after the first or second day and we're like, we might as well call the AI Summit this year because that's all anyone wants to talk about. And it does feel like we're coming from a level playing field where there are certain people who are going to just shoot to the top because they're given access to tools, because they're given access to trainings. Um, and a lot of people, just like you said, Sarah, you think they kind of have it figured out and they're actually like, in last place right now. In terms of, terms of how they're, how they're implementing or how they're um, adopting it. But yeah, so this panel is very dynamic. All three of our, um, experts have really, really interesting things to say about managing and leveraging AI. Our first gu, um, he comes from Ulta Beauty. He's an E commerce expert. His CV includes Dell Technologies and Neiman Marcus, the SVP of Digital and Ecom at Ulta Beauty, Josh Friedman. He was great. We also have a beauty executive on the panel with over 15 years of experience. She's worked for brands like Dry Bar and First Aid Beauty. She's been with Tarte Cosmetics for nearly seven years. VP of um, Digital Marketing and TikTok Shop. Jenna Manula Linares. She has some really dynamic things. And then David Baker is our last one. He was kind of little star of the, of the summit. He had a lot of really good insights that he shared, um, widely and people were always calling on him and asking him questions. Um, he's a former l' Oreal Group and Google executive. He has spent the last four years at skincare brand Beakman 1802, first as their Chief Digital officer and now as their Chief Revenue Officer. So David Baker lended a lot of insight as well.

Speaker B: Amazing. Sounds like a fascinating panel. And I mean it's, it's really interesting too what you said about like other companies not being allowed to use it. Like I'm sure we'll see that, or I wonder if we'll see that change. But yeah, excited to hear.

Speaker C: Yeah.

Speaker A: All right, well, here's my conversation with John, Jenna and David. All right, guys, let's take a seat. Thank all three of you so much for joining us. What a powerhouse. Yes, my pleasure. I'm so happy you're here. Um, all right, so I want to get right into it. Let's talk about internally how you're using AI today. And I really want to talk about your day to day operations. I want to talk about productivity outcomes, all that good stuff. Who wants to start?

Speaker D: Yeah, I'm more than happy to. Um, uh, I was at uh, a lunch earlier and I was like, you can't put me in a room without me opening my mouth. So I fill the silence. But I think we're using it in a myriad of different ways. I think it's, how are we not using it? But in particular a concrete example. Uh, as we heard, the rise of influence marketing and the focus there has always been incredible and continues to be. But finding and sourcing those creators gets really hard. So we've built, um, an agentic, uh, staffer. Her, uh, name's Zoe. Uh, and Zoe, uh, is designed to go and source that, uh, draft personalized outreach to creators so that we can do and find people who fit our ethos and fit our brand voice really, really well at scale while we sleep. Um, and it's changing the way we're thinking about, like your traditional influencer marketing agency approach or, uh, even how we long term, uh, retool and empower our team to do more in person, authentic community events, uh, instead of spending time just trolling through and being like, is this the right person?

Speaker A: Sure.

Speaker D: So we've built that using Claude to help power a lot of that.

Speaker A: Wow, that's so interesting. Do you think that people know that it's AI when they're first getting approached?

Speaker D: Well, the AI drafts the message, but it's still human in the loop.

Speaker C: Love it.

Speaker D: To review it, we still really believe that in that authentic human connection. But also an influencer on TikTok shop, who's selling anything gets probably, what, 50 messages from a Yuca app, uh, user, ah, every day, probably 49 from me. Uh-huh. Exactly. So I think they're used to the inbound, that initial approach being at scale. And I mean, do you know that every email that you get isn't a cold outreach run by AI? Sure, some of them are, but the good ones aren't.

Speaker A: Yeah. Jenna, I know you had insights here too. How's your team using AI right now? Internally?

