From Model to Production: Extending Company Z’s Manufacturing Taxonomy Into Live Deployment.

About the Client

Company Z's manufacturing division is responsible for formulating, packaging, and distributing products to more than 140 markets worldwide. Company Z's internal knowledge management team wanted to enhance and apply a custom taxonomy and auto-classification engine to its content and search tools, enabling its scientists and engineers to find relevant content more quickly across its manufacturing and research operations.

Industry Context

Moving a validated taxonomy model into a safe, reliable production deployment within a live enterprise content system is a distinct challenge, separate from building the model itself. Many taxonomy projects succeed at the modeling stage but stumble at deployment because production systems carry operational risks that development or test environments do not — a live SharePoint site supporting an active pharmaceutical manufacturing organization cannot be taken offline or risked for testing. This project addressed exactly that “last mile” gap for Company Z, building directly on Iknow’s earlier taxonomy development work for the division.

Current Situation

Iknow had developed Company Z's taxonomy on a project earlier in the year, and this project was a continuation of that initial taxonomy development work, focused on operationalizing the model, extending its coverage, and validating its performance in a live production environment.

• Moving a validated model into a live system required close coordination with IT. Company Z's internal IT team needed direct support to set up a proper development instance for autoclassification in SharePoint, work that fell outside the scope of pure taxonomy design.
• Specific facets required deeper real-world validation before scaling. In particular, the Analytics and Drug Name facets required further testing and enhancement with real manufacturing content before Company Z could rely on them broadly.
• Years of legacy classification data could not be discarded or blindly migrated. Some legacy classification data from Company Z's prior KM repository was worth preserving in the new system, while other data was not; Company Z needed a deliberate strategy for what to bring forward.
• Classification changes had to be tested without risking the live production system. Any changes to the taxonomy model and rules required rigorous testing, but Company Z could not afford to experiment directly in its live SharePoint environment.

• Support Company Z's internal IT team in setting up a SharePoint development instance for content autoclassification.
• Further test and enhance the taxonomy model, particularly the Analytics and Drug Name facets.
• Develop and implement a strategy to selectively include legacy classification data.
• Conduct an overall post-classification quality review in SharePoint, including recommended model and rule changes and validation testing.

How Iknow Structured the Work

Iknow combined direct IT enablement, targeted facet testing, a deliberate legacy-data strategy, and a full quality review, all built around a safe development-to-production testing approach that protected Company Z's live system throughout.

Key Activities & Decisions

• IT Team Support. Iknow worked directly alongside Company Z's internal IT team to help establish a development instance for auto-classifying content across SharePoint sites, providing the team with a safe environment to build and test before touching production.
• Facet Testing and Enhancement. Iknow performed further testing of the taxonomy model, focused primarily on the Analytics and Drug Name facets, and made enhancements as needed by drawing on manufacturing deviation reports, initial SharePoint document testing results, and other sources.
• Legacy Data Strategy. Iknow developed a strategy to determine which legacy classification data from Company Z's prior KM repository warranted inclusion in the new system and which should be excluded. Iknow then implemented that strategy in the taxonomy model and rules and tested it within the SharePoint development instance before rollout.
• Quality Review and Validation. Iknow conducted an overall post-classification quality review directly in SharePoint, including recommended changes to the taxonomy model and rules, followed by testing to validate that the changes performed as intended.
• Novel Testing Methodology. Iknow built a new testing approach that allows the taxonomy to synchronize with either the development or production term sets in SharePoint Online, enabling the team to review real SharePoint classification results without affecting the live system.

Stakeholders & Collaboration

Iknow worked as a subcontractor to the prime contractor, CGI, collaborating directly with Company Z's internal IT team throughout the establishment of the development instance, testing, and the eventual production rollout.

Enabling Autoclassification Work Without Risking the Live System

Company Z's IT team needed to build and test an auto-classification capability in SharePoint without disrupting the live system supporting active manufacturing operations. Iknow addressed this by establishing a dedicated SharePoint development instance for auto-classification work and then building a novel testing methodology that could synchronize the taxonomy with either the development or production term sets — allowing the team to safely review real classification results before anything touched the live environment.

Deciding What Legacy Data Was Worth Carrying Forward

Migrating legacy classification data wholesale risked importing outdated or incorrect tags, while discarding it entirely risked losing valuable historical classification work. Iknow addressed this by developing a deliberate inclusion-and-exclusion strategy for legacy data rather than a blanket migrate-everything-or-migrate-nothing approach, then implementing and testing that strategy specifically within the safe development instance before it touched production.

Operational Outcomes

• Expanded the taxonomy from roughly 1,400 to roughly 1,800 terms, with corresponding additions to acronyms and synonyms.
• Automatically tagged all legacy content from the prior KM repository in the new system with custom classification rules.
• Enabled Company Z's IT team to successfully bulk-classify all Knowledge Exchange content on schedule.

Strategic and Organizational Outcomes

The project delivered a novel testing methodology that enables the taxonomy to synchronize with either the development or production SharePoint Online term sets, providing Company Z with a safe, repeatable way to validate future classification changes without touching its live system. The new taxonomy’s value was recognized well beyond its original scope: it was extended to other Company Z systems, including Company Z’s research labs repository, even though it was originally built to organize manufacturing content.

Timeline to Impact

Iknow completed the four-month extension engagement on schedule, directly following Iknow’s previous taxonomy development project. Together, the two engagements constitute a continuous, single-year effort that took Company Z's taxonomy from an initial model to full production deployment and expansion.

Core Skills

• Taxonomy extension & enhancement
• Legacy content migration strategy
• Classification quality assurance
• IT enablement & deployment support

Methods & Frameworks

• Facet-targeted taxonomy testing and enhancement
• Legacy data inclusion/exclusion strategy development
• Development/production environment synchronization testing
• Post-classification quality review methodology

Technologies & Tools

• Microsoft SharePoint (development and production instances, SharePoint Online)
• Smartlogic Semaphore (taxonomy model and classification rules)
• Custom classification rules engineering