No black box — classical statistical models, validated in the open.
Every forecast is backtested out-of-sample before you ever see it.
01
From CSV to defensible forecast
Three steps. No notebooks, no parameter tuning, no consultants.
Drop in your demand history
A CSV with dates and quantities is enough. FlowForecast detects the frequency, gaps, and intermittent patterns on import.
| date | sku | qty |
|---|---|---|
| 2025-11-17 | A1234 | 668 |
| 2025-11-24 | A1234 | 689 |
| 2025-12-01 | A1234 | 712 |
| 2025-12-08 | A1234 | 694 |
Models compete on your data
ETS, Croston, TSB and seasonal naive are each backtested against held-out history. The most accurate model wins — per series, not globally.
See why, then export
Every number comes with its reasoning: selected model, detected seasonality, backtest accuracy. Export to CSV and get on with planning.
Weekly seasonality detected
02
Why FlowForecast?
Statistical forecasting that used to require a data science team — now accessible in a few clicks.
Clear explainability
See which model was selected and why. Understand detected seasonality, compare candidates, and review backtest metrics.
Automatic model selection
The system evaluates multiple forecasting methods and picks the best one for your data. ETS, seasonal naive, Croston — you don't need to choose.
Backtesting built in
Every forecast is validated with out-of-sample backtesting. You see the accuracy before you trust the numbers.
WAPE 8.2% · MASE 0.84 · BIAS −1.3%
No data science team needed
Designed for demand planners and analysts. Upload a CSV, configure your horizon, and export the forecast. That's it.
upload → forecast → export
03
Simple, transparent pricing
Start free with single-series forecasting. Upgrade when your team needs batch processing and advanced features.
04
Questions planners ask us
How much history do I need?
Twelve months of weekly data is a good baseline — enough to detect yearly seasonality. With less, FlowForecast still produces a forecast and tells you which patterns it can't validate yet.
Where does my data go?
Your upload is used to compute your forecasts — nothing else. It is never used to train models and never shared. Our infrastructure runs in the EU (Frankfurt). Details are in the Privacy Policy.
We plan in Excel today. How does this fit?
Keep Excel for what it's good at. FlowForecast replaces the fragile forecasting part: upload your history, take the validated forecast back as CSV, and keep your downstream planning sheets.
Which forecasting models are used?
Proven statistical methods: ETS (exponential smoothing) for trended and seasonal series, Croston and TSB for intermittent demand, seasonal naive as the baseline to beat. The winner is chosen by backtest accuracy.
Do I need a data scientist to use it?
No. It's built so a demand planner can do this alone: upload, review the explanation, export. If you do have a data team, the backtest metrics give them everything they need to audit the results.