Skip to main content



Bryan Flietstra, Manager, Advanced Analytics





Business Challenge

Steelcase offers our clients the ability to customize the look and feel of their office furniture through significant choice of products, sizes, finishes and services. This highly flexible, build-to-order business model allows our clients to create unique settings that fit their vision for the performance and aesthetic of their spaces. The business model is also complex, which makes it challenging for even experienced individuals to evaluate the underlying trends and opportunities to create effective pricing strategies. 

The data science and global pricing teams partnered to deliver a data-driven pricing model that assesses the specific mix of products, participants and other factors to recommend the appropriate discount strategy to maximize our win rate by identifying weak spots in our historical pricing practices.

Analytics Solution

Early explorations to understand the relationships between project win-rates and pricing included a static spreadsheet model that required manual inputs. The early learnings from this proof of concept suggested that there was an opportunity to further explore the relationships. Subsequent iterations included an expanded dataset and a dashboard to display the results. Still, there was significant opportunity to improve the pricing recommendation system. The current version improved the machine learning models and leverages API calls to embed real-time analytics directly within our internal pricing system. This system informs pricing decisions for all incoming pricing requests across multiple global sales regions.


Steelcase now has a pricing analytics tool that guides decision making throughout the sales and pricing process. By developing a best-in-class pricing system internally, we deliver analytics insights directly to the users at the point-of-sale. The models automatically score every inbound request for pricing and display the recommended discount strategy. As a result, we increased our speed in approving pricing requests, approve significantly more requests without manual intervention, and identified previous weak spots where over or underpricing was occurring.