Data Science Consulting:
About Darwin Pricing LLC
- Dynamic Pricing software and Data Science consulting since 2013
- Founded in Aarau (Switzerland)
- Exhibitor at the Web Summit in Lisbon (Portugal)
- Investment by the Swiss Startup Factory in Zurich (Switzerland)
- Geo-Pricing solution used by 700+ online retailers, with a focus on the US market
- Development of the Dynamic Pricing solution of OTTO.de (€2.7bn yearly turnover)
- Development of the Demand Forecasting solution of H&M (€22bn yearly turnover)
Goals of Dynamic Pricing
Automatic price adjustments during the season. Business goals:
- Maximizing overall revenue and profit
- Seasonal articles: Selling until the end of the season with the highest possible profit margin
- Permanent articles: Avoiding clearance before the next stock receipt
- Early detection of overstock and bottlenecks and proactive action
- Reducing remaining stocks with effective sales, avoiding liquidation costs
Method
Core of the Dynamic Pricing system: Sales Forecasting
- How do prices affect purchasing behavior?
- Weekly forecast for each article, depending on price list, assortment and seasonality
- Extrapolation until the end of the season, considering stocks and seasonality
- Determination of the optimal price list
- Corrections as needed during the season
Requirements
Quality and size of the data base are crucial for sales forecasting.
Better data ⇨ More precise sales forecast ⇨ Best results and stable prices
- Relevant product details: category, brand, size, color, material, weight, quantity...
- Current and historical sales data: Assortment, prices, stocks and sales on a daily basis
- Current and historical competition prices, when available
- Purchase prices, recommended retail prices
- Date of the seasonal sales and delivery dates for inventory optimization
- Liquidation costs for remaining stocks
Business Rules
Additional pricing restrictions:
- Minimum price: e.g. purchase price plus VAT
- Maximum price: e.g. 50% above RRP
- Uniform prices: e.g. all article sizes for the same price
- Price steps: e.g. price changes in $10 increments for items in the range ± $100
- Update frequency: e.g. no more than one price change per week for each article
- Price lock: e.g. no price changes at product launch or for certain brands
Technical implementation
- Secure web application in the Amazon cloud (USA or EU)
- 24/7 operational security with the cloud platform Red Hat OpenShift Online Pro
- Daily synchronization of product, sales and warehouse databases
- Sales forecast with artificial neural networks
- Inventory optimization by extrapolation based on the seasonality curve
- Daily selection of the optimal price list based on simulations
- Prices as stable as possible, price changes only as often as needed
- Automatic price changes through an API, possibly after manual release
Achieve the same business goals more efficiently and reliably
Dynamic Pricing:
- Systematic, extensive
- Controlled, data-based
- Forward-looking, predictable
- Considers many factors: Demand, seasonality, stocks, delivery dates, sales, gross margin...
Manual pricing:
- Case-by-case, seldom
- Gut feeling, personal experience
- Reactive, hectic
- Addresses goals one at a time: Profit margin, purchase frequency, customer acquisition, sales, stock clearance ...
