What is the cashflow forecast?
The cashflow forecast projects an operating-cash proxy three months ahead: each month's contribution margin, minus that month's ad spend. Refunds are in there too — contribution margin is already net of them — so they are not subtracted a second time. It is not a bank-account model — no opex, no VAT remittance, no payment-provider settlement lags — but it tracks the largest operational flows, and unlike the revenue forecast it is allowed to go negative.
Formula
monthly value = contribution margin (net of refunds) − ad spend
forecast = trend(last 12 complete months), projected 3 months, NOT floored at zero
Worked example
A store's monthly series runs +€9,000 to +€14,000 through autumn, then Q4 ad scaling pushes spend up faster than margin: December reads −€2,000. The projection carries that slope forward to roughly −€5,000 by February. Whether that is alarming depends on why — planned acquisition investment reads very differently from drift — but the widget's job is to make the slope visible while there is still a quarter to react.
How Saldo Metrics computes it
Two monthly series are combined: contribution margin summed from
v_contribution_margin_all (every order that was not cancelled, EUR) and ad spend
from v_blended_roas_monthly (the fact_ad_spend basis, own-store ad channels only —
marketplace ad spend is not subtracted, while the margin series covers every order).
The series covers the last 12 complete months; the current month is left out.
The margin series deliberately includes fully refunded orders, which the order-count and average-order-value metrics leave out. Cash is a booked question: the shipping, marketplace fees and discount on a fully refunded order were spent and are not coming back, so such an order enters here at exactly minus those costs rather than disappearing.
There is no separate refund series, and that is the point. v_contribution_margin_all
reverses a refund against the order that caused it — the refunded revenue, and the
product cost of any units that physically came back — so a refund is already
inside the margin figure. Subtracting refunds again here would deduct every
refunded sale twice. Note that this books a refund into the ORDER's month, not the
month the credit note cleared; the P&L widget does the opposite, on purpose — see
the contribution-margin page.
The combined series is projected with the same selectable methods as the other forecasts, and negative projections are shown as-is — a warning is the point.
Why it matters
Profitable-on-paper stores die of cash timing. This proxy strips the question to its operating core: does a month's margin — refunds already deducted — cover what marketing takes out of it? A declining slope here, months ahead of the bank balance feeling it, is the cheapest early warning the data can give.
Common mistakes
- Reading it as a bank-balance forecast. Opex, VAT, inventory purchases and settlement delays are all outside it; the P&L widget carries the opex side.
- Ignoring the ad-spend basis. Spend here comes from the channel-level spend facts, matching blended ROAS — not the creative-level performance rows, which can differ slightly in coverage.
- Panicking at one negative month. A planned campaign push makes single months negative by design; the projected slope is the signal, not one bar.
Where you see this in the app
The Cashflow Forecast dashboard widget, with the projection method selectable per widget.