Unearthing Customer Insights with AI

A Persistent Challenge in Industry Classification

Imagine manually categorizing millions of customers into industry-specific classifications using North American Industry Classification System (NAICS) codes. Each classification required 20–30 minutes of meticulous human effort, involving sifting through company websites, databases, and public records. For Grainger, a leader in industrial supply and distribution, this was the reality of customer intelligence workflows. As new customers joined or existing ones reached specific thresholds, the team faced mounting pressure to scale their industry classification process. This is Jonny LeRoy's success story.

LeRoy, Grainger’s Chief Technology Officer, explained the scale of the problem: “Working out what industry a customer is in sounds simple, but it’s a painstaking task. The manual workflows were reliable but slow, and as we scaled, the inefficiency became glaring.”

This task was more than just a logistical inconvenience—it was a business bottleneck. Misclassification or delays in understanding customer industries could result in missed opportunities for targeted marketing, sales engagement, and strategic planning. Grainger needed a solution that could scale efficiently without compromising accuracy.

AI to the Rescue: A Targeted, Incremental Solution

Determined to streamline this process, Grainger turned to large language models (LLMs), an advanced AI technology adept at processing and interpreting text data. Rather than embarking on an extensive overhaul, the team applied AI to a specific bottleneck within the customer classification workflow. The goal: automate the initial research and recommendation step while maintaining human oversight.

LeRoy described the approach with a simple analogy: “We treated the AI like a team of eager interns—it did the legwork of researching and suggesting NAICS codes, but under tight guardrails to ensure human oversight. It wasn’t about replacing people but enabling them.”

The AI solution worked by analyzing customer data, generating industry recommendations, and supporting each suggestion with linked sources for human validation. This collaborative framework combined AI’s speed with human expertise to ensure accuracy and accountability.

Navigating Implementation Challenges

The implementation of this AI system wasn’t without its challenges. Like many organizations adopting AI, Grainger needed to address technical and organizational hurdles:

1. Data privacy and security: Ensuring customer data was secure was a top priority. The team implemented robust governance frameworks to protect sensitive information and adhere to strict compliance standards.
2. Accuracy and reliability: AI-generated classifications needed to be defensible. To address this, Grainger established validation protocols where human reviewers would cross-check the AI’s recommendations against provided sources.
3. Integration with existing systems: Incorporating AI into existing workflows requires careful planning. The team focused on a phased rollout, starting with a limited set of classifications before scaling the system.

LeRoy stressed the importance of collaboration in overcoming these hurdles: “We involved stakeholders across customer intelligence, IT governance, and AI teams to ensure the solution was practical and aligned with business needs.”

Transformative Impact of AI

The results of the AI implementation were transformative, both operationally and strategically. While Grainger has not disclosed precise metrics, the qualitative benefits of the solution have been significant.

• Dramatic efficiency gains: The AI reduced classification time from 20–30 minutes to just 2–3 minutes per customer. This represents a roughly 90% improvement in efficiency, enabling the team to process far more classifications without additional headcount. This is particularly impactful when considering Grainger’s customer base spans millions globally.
• Scalability: By automating a critical step in the process, Grainger scaled its classification efforts to keep pace with business growth. The AI system could handle surges in customer volume without creating bottlenecks or delays.
• Enhanced accuracy and consistency: The combination of AI-generated recommendations and human validation led to improved accuracy in classifications. Human reviewers reported fewer errors and higher confidence in the outputs.
• Employee enablement: Teams previously bogged down by repetitive manual research were freed to focus on strategic activities like customer insights, segmentation, and targeted marketing. This shift also boosted morale, as employees found more satisfaction in higher-value work.

LeRoy reflected on these results: “The real power of AI lies in taking something tedious and time-consuming and turning it into a scalable, reliable process. It’s not magic—it’s about applying the technology to areas where it can have the most immediate impact.”

Lessons Learned on the AI Journey

Grainger’s experience with AI offers valuable insights for other enterprises considering similar transformations. LeRoy highlighted several key lessons:

1. Start small, scale strategically: “Don’t aim to solve everything at once. Pick a specific, painful step in a process, fix it, and build from there.” This incremental approach allowed Grainger to demonstrate value quickly and iterate on the system.
2. Embed guardrails: Human oversight isn’t just about ensuring accuracy—it’s about building trust in the system. Grainger’s emphasis on validation ensured that AI outputs were defensible and aligned with business goals.
3. Iterate and improve: AI systems improve over time with use and feedback. By creating a feedback loop where human reviewers refine AI outputs, Grainger ensured continuous improvement in accuracy and performance.
4. Prioritize governance: Addressing data security and compliance upfront minimized risks and facilitated adoption across the organization.

Shaping the Future

While the immediate gains of the AI system are clear, Grainger’s journey is far from over. LeRoy envisions expanding the use of AI across other processes in the organization. “This is just the beginning. There are countless opportunities to apply AI to pain points across our supply chain, customer intelligence, and beyond.”

By focusing on well-defined, high-impact problems, Grainger has built a scalable framework for AI adoption. The company’s approach of combining human expertise with AI’s capabilities has not only transformed its industry classification process but also laid the foundation for broader digital transformation.

Redefining Possibilities

Grainger’s success story underscores the transformative potential of AI when applied thoughtfully and incrementally. The company’s ability to streamline a manual, resource-intensive process through targeted AI adoption is a testament to the power of collaboration, governance, and iteration.

For organizations contemplating their AI journey, LeRoy’s advice is clear: “Dive in now. Waiting for the perfect technology will leave you behind. Start with what you have, focus on areas with immediate impact, and let the results speak for themselves.”

With AI, Grainger has turned a once-daunting task into a seamless, scalable process. As businesses worldwide grapple with efficiency and scalability challenges, Grainger’s approach serves as a blueprint for unlocking AI’s potential.