Zondre Watson, general manager of technology and analytics for adult retail chain Ero-Tech, has developed a new retail analytics system called Retalyz. This system aims to improve inventory management and ordering processes across Ero-Tech's 21 locations by leveraging AI and providing context-specific data.

Addressing Inventory Challenges with AI

Watson, who has a background in finance, chocolate, and controlled chaos, stated that the purpose of software is to learn to understand a business and continuously improve. He noted that AI is no longer an advantage but a necessity for retailers. However, these tools require a tailored environment and specific context to function effectively.

Watson developed Retalyz over the past year to automate and enhance automatable work, particularly in ordering. He described inventory as the "lifeblood of a retailer," with ordering being a critical point. Previously, orders across Ero-Tech's 21 stores were placed manually, often involving a guessing game that consumed hours. The old process involved managers running POS reorder reports, cleaning up discontinued items, and adjusting for sales and clearance spikes before guessing quantities. This was then reviewed by a buyer, and if anything fell outside the usual range, it was escalated to Watson or an owner. Each review took one to three hours and was repeated three or four times for the same order.

A specific incident highlighted the need for a more robust system: an order for six "bullets" resulted in 12 "fishbowls," increasing the order cost by approximately $1,300 due to a unit-of-measure mix-up. This occurred because the vendor sold by the case, the POS system thought in singles, and the tools in place could not prevent the error as they lacked information on how the vendor actually shipped the item. Watson explained that the gap between what software knows and what an operator knows was the central focus of his work.

Retalyz: A Context-Driven Solution

Watson identified a flaw in most reorder math systems, including Ero-Tech's previous one, which calculated units sold divided by days between orders. This method was "blind" to items that sold out quickly and remained out of stock, as it only counted actual sales and not lost sales due to bare shelves. To address this, Watson built a portal within Retalyz to estimate these "invisible sales."

Every item in Retalyz receives a confidence score based on its availability. If an item was consistently in stock, its sales are trusted as real demand and adjusted for seasonality and trend. If an item was out of stock for a period, the confidence score drops, and the projection scales up to estimate potential sales had it been available. Orders are then sized to prevent stockouts before the next delivery.

The system also learns real lead times for each path, including the time it takes for a manager to place an order, a buyer to process it, and a vendor to ship. It adjusts these times as they drift. The portal automatically completes the review process, reading daily changes in sales, receiving, transfers, and returns. It builds orders by dropping discontinued items, adjusting quantities, checking vendor thresholds and minimums, converting single units into vendor-shipped boxes and cases, and flagging when an additional unit crosses a free-shipping threshold. This functionality directly addresses issues like the "fishbowl problem."

The completed purchase orders are then routed for approval and scheduling, allowing purchasing to manage cash flow. This shifts human involvement from performing the work to reviewing exceptions. Watson stated that the one-to-three-hour review process, previously done multiple times, now collapses into a few minutes of reviewing system-flagged items. This results in fewer errors, less idle stock, and fewer empty pegs.

Watson emphasized that while generic, expensive systems exist, they often lack specific context such as lead times, clearance habits, or vendor-specific shipping practices. He concluded that the effectiveness of Retalyz comes not from complex code, but from the operational context provided by someone who understands the business. This approach represents a shift from automation to an "agentic approach," where people monitor exceptions rather than follow strict scripts.

Key Facts

  • Zondre Watson, General Manager of Technology and Analytics for Ero-Tech, developed Retalyz.
  • Retalyz is a retail analytics system designed to improve inventory management and ordering.
  • Ero-Tech operates 21 retail locations.
  • The system addresses issues like unit-of-measure mix-ups and lost sales due to out-of-stock items.
  • Retalyz automates order generation, adjusts for seasonality, and learns vendor lead times.
  • The system aims to reduce manual review time from hours to minutes.