Four programmes, one transaction
The defining feature of agricultural input programmes is stacking. Prepay, early order, volume, brand mix, retailer performance — each is a reasonable instrument on its own, and each is typically designed by a different part of the commercial organisation. What almost nobody calculates before launch is the combined effective rate on the transaction that qualifies for all of them.
That number is knowable. Modelling it against last season’s actual volume mix takes minutes if the programme rules are executable and is effectively impossible if they are prose. The benchmark target is above 95% of programmes modelled before launch; typical practice is barely half.
The accrual that carries the year
Because programmes settle after the season, the liability sits on the balance sheet as an estimate for most of the year. A 15 to 30% true-up variance is common — which is to say, the number the business planned against was materially wrong for two to three quarters.
The cause is almost always that the accrual is computed from aggregate volume against an assumed rate, rather than from the transaction lines against the actual stacked rules. Transaction-level accrual is what collapses the variance to under 3%.
The grower you cannot see
Grower-level rebates are earned by the end farmer but transacted through a retailer. The manufacturer therefore depends on retailer sell-through and grower registration to know who earned what. Where that data arrives late, partially, or in identifiers that do not resolve, grower earning is approximated — and approximation in an incentive programme always resolves in favour of the party holding the data.
How RevUpra runs this
Programme rules become executable, so a stacked combination can be modelled against real historical volume before it launches and monitored against actual volume while the season runs. Accruals are computed from transaction lines against locked periods, so the true-up stops being a surprise. Retailer sell-through is ingested with schedule monitoring that alerts on a missing period rather than failing silently, and grower identity is resolved through cross-reference. Season-end returns reverse against the original accruing lines automatically, because the accrual knows which lines created it.