Trade Promotion ROI

Why 60% of promotions lose money — and how to identify yours. A briefing for VPs of Trade Marketing, Sales, and Commercial in FMCG.

DATAQUANT RESEARCH TEAM  ·  TRADE & PROMO  ·  10 MIN READ

The most reliable finding in FMCG trade promotion analytics, repeated across markets, categories, and brands for the past two decades, is that approximately 60% of trade promotions destroy value. The exact percentage varies — some studies put it at 55%, others at 70% — but the directional answer has been stable across decades of category research.

This is uncomfortable. Trade promotion is the largest line of marketing investment in most FMCG businesses — typically 15–20% of net revenue. If 60% of it is value-destructive, the implications for the P&L are profound. They are also avoidable: not every promotion has to lose money. The promotions that do lose money are systematically identifiable in advance, and the ones that work are systematically identifiable too.

This piece is for the Trade Marketing Director or VP of Sales who wants to move beyond gut-feel promo planning to a quantitative basis for go/no-go decisions — without requiring a full data-science transformation.

Why most promotions lose money

A trade promotion creates incremental volume only if it changes consumer behaviour in one of three ways: pulls a non-buyer into the category, pulls a competitor’s buyer to your brand, or accelerates a future purchase that was going to happen anyway. The third type — forward-buying — is volume that the brand was going to get anyway, sold at a discount it didn’t need to give. The first two are genuine incremental volume.

The financial test of a promotion is whether the genuine incremental volume — minus the cost of the discount across both incremental and non-incremental volume — produces positive contribution. The reason 60% of promotions fail this test is that the non-incremental volume (forward-buying, pantry-loading, brand-loyal repeat) is typically the largest share of total promoted volume. The discount on that volume is pure margin loss with no behavioural payoff.

The 90/10 trap

A typical promo lift looks like this: 100 units of baseline weekly volume becomes 350 units during the promotion week. Out of the 250 incremental units, perhaps 80 are genuinely incremental (new buyers, switchers from competitors). The other 170 are forward-buyers — customers who would have bought next week or the week after but moved their purchase forward to capture the discount. The discount is paid on all 350 units, but only 80 of those units represent genuine incremental contribution.

When the post-promotion period is included — the weeks after the promotion when forward-buyers are not in the market — baseline volume often dips below 100, recovering over 4–6 weeks. This is the structural reason promo ROI calculations done at the promo week alone systematically overstate value: they ignore the dip.

Three diagnostics that identify the losers

Three analytical lenses, applied to your historical promo data, identify the structural losers in your portfolio. They do not require advanced data science — the math is straightforward, the data is in your sales system, and the answers are usually stable enough to act on.

Diagnostic 1: Baseline-corrected lift

For every promotion, calculate the lift relative to the underlying baseline trend (not the prior week). A category growing at 6% year-over-year produces apparent promo lifts that are partially category trend, not promotion effect. Strip out the category trend; what remains is closer to the genuine promotion effect.

Diagnostic 2: Post-promo dip measurement

For 6–8 weeks following each promotion, measure baseline volume against the pre-promotion baseline. The dip is the forward-bought volume that was pulled forward into the promo week. Net promo lift = promo-week incremental volume minus the cumulative dip. Most FMCG businesses do not measure the dip systematically, which is why promo ROI calculations are usually too generous.

Diagnostic 3: Cannibalisation across SKUs

When you promote SKU A, do you steal volume from SKU B in your own portfolio? In most FMCG categories, internal cannibalisation is 15–40% of apparent promo volume — the customer was going to buy from your portfolio anyway, and the promotion just shifted them to a lower-margin SKU. This cannibalisation is invisible at the SKU level and only becomes visible when the analysis is done at the brand or category level.

Building a promo evaluation capability

A working promo evaluation capability requires three components, in order of priority:

Component 1: Pre-promotion P&L modelling

Before approving any promotion, model the promo at the SKU × retailer × mechanic level: expected lift, expected forward-buying %, expected dip duration, expected cannibalisation, all-in contribution including discount cost. The output is a single number: expected promo profit. If it is negative, the promotion needs to be redesigned or rejected.

Most FMCG businesses do not do this. They approve promotions on the basis of “we did this last year and it worked” or “the retailer is asking and we don’t want to upset them.” Both reasons are operationally legitimate and analytically inadequate.

Component 2: Post-promotion measurement

For every promotion that runs, measure the actual outcome 8 weeks post-event. Compare to the pre-promo model. The variance is the learning. Promotions where actual outcome materially undershoots the model should be redesigned or retired in the next planning cycle.

Component 3: Promo portfolio optimisation

With a year of pre-and-post promo data, the next step is to optimise the entire promo portfolio: which mechanics work for which brands at which retailers, which depths produce best return, what timing in the calendar produces best lift. This is where machine-learning models add value — but only after the basic measurement infrastructure of components 1 and 2 is reliable.

What this looks like in practice

A €280M personal-care manufacturer we worked with had a promo budget of €38M annually. Pre-engagement evaluation showed:

• Approximately 22 of 51 evaluated promotions were value-destructive on a baseline-corrected basis — cumulative loss of €4.1M annually
• Internal cannibalisation across the portfolio was running at 28% of apparent incremental volume
• Forward-buying was averaging 51% of total promoted volume — most apparent lift was timing-shift, not behaviour change

The remediation was structural rather than analytical. Three of the value-destructive promo mechanics were retired entirely. Five were redesigned with shallower depth and broader pack range. Two were repositioned to different retailers where the elasticity profile was more favourable.

In the next 12 months: net promo profit improved by €6.8M on a flat overall promo budget. The gain came roughly half from retiring losers, half from improving the design of the keepers. The total volume of promotional activity was almost unchanged — the change was in which promotions ran and how they were structured.

Three pitfalls

1. Measuring promo at the promo week only. Almost all promo evaluations done internally measure promo-week incremental volume and stop. This systematically overstates value because forward-buying — which is captured in the promo week as apparent lift — is structurally non-incremental at the customer level. Always measure across the full pre-promo + promo + post-promo window.
2. Comparing across retailers without controlling for category dynamics. A promo in Retailer A may produce 4x more lift than the same promo in Retailer B — not because Retailer A is better, but because the category at Retailer A was already trending positively. Cross-retailer comparisons require category-trend adjustment to be analytically meaningful.
3. Treating retailer requests as non-negotiable. A common pattern: the retailer asks for a specific promo mechanic, and the trade marketing team approves it because saying no is uncomfortable. The financial analysis that shows the promo will lose money is rarely surfaced to the retailer conversation. Conversely, retailers respect counter-proposals supported by data — a redesigned promo with quantified profitability often lands better than the original ask.

Closing thought

Trade promotion is one of the few P&L categories in FMCG where the analytical answer to “are we doing it right?” has been clear for decades and is still ignored at most businesses. The reason is not that the analytics is hard — the diagnostics described here can be built on existing sales data in 6–8 weeks. The reason is that the analytics produces uncomfortable findings — specific named promotions that are losing money, specific retailer relationships where the trade-spend rationale is weaker than the team has assumed, specific brand-level cannibalisation that the SKU-level analysis hides. Acting on the findings requires commercial leadership willing to renegotiate retailer relationships and retire promo programmes that are sacred internal cows. Without that leadership, the analytics produces a deck that no one operationalises. With it, 6–9 percentage points of trade-spend efficiency become available within a year.