/ Case Studies / U.S. Metal Distributor Reducing Sales Complexity with Automated Product Search Case Studies U.S. Metal Distributor Reducing Sales Complexity with Automated Product SearchUsing Natural Language Processing to Transform Product OrderingThe primary goal of this case study was to significantly reduce the time it took to provide quotes for over 100,000 different types of metal products, enhancing the efficiency and competitiveness of a leading U.S. metal distributor. The solution involved the deployment of a cloud-based system powered by Microsoft technology and natural language processing (NLP). This innovative system automated product searches and rapidly generated near-real-time quotes for sales representatives to present to customers.Objective: Reduce the time it takes to quote customer orders of 100,000 different types of metal products. Solution: Deploy a Microsoft cloud-based and natural language processing solution that automates product searches and presents near-real-time quotes for sales representatives to present to customers. Results: Decrease time to quote from five minutes to less than 30 seconds. Enable sales representatives to quote and fill more orders. Gain a solution that can scale with new products and customer demand. The solution is predicted to drive X% more revenues for the metal distributor this year. Metal distributors and processors operate a complex business. They must source, buy, and distribute stainless, aluminum, carbon, and alloy products in all shapes and sizes. A leading U.S. metal distributor has thrived and grown by serving as an expert on metal products, facilitating the purchase or more than 100,000 different items. The company’s sales representatives receive customer requests via email on average every 4.5 minutes. They then seek to fulfill requests at speed, to capture new business and keep existing accounts. Customers typically send their requests to multiple distributors, meaning that the first to respond and fill the orders wins the business. Most orders take around five minutes to fill, meaning that there is little margin for error. One challenge with the company’s business model is that it doesn’t scale easily and is vulnerable to issues such as staff departures and shortages. Even employee lunches, breaks, or vacations could harm sales if demand spikes and the company doesn’t have enough staff on hand to cover customer requests. Another issue is that it takes significant time for new hires to learn the company’s business and product list. Customer requests are presented in a variety of ways, such as product SKUs or attributes. Staff have to understand this terminology and be able to quickly search the inventory of items, so that they can find the correct products and get customer approval to order products. Using Natural Language Processing to Transform Product Ordering The company’s leaders wanted to create competitive advantage by accelerating its time to quote. Being able to automatically find and quote costs for metal products would enable the company to win more business. The metal distributor engaged Concurrency to build an automated quoting system, code-named Metal Detector. The solution uses Microsoft cloud technology, including Microsoft Azure, Azure DevOps, and Azure Databricks; the Spark NLP library; and Azure Kubernetes Services; and integrates with Microsoft 365. Metal Detector scans email in real-time and uses entity recognition to look for relevant information, such as SKUs and product information. As an example, the model is trained to recognize “steel,” “stainless steel,” and “SS” as the same description. The model can also interpret missing information, such as understanding that W1 means beam width and that if no product is specified, it means carbon. Next, Metal Detector parses this information into a usable format, and queries it against the metal distributor’s database. Sales representatives then receive a near-real-time quote that they can present to customers. Customers can click on the order button to approve the quote and initiate the purchase. Metal Detector also integrates with SAP and Oracle ERP systems, keeping business and financial data up-to-date. The company’s leaders can use Power BI Reporting to assess sales in real-time. They can determine Metal Detector’s ROI by measuring increased sales against baseline data. Delivering a Better Customer and Colleague Experience In the solution’s next evolution, customers will be able to receive quotes directly, enabling them to use self-service to navigate the purchasing process. As a result, company sales staff will be able to focus more on delivering consultative service or solving problems, rather than just quoting products. It also will be easier for the company to hire and retain staff, since they won’t need to master a 30,000-line-item database. Automating quoting and ordering with Metal Detector will enable the metal distributor to deliver a better customer experience and capture more business from competitors. Generative artificial intelligence (AI) advances mean that the company can easily add more products, without the need to train machine learning models on new data. Generative AI uses transfer learning to learn how to process new data sets, enabling organizations to deploy new capabilities that previously took months or years in just a few weeks. The company’s leaders also envision using Metal Detector for other applications, such as improving credit reporting on aging invoices and predicting customer demand changes. By creating new efficiencies, fulfilling more demand, and improving invoicing returns, teams can deliver more profitability to the business.