Speaker C: Yes. Well, I want to meet Zoe because I feel like she can make my life a lot easier. I would say AI has permeated every team and workflow that we have at Tarte. I'm constantly telling my team, if it takes you longer than 15 minutes to do something, there's a faster way and you should learn and try to figure out via AI. For us, um, one big example that's been really helpful is just really concepting. So using AI to come up with new creative activations for influencers, for events, and then also with our creative team. So when my team is briefing into creative and we produce all of our creative in house, I tell them, you cannot submit a brief unless you actually submit a designated output that you'd like to receive. So it really helps us go from concept to first draft, not necessarily creating the final output, because within beauty, um, authenticity and trust is really important. Although an AI unit can spit out or an agent, like 100 pieces of creative, it still takes a human element and approach to make it feel authentically tart. But it really helps us get from that blank page that first draft pretty instantly. And that's cut down on weeks, if not months of work, depending on what the actual activation is. If you're looking to redesign a website or new landing page, that saves us months of work. Wow.

Speaker A: And do you have any. I know we talked about Claude. Do you have any, um, favorite programs, any sort of name dropping?

Speaker C: Oh my gosh. I find that there's AI built into all the tools that we're already leveraging in our third parties. Claude has been my best friend. I think the thing that everyone should do is pay for the paid version of Claude if your company's not doing it and create a project. And when you create that project, label it like mine is like my tarte marketing assistant. I uploaded my brand guidelines, our tone of voice, my favorite presentations. Um, I drop in links to our social media handles to our website and essentially it becomes a guitar marketing agent or analyst for me. That's in my back pocket. Then anytime I need to start a new project, not starting from scratch, it's learning every time and getting smarter and smarter along with me. Wow. So smart.

Speaker A: Um, I want to talk about externally facing AI too, so I feel like we have to start with you, Josh, because in April Ulta Beauty announced a partnership with Google's Gemini. You guys now have an AI powered chatbot on the site. You also are going to be showing up with Gemini results, which is huge. Can you share learnings about what consumers want from AI and sort of, I don't know how you're giving it to them.

Speaker E: Yeah, it's gone well so far. I think a lot of the stats that I've seen up here so far ring true for us as well. Engagement is high. Time on site is higher. Certainly conversion is higher whether it's off site or on site. So we're seeing good results. Uh, obviously much more conversational and much more natural language based than keywords, uh, are. So all the typical patterns exist. Uh, we're really seeing some interesting things happen on site that are a little different than off site. But we're really looking forward to what Google is doing, especially off site to go build, uh, multi sku, multi brand universal cart, uh, uh, other protocols like uh, hooking up to our loyalty program with 47 million loyalty members. So we want to make sure that they see and receive the value of the loyalty program in all those services.

Speaker C: Sure.

Speaker A: When your team was building this, did you come across any like, I guess maybe surprises or anything that sort of like piqued your interest about how either it went or how consumers are using it?

Speaker E: Uh, we continue to find new data sets that we need to put in, uh, store locations, store hours, a lot of those things where uh, customers uh, are just asking generic questions where they'll think about it in product or solution mindset, but they're really asking lots of questions about the brand.

Speaker A: Sure.

Speaker E: Uh, and we're seeing some really good use cases too now in our customer care agent as well.

Speaker A: Interesting. Yeah. I think that in our world we all think everyone knows all about AI but then I think some consumers at home are like what's your return policy like?

Speaker E: Just kind of all that stuff.

Speaker A: Yeah. Um, what about you guys? How are you using AI in a consumer facing way?

Speaker C: I would say for us when you come to our website, if you receive an email and sms those dynamic product recommendations, that's like the number one way if you're going to checkout and you get a bag upsell. That's all driven by AI. So most of our product recommendations are there. We're also leveraging it a ton for media buying. We do all of our media buying in House and TikTok, Meta, Google, they've done a really nice job of building a lot of AI already into their offerings. So leaning into their automated and AI powered targeting, bidding algorithms, creative um, delivery. That's all been really helpful for us as well. Cool.

