Lowe's: Mylow Users Convert at 3x the Rate

Mylow and Mylow Companion, the AI assistants that US home improvement retailer Lowe's built on OpenAI models, have answered millions of questions from customers and store associates since launching in spring 2025, OpenAI said in a customer story published on September 25. According to Lowe's Q2 2026 earnings call, customers who used Mylow on Lowes.com converted at three times the rate of online shoppers who did not. Mylow Companion, the version for store staff, is used by roughly 300,000 associates across 1,700 stores in all 50 states.

The two services play different roles. Mylow is a conversational advisor on Lowes.com that answers open-ended questions with step-by-step guidance, product recommendations and relevant how-to content. Mylow Companion gives associates conversational answers across departments on mobile and desktop. OpenAI's customer page lists the products Lowe's uses as ChatGPT and the API. According to reports, Lowe's introduced the customer-facing Mylow in March 2025 and rolled out Mylow Companion, built on the same foundation, to all 1,700-plus stores that May.

The starting point is the project, not the product. "When you're buying a t-shirt and it doesn't fit, you just return it. No big deal," said Seemantini Godbole, Executive Vice President and Chief Information and AI Officer at Lowe's. "But if you're renovating a kitchen or redoing your floors, those are expensive decisions. You want to feel confident. And that requires expertise." Lowe's stores run up to 130,000 square feet with tens of thousands of SKUs, so even seasoned associates can get stuck on questions outside their department. Online shoppers face a similar problem, the company said, because traditional search only helps if they already know what to look for.

Customer expectations shifted in 2023. "ChatGPT changed expectations," said Joe Cano, SVP of Digital Commerce. "Customers now expect fast, expert answers 24/7. They want to describe their project in natural language and get help, not open ten tabs to figure out what mulch works in their zip code." Cano said most visitors arrive not with a product in mind but with a problem, such as hanging shelves or finding paint that holds up to three kids, and that Mylow's strength is understanding that intent.

In stores, the change runs through voice. According to OpenAI, nearly half of Mylow Companion interactions happen by voice. Without leaving the customer, an associate can ask aloud something like "A customer is painting their living room and wants a finish that's easy to clean. What should I recommend?" and use guidance grounded in Lowe's project knowledge to talk through options. "Before, if a customer needed to replace a hose bib and the plumbing expert wasn't there, you'd have three people spending 30 minutes trying to help. Now, one associate can get it done in minutes," said Que Vance, EVP of Stores.

Performance figures were disclosed as well. When associates use Mylow Companion to help shoppers in the aisle, customer satisfaction has risen 200 basis points. Internal usage patterns show newer associates turning to it frequently, while experienced associates use it for questions beyond their usual departments. "What if each associate asked just one question a day? That's 300,000 questions. That's how we measure success now," Godbole said. Lowe's has added natural voice interaction to Mylow Companion and adopted ChatGPT Enterprise for its Store Support Center, extending its use of OpenAI beyond the store aisle.

METAL has previously reported that US grocery chain Albertsons expanded its retail partnership with OpenAI. Large retailers are deploying generative AI on customer screens and with store staff at the same time. The OpenAI customer story METAL reviewed attributes the threefold conversion figure separately to Lowe's Q2 earnings call, and describes the millions of questions as a cumulative total for both services since launch. The idea of 300,000 associates asking one question a day is a success benchmark set out by Godbole, not an actual usage count.

From a marketing perspective, what this case changes is the role of the search box. When customers needed exact keywords to reach a product, the product taxonomy was effectively the customer journey. Mylow lets customers describe what they are trying to do and pulls product recommendations and project guidance from that description together. Godbole pointed to this structural shift when she said, "We didn't want to just update our systems. We wanted to rewire the entire business to support projects, not just products." The Digital Home Platform Lowe's launched in November 2024, which gives loyalty members personalized information about the products and appliances in their homes, points in the same direction.

The threefold conversion rate and 200-basis-point satisfaction gain serve as evidence that conversational advice can feed into sales metrics. Because the comparison group is online shoppers who did not use Mylow, the published figures alone cannot rule out that customers with stronger purchase intent were more likely to use the tool. How Lowe's carries these numbers forward in its next earnings report will be a benchmark for measuring the real effect of conversational commerce.