A multi-channel brand’s bank balance is a poor guide to its cash position, because most of what determines the next 90 days has already happened and is sitting in other systems. A useful forecast reads those commitments rather than projecting a trend.
What the bank balance leaves out
Purchase orders already placed, which will become payments on supplier terms. Inventory already paid for, which will become revenue weeks later. Marketplace payouts in flight, arriving net of fees on each channel’s own schedule. Receivables from wholesale accounts, weighted by how late each one usually pays. Recurring outflows that only appear in the bank feed.
Why channel behavior matters more than the total
Two brands with identical revenue can have opposite cash positions, depending on which channels that revenue came through. A channel that pays out quickly with low fees and few refunds funds the business differently from one that holds a reserve, charges more, and refunds heavily after a promotional week. A forecast that treats all revenue the same way cannot see that difference.
What a forecast needs to read
Open purchase orders and their terms. Receivables and payables with aging. Bank and card accounts. Payout timing and actual fee rates by channel, from history rather than from the channel’s published rates. Refund behavior, again from history.
All of that already exists in most brands. The problem is that it exists in five places, so building the forecast means exporting from each of them and reconciling the result, which is why the forecast gets built once a quarter instead of continuously.
What changes when it comes from one system
The forecast stops being a project. The question shifts from whether the model is current to what to do about what it shows: whether to move a purchase order, chase a wholesale invoice, or hold off on a reorder for two weeks.
Focal models cash 90 days out from the purchase orders, receivables, payables, bank and card accounts and payout patterns already recorded, so the inputs do not have to be assembled first.
