The forecast is editable because no model knows about the things you have planned. This guide is for whoever owns the number the business is working to. It covers what the forecast is built from, which assumptions are worth your attention, and how confirming a month changes what happens afterwards.
What the baseline is built from
The forecast starts from the previous month’s real result rather than from a target. That is deliberate. A forecast anchored to what you achieved stays honest, and one anchored to what you hoped drifts further from reality every month it is missed.
From that baseline it projects forward using your growth assumptions, seasonality where there is enough history to establish it, and, if subscriptions are connected, the recurring revenue already contracted for the months ahead.
That last part is the most valuable input a forecast can have, because it is the only component that is close to known rather than estimated. Everything else is a projection. Locked subscription revenue is a commitment.
What is worth editing
Not everything. The temptation is to adjust the forecast whenever it disagrees with your instinct, which converts it from a model into a wish.
Edit it when you know something the model cannot. A promotion planned for next month. A product launching. A supplier problem that will limit what you can sell. A large wholesale order already agreed. These are real events with no historical trace, and the model has no way to anticipate them.
Do not edit it because last week was soft. Weekly variance is normal and adjusting for it produces a forecast that chases noise and is revised constantly, which destroys its usefulness as a planning tool.
Confirming a month
Confirming a month fixes it as the plan. That matters for the same reason locking operating expenses matters: a number that keeps moving cannot be measured against.
Once a month is confirmed, variance against it becomes meaningful. You can ask why you came in under plan, and the answer is informative. Without a confirmed plan, the comparison is against a forecast that has been quietly revised towards whatever actually happened, and every month looks approximately correct.
Confirm the plan before the month begins. Confirming it halfway through, when you already know how it is going, produces a plan you will always hit and never learn from.
Reading variance
When actuals diverge from plan, the useful question is which input moved. Revenue is traffic multiplied by conversion rate multiplied by average order value, and a revenue miss can come from any of the three with completely different responses.
A traffic miss is usually a channel problem and shows up in the marketing mix. A conversion miss is usually a site or offer problem and shows in the website funnel. An order value miss is usually a mix problem, often a discount or a promotion pulling volume towards cheaper products.
The forecast tells you that you missed. It does not tell you why, and treating the number itself as the problem leads to the wrong action, which is almost always more spend.
A practical rhythm
Confirm next month’s plan before it starts. Leave it alone during the month unless something genuinely new happens. Review variance at month end and record which input moved. After a few months you will know which of your assumptions is habitually wrong, and correcting that is worth more than any individual month’s adjustment.