The challenge
The client, a retail business, priced products using a mix of fixed markup rules and manual spreadsheet adjustments made whenever a category manager noticed a competitor's price had moved. Pricing decisions lagged behind demand and competitor changes by days, and there was no consistent way to weigh cost, demand elasticity, seasonality, and competitor pricing together.
Underpriced items left margin on the table, overpriced items sat in inventory, and the pricing team had no repeatable process for catching either problem before it hurt revenue.
The solution
Vizontra built a regression-based price prediction model trained on historical sales, product cost, seasonality, inventory levels, and competitor pricing signals, producing a recommended price band for every SKU instead of a single manually-set number.
The model is deployed as an API the pricing team queries directly, with recommendations feeding a review dashboard rather than auto-publishing — so pricing staff can approve or override suggestions with full visibility into what's driving each recommendation. The model retrains on a scheduled cadence as new sales and competitor data comes in.
Results
- Data-driven price recommendations generated for every SKU, replacing fixed markup rules
- Measurable margin improvement on categories with high price sensitivity after rollout
- Model retrains automatically as new sales and competitor pricing data comes in
- Feature importance surfaced per SKU, so pricing staff can see what's driving each recommendation
- Backtested against historical sales to validate pricing decisions before rollout
- Pricing team moved from reactive spreadsheet edits to a proactive, reviewable workflow