Companies need to think clearly about what outcomes they are trying to achieve when adopting generative AI tools, tying the ‘hyped’ technology clearly to their business strategy and values – rather than just adopting artificial intelligence because they have a ‘fear of missing out’ (FOMO) and wanting to get there first. That’s the warning from Gibu Thomas, EVP of Estée Lauder Online (ELC Online), who suggested this week at Google Cloud’s annual user event in San Francisco that it’s better to learn from other people than it is to be the one making early mistakes.
That’s not to say that Estée Lauder Online isn’t making use of AI tools, including generative AI – it is. However, Thomas and his team are focusing on use cases that can prove value to the business and don’t carry regulatory risks, nor do they risk destroying the trust consumers have with the brand.
Some of the use cases being adopted by ELC Online include classifying calls to the company’s call center, copy generation for SEO or marketing, matching products to consumers through search and recommendations, as well as sentiment analysis amongst customers. Thomas said:
When we started the journey, it was about making sure that we had the basics right and had a very strong foundation. The biggest trap I see is that people get caught up in the hype and there’s massive FOMO. You begin to shoot before you’ve aimed. That’s the trap you have to avoid.
Which is why, even though we’re thinking really big, we’re being very deliberate about the use cases that we’re deploying. And how we deploy those use cases is really anchored on true business value or consumer feedback that we’re getting – that it’s actually working.
My advice would be it’s more important to get this right than to be first. And frankly, sometimes other people doing some of the crazier use cases is good, because there’s nothing better than learning with someone else’s tuition.
Estée Lauder’s differentiator, according to Thomas, is that it is a company that is focused on ‘high touch experiences’. The beauty company’s ethos is tailoring the best products for each individual and has been operating for 75 years. Historically, this high touch approach has been focused on customers in store receiving advice over the counter, but as digital tooling has developed, ELC Online’s priority is to blend this high touch approach with high tech. AI is going to be key to this. Thomas said:
I joined the company about two and a half years ago. And one of the early bets that we made was that we said AI is an area that can really differentiate us. The advantage we had is that we didn’t have any legacy encumbering us, so we could really start from the ground up.
ELC Online already had a partnership with Google that predated Thomas’ joining the company, which meant it felt like a natural progression to reach out to Google Cloud and build out some AI tools. Initially this was focused on open source environments, rather than Google Cloud’s own tools, but that has since changed with the introduction of Vertex.
Thomas said that Google Cloud has been a “fantastic partner” and has provided the company with early access to APIs, which has allowed Estée Lauder to validate a lot of the use cases. Key for the organization is blending the human-AI experience. Thomas said:
Generally our approach is to use this as a tool to augment human judgment, because the human piece is really important. And then we have some more ambitious use cases, which we’re dabbling in. But we’re being thoughtful about it because trust and our brand equity is really, really important. We want to be thoughtful to make sure that the things that we’re doing, we feel really confident that it’s bringing consumer and business value before we put it out there.
Trust
Thomas explained that Estée Lauder Online has invested in high caliber talent to support its AI journey, which has enabled it to couple business judgment with technical judgment. He said that it has hired academics to keep the company grounded on the latest research and has employed applied scientists to test all the emerging models. Additionally, technologists are focused on building a bridge between the business and technology teams. He said:
This is not just sitting in a silo, but really is an accelerator for the business. We invest in that talent, so oftentimes when vendors come to us and pitch things, because we have early access to many of these APIs, we’re more informed than the vendors to know if what they’re saying is true or not.
One tool that ELC Online has in production is focused on classifying calls in its call center. Typically call center environments struggle with a high rate of turnover and there is continuous training, which can be costly for a business and yield poor customer experience results. What Estée Lauder Online has found is that the tool is twice as accurate as solely using human judgment to classify calls. Thomas said:
And so that’s a fantastic use case where we’re able to really deliver immediate value to the business. It didn’t really have a lot of the risks and we were able to verify the data quickly. It also gives the company the confidence that this is something that can really be an accelerant to our business, to be more productive, and serve our consumers better.
There are many use cases like that, which are fairly straightforward and foundational, that don’t require massive leaps of faith or organizational stretching out of the comfort zone.
ELC Online is now experimenting with LLMs for copy generation, where Thomas said it has found that the tools enable the company to easily create different types of copy, use different tones, as well as quickly translate copy into different languages. But the key is going slow and going steady, through a process of feedback and validation. He added:
I’d say we’re taking a phased approach to this. We’re thinking really big and we think this has the potential to redefine how we do business. That said, we are being very deliberate about how we pursue that big vision, with proven use cases that then evolve into something more.
For example, we could use it to inform product development in the future. But that’s something we have to be really careful about, because then you’re speaking to brand trust, product efficacy. Our R&D teams are very much involved in that still. It’s sort of like you’re trying to run a marathon, and you’re going to start on mile 16 versus mile zero.
When questioned on how Estée Lauder is thinking about regulations and compliance, as it relates to AI, Thomas said that “if the government tells us to stop, we stop”. But more than simply complying, Estée Lauder is thinking about the responsibility it has, within its industry context, to use AI ethically. Thomas said:
We’ve been around for 75 years, we want to be around for the next 75. That will only happen if we don’t compromise consumer trust and follow all the applicable laws and regulations. But that said, the important thing is proactively anticipating areas where the technology could be misused.
We have principles that are documented where we don’t want to promote harmful beauty standards. We don’t want to have digital misrepresentations of aesthetic models that give people the wrong idea about what authentic beauty means.
So there are principles that we use to make sure that as we apply this technology in the ways in which we engage with our consumers, and use it in our business, that we’re doing it in a very thoughtful and responsible way. Our brand equity and consumer trust are really the biggest assets of the company.
