AUTOMATED FORECAST ENGINE

Upload financials. Get a validated forecast.

Drop in a CSV with a period column and a numeric value column. The engine tests several forecasting methods against your own held-out history, picks the most accurate one, and projects it forward — with error bounds attached.

Try it free — no signup Book a call

Typically respond within 1 business day

WHAT YOU'LL NEED TO PROVIDE

A period column

One row per period — monthly is typical. "2024-01" or "Jan 2024" both work.

A single numeric metric

Revenue, units sold, or another metric you track consistently.

At least 6 periods of history

12+ is better; 26+ monthly periods lets the engine test for seasonality, not just trend.

Actuals, not projections

Every historical row should be a real, closed-out number — not a target or earlier forecast.

This is a demo build: uploads are processed entirely in your browser and nothing is saved or transmitted.

Drag a CSV here, or

Choose file

CSV format: first column = period label, second column = numeric value

PREVIEW HORIZON periods ahead

This free preview is capped at a short-range forecast. A full engagement includes the complete multi-period forecast, scenario planning, and ongoing monitoring.

ACTUAL VS. BACKTEST VS. FORECAST

Actual

Backtest

Downside

Base case

Upside

SELECTED MODEL

BACKTEST MAPE

BACKTEST RMSE

MODEL COMPARISON

Method	MAPE	RMSE

The engine tests each method against a held-out period of your own data it hasn't seen, then automatically selects the one with the lowest error before projecting forward.

UNDERSTANDING WHAT YOU'RE LOOKING AT

What these numbers mean for your decision

COMMON QUESTIONS

Is my data safe?

In this demo, everything runs in your browser — nothing is uploaded or stored anywhere. In a real engagement, your data stays in a private, access-controlled workspace only you and I can see.

What if I don't have 2 years of history?

You can still get a forecast with as few as 6 periods. You'll just get trend-based models rather than seasonal ones — the engine tells you exactly which methods it could and couldn't test, so nothing is hidden.

How is this different from a spreadsheet trendline?

A trendline is fit and shown with no accountability. This engine holds back real historical periods, tests several methods against them, and reports the actual error rate before you ever see the forward-looking number.

What if my business doesn't show any seasonality?

Then the engine simply won't force a seasonal model on you — it only uses one if it improves accuracy against your real held-out data.

Can this update automatically as new numbers come in?

Yes — that's what the retainer engagement is for: your forecast is re-validated on a schedule so accuracy doesn't quietly decay over time.

Why does the free preview only show a couple of periods?

This demo is meant to show the method and the accuracy score, not replace a full engagement. A real engagement includes the complete forecast horizon, scenario planning, and the ongoing monitoring that makes it useful for actual decisions.

Upload financials. Get a validated forecast.

What these numbers mean for your decision

Let's find out what your numbers are trying to tell you.