Dynamic Pricing in Retail: 10% Profit Growth During Pilot

"Three Prices" is an international retailer with an extensive network of stores across 144 cities in Belarus and Russia. The company is actively growing: expanding its geographic presence while strengthening its position in established regions. To address strategic business objectives, the company needed a solution that would enable flexible price management across its vast product assortment and increase profitability even in highly competitive and regulated conditions.

At the project's launch, the company had its own effective pricing system, but the team understood that reaching the next level required more nuanced, adaptive pricing based on real demand analysis and market dynamics. This became the starting point for implementing dynamic pricing.

The project's goal was to automate pricing processes and, using intelligent price optimization algorithms, unlock the potential for gross profit growth. The target metric was a 5% increase in gross profit across 20 pilot stores.

The pilot project results exceeded target metrics and confirmed the hypothesis about significant potential in managing prices across the entire product assortment:

• Gross profit growth for the pilot store group relative to the median and compared to the control group reached 10.1% against the 5% target

• Positive average check dynamics were maintained - dynamic pricing algorithms found the optimal balance between demand and margins

• The pilot's positive results became the starting point for full implementation; gradual scaling of dynamic pricing to all retail locations is currently underway

The project's main challenge was not technical integration but fine-tuning the approach to the specifics of the Belarusian market and embedding the new process into real business operations. The key challenge was compliance with Resolution No. 713 of the Council of Ministers "On the Price Regulation System." This required developing mechanics for automatically excluding regulated goods from pricing algorithms without losing effectiveness across the remaining assortment.

Equally important was precise data synchronization with the accounting system. Deep verification of gross income calculation methodologies revealed risk factors for discrepancies (such as different principles for accounting VAT and discounts); eliminating these became the foundation for correct algorithm operation. Instead of implementing a standard solution, the team embedded the new approach into existing processes, ensuring system stability and reliability.

The project started with an audit of current pricing processes and data integration from the 1C accounting system: sales, inventory, cost prices, promotional campaigns, product catalogs — all considering the requirements of Resolution No. 713.

Next, test stores were selected where the Imprice system would handle pricing, and control pairs were identified for accurate results comparison — stores most similar in all key characteristics (overall sales dynamics, traffic, product replenishment, store format, etc.). After analysis, 20 stores in Belarus were selected with a control pair for each.

Defining the key metric completed the preparatory stage. The goal set for the pilot project: increase gross profit of the pilot group by 5% relative to the control group while maintaining positive check dynamics in test stores.

Several key pricing mechanisms were implemented to achieve target results:

1. Assortment clustering into baskets (product classes): KVI, BackBasket, and Longtail — 7 strategic segments total

2. ML price optimization with a unique strategy for each segment. Imprice algorithms selected optimal prices for all product categories based on historical data, real-time demand testing, and within the target strategy:

• Hard KVI — focus on maintaining traffic and increasing sales volume

• Back Basket — maximizing gross profit without losing demand

• Long Tail — price experiments to accelerate turnover for slower-moving items while maintaining margin levels

3. Intelligent repricing queue. The system considered how many price tags staff could replace in the sales floor and prioritized products with the greatest growth potential.

Beyond core mechanics, special attention was paid to operational aspects during implementation:

• Created a double-check protocol for system recommendations to avoid operational failures and give the team confidence in each price tag

• Ensured seamless integration into merchandising processes

Testing lasted 2 months — July through September 2025. Twice weekly, the Imprice system calculated new prices, which after verification were transmitted to the 20 pilot stores. Work on the project was collaborative: "First BIT" formed the strategy and analyzed key indicator dynamics, the Imprice system provided flexible algorithms, and the "Three Prices" team daily confirmed results with actual sales.