Revenue Forecaster

Revenue Forecaster

Detect trends and seasonal patterns in your historical data, project future outcomes with uncertainty ranges, and test scenarios before committing to a plan.

Include a header row with column names (date,value). Dates in YYYY-MM or MM/YYYY format.

Loads 12 months of example revenue from a neighborhood bakery and generates a forecast.

Forecast

Add historical data and generate your forecast.

How the forecaster works

The Revenue Forecaster uses a simple two-step decomposition to project future outcomes:

This is a simplified forecasting approach — not a machine learning model. It works best for data with a clear direction (growing or declining) and regular seasonal patterns. Structural breaks, campaigns, or market shifts may cause significant deviations from the forecast.

Forecasting FAQ

What data format is accepted?

Paste a CSV or tab-separated spreadsheet with two columns: a date column (header must include "date", "month", or "period") and a value column (header must include "value", "revenue", "amount", "conversions", or "sales"). Alternatively, use the manual entry option.

How much data do I need?

At least 3 months of data is required for any forecast. More data produces better results — 12+ months is ideal, especially if you want seasonal detection. With less than 6 months, the confidence bands will be wide.

Why are the confidence bands wide?

Uncertainty grows the further you forecast into the future. The model uses historical residual variance to calculate bands — if your data is volatile month-to-month, the bands will be wide. This is honest, not a flaw.

What does the growth target probability mean?

If you enter a growth target, the model calculates the probability of hitting that average monthly level over the forecast horizon, based on your historical variability. A low probability means the target is ambitious relative to your current trajectory.

Can I model the impact of a campaign or price change?

The base forecast assumes your current trajectory continues. You can approximate campaign impact by entering an adjusted target and comparing scenarios, but the model does not explicitly handle step-change interventions.