Optimizing Inventory & Catalog Management

The purpose of this engagement was to design and build a centralized product configuration engine that would serve as the single source of truth for our client's sales operations. Our goal was to eliminate the inefficiencies caused by fragmented data by creating a platform that could not only unify complex product information but also understand the relationships between them. Ultimately, we aimed to provide the sales team with a powerful tool to quickly and accurately build optimal leasing solutions, backed by intelligent, data-driven recommendations.

Our client's sales process was hindered by several key problems stemming from their fragmented data:

• Slow & Inefficient Workflows: Salespeople had to manually consult multiple disconnected data sources to build a single quote, drastically slowing down the sales cycle.
• High Risk of Errors & Inconsistency: The manual process led to frequent configuration errors with incompatible products and resulted in inconsistent pricing and proposals across the sales team.
• Over-reliance on "Tribal Knowledge": Critical configuration expertise was not documented, creating a major business risk when experienced employees left and making it difficult to train new staff.

a. Centralized Product & Data Management

• Unifying product data from three primary sources: trusted industry databases, internal legacy catalogs, and the sales team's expert knowledge base.
• Enriching the unified data with proprietary information such as custom pricing tiers, bundling rules, and internal service notes.
• Modeling key entities including manufacturer metadata, detailed product specifications, and a comprehensive list of all compatible accessories.
• Implementing a process for regular data refreshes to incorporate new products, pricing updates, and lifecycle changes.

b. Complex Relationship & Compatibility Modeling

• Enforcing a complex set of configuration rules (e.g., "must be coupled with," "incompatible with") between all engines and accessories.
• Enabling real-time compatibility queries, such as identifying all valid accessories for a specific copier engine.

c. Intelligent Sales & Configuration Tools

• Recommending optimal product configurations and bundles based on a customer's specified budget.
• Tracking product lifecycles to proactively identify obsolete items and suggest modern, equivalent replacements.

d. API-Driven Architecture & Interoperability

• Exposing a robust, well-documented API to allow seamless integration with other business systems like CRM or ERP.

To overcome these challenges, we engineered a platform with three core components:

• We implemented a sophisticated ETL pipeline using AWS Glue and Redshift to automatically ingest, cleanse, and unify inconsistent data from all sources into a single source of truth.
• We used Amazon Neptune, a graph database, to efficiently model and instantly query the thousands of complex compatibility rules between products.
• We built an intelligent recommendation engine using Elasticsearch and vector technology. It allows us to answer tough questions for the sales team:
  • Which configurations are replacements for a particular configuration?
  • Which configurations are cost-effective for customers?
  • Which configurations are good for supply and service in a physical location?

The Results: A Transformative Impact

• The impact of this new platform was immediate and significant. Leveraging our collaborative ODC model, we successfully delivered a system that has become the backbone of the client's sales operations.
• Today, the platform is actively serving 40 of our client's branches across the US, empowering 200 salespeople daily to build accurate, complex quotes with a speed and confidence that was previously impossible.

To bring this platform to life, our team needed to solve three core challenges that went far beyond typical software development. Each required a unique and modern approach.

a. The Data Integrity Challenge: Our biggest initial hurdle was the data itself. It wasn't just siloed in different places—it was inconsistent and often unstructured, with the same product having different names, specifications, or pricing across industry databases, legacy systems, and internal notes.

b. The "Compatibility Maze": Once we had clean data, we faced a complex architectural problem: how to model the thousands of intricate compatibility rules. A traditional database would have buckled under the weight of querying a "spider web" of relationships.

c. The Intelligent Search Problem: Finally, the goal wasn't just to store data, but to make it truly useful. The sales team needed to ask complex, human questions like, "What's a cost-effective replacement for this obsolete printer?" which a standard keyword search simply can't answer.