You ran a 30% off promotion on your hero SKU last month. Volume spiked 4x during the promo window. Your sales team called it a win. Then the next three weeks happened — orders cratered below baseline, your 3PL scrambled to rebalance inventory, and your contribution margin for the quarter came in worse than if you’d done nothing.

For most CPG brands running $2M+ in annual trade spend, this is a routine occurrence.

Promotions do work. The trouble is that most brands can’t tell the difference between demand they created and demand they borrowed from next week. Without that distinction, every promo decision is a coin flip with six-figure stakes.

Baseline Demand vs. Promotional Lift

Every SKU has a baseline — the volume it moves without any promotional activity. Everything above that baseline during a promo window is your gross lift. But gross lift is a vanity metric. What you need is incremental lift: the volume that wouldn’t have happened at all without the promotion.

The gap between gross lift and incremental lift is where margin goes to die.

Three forces eat your gross lift:

Force	What Happens	Typical Impact
Pantry loading	Consumers buy more now, buy less later	20–40% of gross lift
Forward buying	Retailers stock up at promo price, reduce future orders	15–30% of gross lift
Cannibalization	Promo SKU steals volume from your other SKUs	5–15% of gross lift

On a typical CPG promotion, 40–60% of gross lift is time-shifted demand wearing a costume rather than anything incremental.

Building a Baseline Model

You can’t measure lift without a credible baseline. Here’s a practical approach that doesn’t require a data science team.

Step 1: Collect 52 weeks of weekly sell-through data per SKU per channel. If you don’t have 52 weeks, use what you have — but flag anything under 26 weeks as low-confidence.

Step 2: Strip out promo weeks. Tag every week where a promotion was active and remove those data points. What remains is your non-promotional demand signal.

Step 3: Fit a trend + seasonality model on the remaining weeks.

Baseline(week) = Trend(week) × Seasonal_Index(week)

Where:
  Trend(week) = Average_Weekly_Units × (1 + Growth_Rate)^week
  Seasonal_Index(week) = Avg_Units_in_Same_Period / Avg_Units_Overall

Example for a skincare SKU:
  Average weekly units (non-promo): 820
  Annual growth rate: 12%
  Q4 seasonal index: 1.35 (holiday bump)
  
  Baseline for Week 48: 820 × (1.12)^(48/52) × 1.35 = 1,188 units

Step 4: Project that baseline forward through your promo weeks to see what would have sold without the promotion. The difference between actual sales and projected baseline is your gross lift.

The Post-Promo Dip

Here’s where most analyses stop too early. Gross lift during the promo window tells you half the story. The other half lives in the 2–6 weeks immediately after.

Consumer packaged goods have a consumption rate. Your customer doesn’t use more shampoo because they bought three bottles at a discount — they just don’t buy shampoo again for three months. That pantry-loaded volume comes directly out of future periods.

To measure the real cost:

1. Calculate your baseline for the 4 weeks following the promo
2. Compare actual post-promo sales to that baseline
3. Sum the shortfall — that’s your demand borrowed

Promo Period (2 weeks):
  Baseline forecast:    1,600 units
  Actual sales:         6,400 units
  Gross lift:           4,800 units

Post-Promo Period (4 weeks):
  Baseline forecast:    3,200 units
  Actual sales:         1,920 units
  Demand borrowed:      1,280 units

True Incremental Lift: 4,800 - 1,280 = 3,520 units
Borrow Rate: 1,280 / 4,800 = 26.7%

A 27% borrow rate is decent for CPG. Brands with high pantry-loadable products (paper goods, cleaning supplies, shelf-stable food) regularly see borrow rates above 50%. If your borrow rate consistently exceeds your margin on the incremental volume, you’re paying customers to shift their purchase timing.

The Promo P&L Nobody Runs

Most brands evaluate promotions on revenue during the promo window. That’s like evaluating a loan by looking only at the disbursement. Here’s the full accounting:

Line Item	Calculation	Example
Gross lift revenue	Lift units × Promo price	4,800 × $18.90 = $90,720
Less: Revenue at full price	Baseline units × Full price	1,600 × $27.00 = ($43,200)
Less: Post-promo dip revenue loss	Borrowed units × Full price	1,280 × $27.00 = ($34,560)
Less: Cannibalized SKU margin	Cannibalized units × Margin	240 × $11.50 = ($2,760)
Less: Promo funding / trade spend	Retailer allowance + slotting	($12,000)
Less: Incremental fulfillment cost	Surge picking, overtime, expedited freight	($4,200)
Net promo contribution		($6,000)

That 4x volume spike just cost you six thousand dollars. And this doesn’t include the brand equity erosion of training your customers to wait for sales.

Five Promo Archetypes and When Each Works

Not all promotions borrow equally. The structure of the offer determines how much demand shifts forward versus how much is genuinely new.