Speaker D: I um, think for us we're using it a lot. When I think about consumer facing, definitely everything you're saying, uh, what I think as well, we're using it and testing it in how we can wireframe or plan our shoots or plan our thinking a little bit better because then we can look at, okay, how do we wireframe and put together a proof of concept of a new landing page to talk about our incredible launch. And we'll look at three, four, five different examples that are spun up immediately instead of waiting for a few weeks in design time to get it out and we pick that direction and go um, what we are very keen not to do and I think um, because you had also asked how people are reacting to it, I think we're very keen not to or be very careful with generative AI assets and images. We're a science backed skincare brand. I never want anyone to be thinking is that before and after a fake.

Speaker A: Sure.

Speaker D: Because it never is and it never will be. Uh, but I think uh, what's really interesting is we're in this time where as you said we're all so in this AI space but the consumer has a little bit of a backlash at the moment. You look at any comment, they're like, why are you destroying the planet? Or I'm on Team Pope. Um, uh, and really there is this resistance and the only way I can think about it is, uh, similar to film, a lot of amazing effects that have been done in film were originally practical. You built the dinosaur in Jurassic park and someone was inside the dinosaur. Uh, and we've moved into computer generated images, cgi, um, and for a while in that early age it was really ugly and everybody hated it. Uh, and then we've shifted into some really amazing CGI and practical that have mixed, that tell amazing stories. So what I consistently see is the consumer doesn't care if you're using AI to run the back end. They care if the story is bad.

Speaker A: Mhm.

Speaker D: And so it really, to what we were saying earlier, uh, there's that authenticity that's really needed and wanted and that comes through in the storytelling, not in the asset.

Speaker A: Well, that leads me to a question that I was going to do later. But I mean, you must have a policy, for example at Beekman or maybe you guys do too, about using sort of a generated human or I don't know what's the sort of line in the sand that we're at now around using because you obviously don't want backlash.

Speaker C: For us, it's no generated models or no generated product. So we will shoot all of our models in house. We'll use our photo team to shoot products. We'll use AI to manipulate the world in the back to add cool topography to the front of the graphic. And for us right now it's really to come up with that first concept and then initial draft and then also too like we tap into Perplexity a lot to look at competitor insights. Like pull what were like the top pieces, um, of creative on Instagram last month as it relates to this concealer or the specific product and then it helps our designers. Mhm.

Speaker A: That makes sense.

Speaker D: I fully agree. Like, no, it's actually pretty much the exact same rules we all had for Photoshop. Uh, we're not going to use fake product or fake images of the people, but absolutely use it for layers. Use it for a different expansive, uh, background. Um, so a lot of that is still there where I think there is similar, uh, to Photoshop or any sort of 3D animation. That's where. Does the consumer know if we animated that in uh, After Effects or in Gemini? No. So if it's this very whimsical cartoon area that generally tends to be fair games. We have our mascot Goatee who oftentimes is either computer generated or AI generated, but it's obviously clearly not real because it is a fantastic animation.

Speaker A: Yeah, for sure.

Speaker C: Well, uh, we have to talk about

Speaker A: personalization because we are here and we're talking about AI, but I also feel like it's kind of losing its meaning a little bit, the word. Um, Josh, maybe I'll start with you. How are you guys using AI to sort of supercharge personalization right now and what's working best?

Speaker E: Sure. Uh, it's definitely helping in the algorithms. So we know about beauty, we know about you. We want to make sure we're bringing right content to the right person at the right time and it's helping us whether it's identify next best behavior, next best action, next best content. Um, we're using it to make all those types of uh, not full decisions, but certainly putting together the weightings for those.

Speaker D: Mhm.

Speaker A: Okay, cool. How are you guys using AI to supercharge personalization all the way to.

