Incremental vs. Base Volume in CPG Promos

Incremental volume is the portion of promoted sales that would not have happened without the promotion. Total sales during a promo event = base volume (what you would have sold at everyday price) + incremental volume (the lift the promo generated). The ratio between the two determines whether your trade dollars are buying real growth or just subsidizing purchases consumers were already going to make.

This post walks through how CPG sales and insights teams define, model, and act on the base-versus-incremental split. It covers baseline modeling, lift calculations, cannibalization, pantry loading, and a worked numeric example you can adapt to your own data.

Base vs. Incremental Volume

Every promoted week has two components stacked on top of each other. Base volume represents the demand a brand would capture through normal velocity at full price, with no feature or display support. Incremental volume sits on top: it is the additional units moved specifically because of the promotion.

The split matters because trade funding is sized against expected incremental. If a brand ships 10,000 cases during a TPR week but only 1,500 of those are incremental, the other 8,500 are base units the brand gave away at a lower price for no revenue upside. That gap is the difference between a profitable event and one that destroys margin.

The table above shows how each figure flows from the one before it. Base volume is the only number that is modeled rather than observed, which is why baseline methodology is the most contested element of any promo analytics conversation.

How Baselines Are Modeled

A baseline is a counterfactual: what sales would have been if no promotion ran. Because you cannot run the same week twice, the baseline must be estimated.

Common baseline methods

• Pre/post moving average: average the N non-promoted weeks before and after the event. Simple, but it absorbs seasonality as if it were noise.
• Comparable-period method: use the same calendar period from the prior year, adjusted for trend. More reliable for seasonal items (holiday candy, sunscreen) but requires clean year-over-year comparables.
• Regression-based models: control for price index, ACV, season, holidays, and retailer factors simultaneously. Syndicated data providers like Circana (IRI) and NielsenIQ embed proprietary regression baselines in their promoted-sales decompositions.
• Account-level customization: some brands build retailer-specific baselines because a Kroger baseline behaves differently from a Walmart or natural-channel baseline at the same price point.

No method is perfectly neutral. Regression models trained on historical data can understate the baseline if past promotions were very frequent (the model treats promotional price as "normal"), which inflates apparent incremental. Teams that run 60%+ ACV on promotion for long stretches often discover their baselines are soft.

Why baselines differ across data providers

When you pull promoted-sales decomposition from SPINS versus Circana versus your own retailer portal data, the baseline figures will rarely match exactly. Each provider calibrates to its own panel, applies different seasonality adjustments, and runs proprietary smoothing. The practical implication: pick one methodology as your measurement standard, apply it consistently across your portfolio, and compare programs against that standard rather than against other sources. Scout surfaces the raw scan data across providers in a single view so teams can see which baseline assumption is driving the incremental story.

Lift and Incrementality

Lift % is the most common shorthand for incremental performance. It tells you how many times over base volume the promoted event generated.

Lift % = (Incremental Volume / Base Volume) x 100

A 200% lift means a brand sold 3x its baseline during the event: 1x base plus 2x incremental. A 50% lift means it sold 1.5x baseline.

Lift varies enormously by category, retailer format, price depth, and feature support. A mid-tier snack brand running a 20%-off TPR with a feature ad at a club channel store might see 300-400% lift. The same brand running a 10%-off shelf price reduction at a conventional grocery chain with no feature might see 40-80% lift. These are not interchangeable. Comparing lift across programs without controlling for depth and merchandising is one of the most common mistakes in trade planning.

Incremental efficiency

Raw lift % rewards large displays and deep discounts almost automatically. A more useful metric is incremental units per dollar of trade spend, or incremental units per point of ACV on promotion. These normalize for how much you invested to generate the lift.

In the table above, Retailer C has the lowest absolute lift but the best incremental efficiency. Retailer A generates the most incremental units but at a higher cost. A brand with a constrained trade budget would allocate more to programs like C and less to programs like A, a conclusion hidden when you look only at lift %.

Cannibalization and Pantry Loading

Even when the incremental calculation is clean, two forces can erode the real value of that lift: cannibalization and pantry loading. Both shrink true incrementality below what the promoted-sales decomposition shows.

Cannibalization

Cannibalization occurs when a promotion on one SKU steals volume from a sister SKU in the same brand portfolio. A consumer who would have bought the 18-count regular variety buys the 18-count promotional variety instead. The brand's total shipments did not grow; volume just shifted between items. If you measure incrementality only at the promoted item level, the lift looks real. At the portfolio level, it is partly or entirely offset.

Cannibalization rates differ by how closely items substitute for each other. Flavor variants within the same pack size cannibalize heavily. A premium-tier item and a value-tier item in the same brand family cannibalize less. A Guide to Trade Promotions Effectiveness Analysis covers how to measure cross-item effects when evaluating a full portfolio event.

