Before
- Weekly gut-feel forecasting for hundreds of promo SKUs
- Forecast variance swinging wildly up to ±44%
- Simultaneous overstocks and stockouts in the same store
- 30 hours/week locked in spreadsheet forecasting
Customer stories
Forecasting & execution
Forecasting & executionRohlik protected €593K of promo sales.
Measured result
€593K
promo sales protected
303
sold-outs prevented
Weekly promo forecasts produced variance of up to ±44%.
Rohlik planners forecast hundreds of promo SKUs each week across five country operations.
Forecast variance ranged from ±20% to ±44%, creating simultaneous overstocks and stock-outs in the same store.
Spreadsheet forecasting also consumed 30 planner hours per week, leaving less capacity for longer-term demand work and category strategy.
After Duvo
- Daily, agent-generated forecasts across all 5 markets
- Orders placed automatically from primary signals
- Forecast variance tightened to a consistent ±22–26%
- Planners only review exceptions via Slack
Our planners used to spend 30 hours a week trapped in spreadsheets, running gut-feel forecasts that often led to simultaneous overstocks and stockouts. Now that Duvo runs daily forecasting from primary signals and automatically places the orders, we've eliminated quarterly disaster weeks and returned nearly four FTEs of capacity to long-horizon demand planning.
Supply Chain Director
Rohlik Group
The workflow from signal to outcome.
01
Pulls the data
Duvo pulls campaigns, sales history, availability and pricing data daily across all 5 markets.
02
Generates the forecast
Duvo turns the campaign, sales, availability, and pricing inputs into a daily per-SKU forecast.
03
Places the order
Places the order automatically based on the generated forecast.
04
Flags exceptions
Flags missing data via Slack for the planner to resolve, so they spend time on exceptions, not running the forecast itself.
Evidence status
Measured result
Rohlik directly measured 303 sold-outs over the live forecasting period. The €593K result equals sold-outs multiplied by average promo-SKU revenue contribution.
Systems involved
Data warehouse, Ordering system, Slack
Control points
- Missing data routes to planners in Slack instead of being used silently.
- Orders are placed from the daily forecast built from campaign, sales, availability, and pricing signals.