How to find the 15–25% hiding in your promotional history

Ask most FMCG suppliers how their promotions performed last year and you will get an uplift number. The event sold 340% of base. It sounds like a result.

It isn't one. Uplift tells you what happened to volume. It tells you nothing about whether the money you spent to buy that volume came back, and in a lot of promotional calendars a meaningful share of it doesn't.

The number that actually matters

The question a promotion has to answer is narrow. Did the incremental margin it generated exceed the investment it consumed?

That requires three things most reporting doesn't separate:

A defensible baseline. What would have sold anyway, without the promotion. Not last year's same week, which was probably also on promotion.
Incremental volume rather than total volume. Everything above that baseline, and only that.
The full cost of the event: off-invoice, scan spend, co-op contributions, and the margin given away on the units that would have sold at full price regardless.

That last one is where most promotional maths falls apart. If a third of the volume moved on deal would have sold anyway, you have funded a discount on it for nothing. That subsidy is real money, and it rarely shows up in a post-promotional review.

Three leaks that show up almost every time

Pantry loading

A deep promotion on a storable category pulls forward demand rather than creating it. Volume looks excellent in week one and poor for the following three. Evaluate the event in isolation and it passes. Evaluate an eight-week window around it and the incremental volume shrinks, sometimes to nothing.

Cannibalisation within your own range

The promoted SKU performs. The two adjacent SKUs it stole from don't get counted. Portfolio-level incremental volume ends up well below the SKU-level number that got reported to the team.

Frequency erosion

A line promoted every third week has effectively repriced itself. Shoppers wait. The base you are measuring uplift against has already been dragged down by the promotional pattern itself, which flatters every event that follows.

What the audit looks like in practice

You need twelve to twenty-four months of promotional history, scan data at SKU level, and the actual investment by event. Most suppliers have all three. They are just held in three different places by three different people.

From there:

Rebuild the baseline from non-promoted periods, adjusted for seasonality and distribution changes.
Calculate incremental margin per event: incremental units at the promoted margin, less the discount funded on baseline units, less all direct investment.
Rank every event by ROI rather than by volume. The ranking is almost always a surprise to somebody.
Look at the bottom decile honestly. In most calendars a group of events has never paid for itself and keeps running because it is on the plan.

Why the ranking matters more than the average

An average promotional ROI across the calendar is a comfortable number that hides the shape of the problem. The distribution is what you act on, and it usually looks like this: a set of events that pay well, a large middle that roughly breaks even, and a tail that loses money reliably.

Reallocating from the tail toward the top of the ranking is the fastest margin improvement available in most FMCG businesses. It costs nothing to implement, because the spend already exists. It doesn't require a price increase or a retailer concession either.

The harder conversation

The reason this analysis often sits undone isn't technical. It is that the tail of the calendar usually contains events somebody promised a customer, and retiring them means a conversation.

That conversation is easier with the evidence in hand. "We're reducing promotional depth" is a position you have to defend. "This event has run six times, and here is what it returned each time" is a position the numbers defend for you.

Northform works with New Zealand FMCG suppliers on pricing, trade investment, portfolio mix and customer terms. If your promotional history has never been audited this way, that's a normal place to start.