AI Transformation for Consumer Brands: A Case Study With Madison Reed

As a founder of a consumer brand that’s been growing for over a decade, I intimately ‘get’ both the challenges and opportunities of AI automation in business operations. Madison Reed has used AI since its inception to match a consumer’s color for their hair, however our business is not AI-native — we make tangible products that our customers pick up with their hands and use on their hair… There’s nothing more personal than that.

In 2016, we expanded AI capabilities with an augmented reality app that let customers virtually try on hair color and ultimately help them find their hair color match. It was an obvious way to use AI and AR in a fun, creative way to benefit our customers, and at the time, it was innovative. Needless to say, looking back, it was just the tip of the iceberg in terms of how AI would later impact our bottom line. We’ve found that if a consumer uses this tool, they have a 33% greater conversion rate… Meaningful!

Six months ago, we went deeper and began working with an AI software company called Sierra to do three things: 1) train an AI agent skilled in customer support chat to reduce churn, 2) make product discovery more accessible, and 3) book appointments with ease for our 95 Hair Color Bars. We affectionately named our support agent Madi. 

We went live on our website in February and the revenue and savings resulting from this agent – no exaggeration – will pay for half of what we spend on customer support in a year. The agent is already improving our customer lifetime value, increasing bookings and reducing our cancellation rate by 5x of what it was before. 

Because my team and I are ecstatic about the results, I wanted to share what we’ve learned for other founders running businesses that aren’t AI-native but have massive opportunities to apply automation in ways that actually move the needle beyond productivity hacking.

Identifying Use Cases for AI Experimentation 

We partnered with Sierra to launch three AI agents over the course of four months. We chose to experiment with the three use cases to reduce customer service live interaction because it’s one of the simplest and most straightforward cases for testing traction and learning. As noted above, each of the three AI agents we built has a specific function: 

• Interact with customers managing hair color subscription memberships and account information 

• General product discovery

• Book hair color service appointments in our in-person stores (Hair Color Bars)

To manage the implementation effectively, we formed three dedicated working groups in November 2024, each focused on one of those use cases and staffed with team members with relevant expertise. 

After three months of development, we launched our AI agent in early February 2025 – a realistic timeline for what companies of our size can expect when implementing similar AI solutions. The new AI agent now handles 90% of our web traffic. It started at 30% and, as we learned more, we increased to 50%, then 90% as it stands today – and the impact has been immediate. 

And for us, this isn’t just about rebooting customer support — AI is changing how we operate across the board with creative too, as you’d suspect. With AI, we can take both older and current owned and licensed photography and written content, modify and enhance it, and save significant photoshoot and creative costs to get more use out of a single image, article, or social post. It’s a massive cost savings and lets us work in a smarter, more scalable way. And it reserves human talent for the creative act of ideation, not the tedious amplification of output. We can change hair color, outfits, and backgrounds at record speed – this has real impact.

Selecting AI Technology Partners & Learning Limitations 

In choosing to work with Sierra, we evaluated a number of other AI technologies focused on customer support and felt Sierra’s functionality was best equipped to customize for our specific use cases as well as our omnichannel business strategies. There are so many new AI tools every day to consider. 

My best advice for selecting the right partner is to really ask a lot of questions around capabilities of these companies. Explore their ability to customize the agents for your use cases, and while they may seem inexpensive at launch, understanding the cost of the ongoing platform is critical. We found a lot of inexpensive options that became expensive after launch. In our case, we have in-house development and engineering as well as digital product teams, so it was crucial to get their buy-in. They’ve been critical to driving this project. 

Once your AI agents are launched, you’ll continue to learn and encounter scenarios that unveil what the agents aren’t trained for. That might result in frustrated customers from time to time but it will give you what you need to improve. For example, our customers sometimes want to have consultations with our human stylists through our AI agent, which our agent can’t coordinate yet – so we need a clear plan for directing those inquiries to appropriate channels. 

Every consumer brand should be thinking about how to use AI not just for efficiency, but to solve the real business problems of the leaders of various departments on your teams. In addition, we have begun running company-wide training and webinars that teach folks AI tools and how to be more effective at your jobs with them. This has resulted in better, faster quality of work in all functional areas.

When you align and design AI solutions to meet the needs of departmental leaders on your teams, you will spark greater internal enthusiasm for experimentation, not resistance, and discover new and interesting ways to increase your output, potential cost savings, and potential profit.

Amy Errett is the founding CEO of hair color and care company Madison Reed and an investor at True. She serves on several boards spanning the startup ecosystem and academia – and believes in the power of leading with love. Once you’re backed by True, Amy’s one of the many experts in your corner.