Pantry loading

Pantry loading (sometimes called pull-forward) happens when loyal buyers who would have purchased next week or next month stock up during the promoted week because the price is attractive. Total category consumption does not increase; the purchase simply shifts earlier in time. The promoted week shows strong lift, but the two to four weeks after the event show below-baseline velocity as those stocked-up consumers work through inventory. This is the classic post-event trough visible in a 52-week weekly scan chart.

Pantry loading is more severe for non-perishable, high-velocity items with consumers who are already loyal (they have storage capacity and will use the product eventually). It is lower for fresh, short-shelf-life items or for new users acquired through the promotion.

True incrementality accounts for the full event window including the post-event trough. A promotion that shows 180% lift in week 1 but a 60% below-baseline drag for three weeks after has much lower net incremental than the event-week lift suggests.

Why Incremental Volume Drives Promo ROI

Promo ROI is a function of incremental margin, not total promoted margin. The math is straightforward: base volume would have sold at full price regardless. Trade funding applied to base units is a pure margin give-up with no offsetting revenue benefit. Only the incremental units represent additional revenue the brand would not have captured without the promotion.

ROI = (Incremental units x gross margin per unit) / Total trade spend

A program with strong total volume but weak incrementality can still post negative ROI because the trade spend is subsidizing base. How to Tell If a CPG Promotion Actually Worked walks through the full ROI framework in detail.

This is also why "promoted volume" as a success metric is misleading. A brand that doubles a retailer's display space and halves the price will ship a lot of cases. The question is how many of those cases represent net-new demand versus base units sold at a deep discount to consumers who would have bought them anyway.

Understanding what trade promotion is trying to accomplish at each stage of a brand's lifecycle helps set realistic incrementality targets. A new item in year one needs true trial: high lift is expected and the baseline is thin. A mature item at 80% ACV in its category is much harder to move incrementally because loyal users are already buying regularly.

A Worked Example

Suppose a mid-size snack brand runs a two-week TPR event at a regional grocery chain covering 320 stores. Here are the inputs:

• ACV on promotion: 68% (the event runs in 68% of the chain's ACV-weighted stores)
• Total promoted units shipped during the two-week window: 94,000
• Modeled baseline for the same two-week period: 31,000 units
• Price discount: 22% off regular shelf price
• Merchandising: feature ad, no display

Step 1: Calculate incremental volume.

Incremental = 94,000 - 31,000 = 63,000 units

Step 2: Calculate lift %.

Lift = 63,000 / 31,000 x 100 = 203%. The event approximately tripled baseline velocity.

Step 3: Adjust for cannibalization. The brand has a sister flavor SKU that dropped 8,400 units below its own baseline during the same two weeks. Net portfolio-level incremental = 63,000 - 8,400 = 54,600 units.

Step 4: Adjust for pantry loading. The three weeks following the event run 4,200 units below baseline per week on average. Post-event drag = 12,600 units. Net true incremental after pantry loading and cannibalization = 54,600 - 12,600 = 42,000 units.

Step 5: Calculate ROI. Assume gross margin per unit is $0.85 and total trade spend for the event (TPR funding + feature allowance) was $31,500.

ROI = (42,000 x $0.85) / $31,500 = $35,700 / $31,500 = 1.13x. The brand returned $1.13 for every dollar of trade spend, a modest but positive result.

The gap between gross incremental (63,000) and net true incremental (42,000) is 33%. That gap represents real margin erosion. If the brand had used the gross figure to calculate ROI, it would have reported $1.70 per dollar, a significant overstatement. Promo post-mortems that skip cannibalization and pantry-load adjustments will consistently overstate program performance and lead to over-investment in repeating the same events.

Putting It Into Practice

A few practical habits make this analysis tractable at scale:

• Track the full 8-10 week window around each event, rather than only the promoted weeks. Post-event trough behavior is visible only in weekly time-series data.
• Pull promoted-sales decomposition at the item level, rather than only at the brand or category level, so cannibalization within the portfolio is visible.
• Compare baselines across syndicated providers when you first set up a measurement framework. Diverging baselines on the same item usually point to how each provider weights panel versus POS data.
• Set incrementality benchmarks by retailer format. Club is structurally different from conventional grocery; mass merchandiser lift profiles differ from natural/specialty.
• Feed the net true incremental figure back into annual trade planning. Programs that looked good on gross lift but underperform on net incrementality should be resized or redesigned before the next plan cycle.

Harmonized syndicated data makes the 8-10 week view significantly faster to build. When SPINS, Circana, and retailer portal POS data are normalized to a common item and week grain, analysts can run the full incremental analysis in a single environment rather than stitching together spreadsheets from three separate exports.