Acquire / Dispose
Book one-off purchases and sales into the forecast on the exact date they fall.
Most of a forecast is modelled: a projection of what usually happens. One-time purchases and sales are the part that is not — a stack of marketing flyers, a second-hand desk, a one-off consulting invoice. Each one has a known amount and a known date, and no rhythm for a model to project from.
What it changes
Every entry carries an amount, a category and a date, and the amount is added to that category's forecast in the period holding that date. Nothing is averaged into a monthly rate: the purchase shows up as a bump where it really falls, on top of whatever the model projects there.
The cells an item moves are highlighted blue in the forecast table, and the committed amount is shown under the number, so you can always see how much of a cell is a commitment rather than a projection.
Where the accuracy actually comes from
A modelled cost line spreads spending smoothly. Irregular spending does not arrive smoothly — three supplier invoices due the same week is an ordinary occurrence and a modelled average will never show it. That clustering is the thing that puts a business into an unplanned overdraft, and it is exactly what one-time items surface.
Dates need a judgement call
An invoice's due date and its payment date are not the same thing, and a forecast built on due dates is optimistic by however long your customers actually take. Where a customer reliably pays late, dating the sale to when they pay rather than when they should gives a forecast you can act on.
This extension earns its keep in businesses with lumpy spending — agencies between project payouts, shops between restocks. See restock planning for the inventory case.