Speaker C: Josh mentioned, um, product recommendations, send time optimization. When it comes to like email and SMS marketing, like making sure that we leverage AI to determine the time someone is most likely to engage with that marketing correspondence has been really helpful for us. We one way that we're looking at it too for this year and we tested it last year which is really interesting. The holidays are right around the corner and last year instead of rolling out like a formal holiday gift guide, we did an AI powered guide and that was really successful just because the way people are searching, it's more with natural tones and language than typing out best gift for my 16 year old daughter. So that worked really well for us last year and that's currently what we're working on. Wow.

Speaker A: So did it look different for consumers or like how did it, what did it look like?

Speaker C: I would say you had your same kind of category header page or your homepage banner, but then the products would surface based off of your results. So instead of just limiting the gift guide to be 10 products on site or just our holiday collection, you open the door to your entire product portfolio.

Speaker D: Oh cool.

Speaker E: That's really cool.

Speaker D: Um, same way that everyone else is using and then the other way and I think they're actually a sponsor so I'll give them a little heads up. But postscript from an SMS perspective, we've leaned in with their AI personalization tool. So um, a lot of the responses we're getting back from our consumers when we Send out a campaign, then our AI automated personalized responding back. And so they're chatting with, uh, a large language model. Mhm. Um, that's powered and trained on our brand voice. And what's kind of crazy is we're seeing a 25x increase in ROI from the automated versus not because she's having a conversation with Beekman, whereas before we didn't have the ability to do that. So it's at scale. It's been incredible for us.

Speaker A: Wow. Jenna, when we all first chatted on a Zoom, you said that it's a fine line between personalization being really helpful and getting a little creepy. I want you to explain where that line is right now.

Speaker C: Well, I would say just with AI and all the data points you're able to collect, you definitely want to make sure that you're helping the customer get to their point of sale faster and more efficiently. But you don't want to pull out too many, I would say, personal identifiers that can make them feel uncomfortable or like Big Brother esque. So just because you have access to information doesn't mean you have to use it or you have to make it like front and center or you have permission to correct.

Speaker A: Oh yeah, that's so true. If it knows something about you and you're like, wait a minute, they know that or.

Speaker D: And often I find it's that information that they think they know about you that's not relevant. And I see this in the BDR being pitched, um, with these tools, um, like the number of times they're like, oh, I see you went to this university that I haven't been back in that state for 20 years. And they're like, I also love the mighty Ute students section. Do you get to go see games often? And I'm like, you literally know nothing about me. This is just a bad script.

Speaker A: Sure.

Speaker D: And so I think it goes back to like, if it's good, amazing. The bar for it being good is high. And so it's making certain that you are telling a good story, being able to be that connective tissue. There have been many AI generated BDR pitches that I've been got by because they actually felt truly personalized. Mhm. And that to me is the inspiration of like, how do I do that for. So Zoe reaches out in a more personalized way or so that we talk to all of our consumers in a more personalized way.

Speaker C: Yeah.

Speaker A: I'm curious, real world takeaways here. How are you training your teams to be better at AI? There must be People who are resistant, there must be people who aren't getting it as quickly. Um, what's the process look like? How are you training people and whoever wants to jump in?

Speaker D: Um, I think first and foremost it's finding the people who have an interest in it and giving them the room and space to play and sitting down with them and saying, look, what's your hardest problem? What's your thing that takes you more than 15 minutes? Let's together build something so you see what's possible. And if that is enough to tip you over, they're going to keep going with it. Um, but then we have the detractors who are very resistant to it for many different ways and some of it's just repetition. But I think where I'm starting to articulate this more, we have not articulated as loudly as I'm about to right now. But I feel like, uh, everyone's afraid that AI is coming for their jobs. It's literally headlines everywhere. I firmly do not believe that AI is coming for anyone at Beekman's job. But someone who knows how to use AI is going to come for someone who doesn't know how to use AI's job. And if you are still there, if you hired a finance person and they were doing spreadsheets with pen and paper like they used to do in the 50s and 60s, you'd fire that person. And I think we are getting to the level where we have incredible tools at our disposal and someone choosing not to use them, that becomes a business risk. Mhm.

