Nirvana Case Studies - Product classification services for a very large portal's shopping site

Back Office Processing

< Previous | Next >

Product classification services for a very large portal’s shopping site

Client Profile
One of the largest portals in the world

Situation
The client had over 100,000 storefronts on its portal. In order to facilitate shoppers’ search, the portal used a classification system with over 1,400 categories. These categories were meant to mirror the average shopper’s use of search terms for the product. As new products were added to these stores, they needed to be classified to facilitate shoppers’ search. Approximately 90% of the classification was automated. The residual stock keeping units (SKUs) required trained human judgment to classify in a manner which would best serve the average shopper’s search approach i.e. the classification should best reflect how customers would view the product. Opinions could and did vary on how to classify products which could justifiably belong to multiple categories.

Productivity and Accuracy
Productivity and accuracy while processing large volumes meant that scalability of the process was a key concern. The automation developed for classification had a ‘learning’ feature which held the potential for reducing volumes required for manual classification; however future volumes in new SKU introduction were unclear – in effect making the outsource contracting process difficult (committing work volumes and volume variations).

A further complication was that the human resource requirements were such that a career path was difficult to provide. After considering the pros and cons, it was apparent that outsourcing to a specialist service provider was the best way forward. The outsourcing process would require considerable effort to stabilize and required a service provider who could devote adequate management attention to a relatively small but tricky process.

Nirvana Solution

Developed the resource profile (skills/aptitude/qualification) for carrying out such as process effectively.
Starting with a larger team, progressively whittled the team down to the best fit resources to invest on for intensive training.
Designed, documented and deployed the process, this involved establishing resource profiles, metrics, decision support tools, training and certification of associates, QC methodology, escalation process, reporting, billing etc).
Refined existing tools for applying consistent classification included a ‘tip sheet’ tool.
Productivity/cost ‘discovery’ based on the pilot followed by a transparent pricing and contracting process for scaled up service delivery. Made possible by close senior management involvement in developing the partnership environment which enables ‘non standard’ services to be outsourced
successfully.