Speaker C: Well said. I completely agree. I would say for us, we don't have any formal training in place at tart, but we're creating a culture of experimentation. So what I challenge my teams to do each week is to use AI in a new or different way. It doesn't need to be new AI tool, it can just be a different prompt that you're using with cloud or ChatGPT or perplexity, whatever it may be. And then in our weekly team meetings, we set aside 15 minutes at the very end and we go around in a circle and we call on each person and they have to say and explain to the group how they leveraged AI and if it's helpful, if it was beneficial or not. And then we keep track and just honestly, a spreadsheet, like a Google spreadsheet, like all the prompts that worked, all the ones that that didn't, and um, we just keep doing that and we're learning from one another.

Speaker D: We're going to need to talk later. Because we were doing that. And then we gave it one more where we said, okay, Claude, help us build a training program for the team. Give them homework assignments and they have to turn it into you and you grade it. So then we're not keeping track on the spreadsheet anymore. We're able to look at where that's coming through.

Speaker C: Let's chat after this.

Speaker D: Wow.

Speaker C: Just like that.

Speaker A: Next up, do you want to take that one too?

Speaker D: Sure.

Speaker E: The best thing is just sharing, right? Uh, we do have a little bit of training on, um, how to use a couple of different tools, but it's really peer to peer. Uh, sharing is really the best way. Um, and hearing how your peer is doing it at the desk next to you or someone across the hall who might be doing a different job is just the easiest way. And everybody's using it at home now too. So, um, I don't think there's too many people still using typewriters or whatever the analogy would be. So, um, nobody's coming in too resistant. And uh, now it's just a matter

Speaker D: of how to apply it.

Speaker A: Yeah, um, I've got one more, um, for you guys. Uh, maybe let's start with you, Josh. Talk to me a little bit about what the future holds. Let's look ahead. Let's put on imaginary, uh, caps here. One year, five years, six years, seven years, um, what will AI continue to change?

Speaker E: Uh, I think work is probably one of the bigger ones. Certainly how we work and how we manage our agents at work will probably be one of the biggest ones as far as guest facing. Um, I do feel like folks, uh, will use it a lot more to assist them in finding a solution or maybe more articulating the problem. But uh, but I think they'll still want to know, uh, what they're buying and why and how much it costs

Speaker C: in five years. I don't know if we'll be talking about AI in the context that we are. I think it's just going to be built into the infrastructure of everything out there. We're not going to be talking about one off tools.

Speaker D: Mhm.

Speaker C: But from the customer standpoint, I do think everyone's just going to be a lot less patient if you could even imagine that. It's just going to be table stakes. They're going to want an experience that is relevant, personalized and fast. And I just think that's going to be the nature of it.

Speaker D: Um, I'll go with my controversial one. I think that we are on the verge of new companies starting right now. Who their entire backend architecture is agentic. Forward is as revolutionary transformational as digitally native vertical brands were 15, 20 years ago in transforming the retailer relationship, the D2C relationship, the entire UM way of working. Because if everything is centralized and all context and memory of the brand is in one place instead of dispersed, that that actually allows for new brands to accelerate at a very different pace. And I think you'll see M and A activity buying into those brands. I think you're going to see transformation across the board and I think I'm going to counterpoint on yours. I think we will be talking about AI in the same way we are still talking about D2C and social. It's just not the new thing. It's how is everyone doing it a little bit differently and better.

Speaker A: Thank you guys so much. Give it up for Josh and David and Jenna. Thank you guys so much. And thank you for listening to this episode of the Glossy Beauty Podcast. If you enjoyed it, please leave us a rating and a review on Apple Podcasts, Spotify or wherever you're listening. And for even more coverage on the beauty industry and more, please visit Glossy Co Beauty.