1. Percentage-Off (30% off)

High borrow rate. Existing customers stock up. Works for trial on new SKUs where you’re buying first purchases, not accelerating repeat purchases. Terrible for hero SKUs with established purchase cadence.

2. BOGO / Multi-Buy (Buy 2 Get 1)

Highest pantry loading of any format. The average consumer overestimates their consumption rate by 30–40%. Use only when you need to clear short-dated inventory or when the product has genuine elastic consumption (snacks, beverages).

3. Bundling (Hero + New SKU)

Low borrow rate when structured correctly. The hero product sells at or near full price, and the trial product gets the discount. The incremental value is the trial conversion, not the hero volume.

4. Retailer-Funded Display (Endcap, off-shelf)

Moderate borrow rate, but the mix shifts. You get more new-to-brand buyers from impulse purchase. Measuring this requires tracking household penetration, not just unit velocity.

5. Loyalty / Rebate (Cashback via app)

Lowest borrow rate because the friction filters out pantry loaders. The people who clip digital coupons and submit rebates tend to be genuine deal-seekers making incremental purchases. Higher administrative cost, better incremental economics.

Building Your Promotional Calendar Around Incrementality

Once you can measure borrow rates by promo type, you can build a calendar that maximizes true incremental volume instead of chasing gross lift vanity metrics.

Rule 1: Space promos at least 2x your category’s purchase cycle apart

If your average customer buys every 6 weeks, don’t promote more frequently than every 12 weeks on the same SKU. Closer spacing means you’re subsidizing purchases that would have happened anyway.

Rule 2: Rotate SKUs instead of repeating hero promos

Your hero SKU already has the highest baseline velocity. Promoting it has the highest borrow rate because existing buyers are the most likely to load up. Promote trial-stage or growth-stage SKUs where the incremental buyer acquisition has long-term LTV.

Rule 3: Require a post-promo forecast, not just a promo forecast

Before approving any promotion, demand planning should model both the lift and the dip. Your production plan, warehouse labor schedule, and cash flow forecast should account for the 3–5 week valley that follows every spike.

Promo approval checklist:

□ Baseline model for promo SKU (last 26+ weeks)
□ Estimated gross lift (units)
□ Estimated borrow rate (% based on promo type history)
□ True incremental units = Gross lift × (1 - Borrow rate)
□ Incremental contribution margin
□ Trade spend / promo funding
□ Net promo P&L (must be positive)
□ Post-promo demand dip modeled in supply plan
□ 3PL / warehouse notified of volume swing

Rule 4: Kill promos that fail the incrementality test twice

If a promo archetype on a specific SKU shows a borrow rate above 50% in two consecutive runs, it’s a margin leak with good packaging. Move the trade dollars to a different SKU, channel, or tactic.

Measuring Incrementality Without a Control Group

The gold standard for measuring promotional lift is a matched-market test: run the promo in some stores and hold others as a control. But most brands under $100M don’t have the retailer relationships or data access to run clean holdout tests.

Here are three practical alternatives:

Time-series comparison. Compare the promo period + post-promo period to the equivalent period in the prior year, adjusted for trend and seasonality. This is the baseline model approach described above. It’s imperfect — external factors like weather, competitive activity, and distribution changes create noise — but it’s directionally reliable over multiple promo cycles.

Cross-channel comparison. If you promote in one channel (say, Amazon) but not another (DTC), compare the lift in the promo channel against the movement in the non-promo channel during the same period. If your DTC channel shows a dip while Amazon spikes, some of that “lift” is channel switching, not incremental demand.

Cohort analysis. For DTC brands with customer-level data, compare the purchase behavior of customers who bought during the promo to their pre-promo and post-promo purchase frequency. Did they buy earlier than their typical cadence? Did their next purchase take longer than usual? This directly quantifies pantry loading at the individual level.

What Good Looks Like

After tracking promotional incrementality for 4–6 quarters, high-performing CPG brands converge on a few benchmarks:

Metric	Poor	Average	Strong
Average borrow rate	>50%	30–50%	<30%
Promos with positive net P&L	<40%	50–65%	>75%
Trade spend as % of gross revenue	>20%	12–18%	8–12%
Incremental buyer % of promo volume	<15%	20–35%	>40%
Post-promo dip duration	>5 weeks	3–4 weeks	1–2 weeks

The brands in the “strong” column spend just as much on promotions. They spend it on different promotions — ones that acquire new buyers instead of subsidizing existing ones, ones that drive trial on growth SKUs instead of spiking volume on products that already sell.

When Promotions Genuinely Create Demand

To be fair: some promotions do create real, lasting demand. They share a few characteristics.

First, the product has a low household penetration in the channel. If only 8% of the retailer’s shoppers have tried your product, a well-placed endcap with a modest discount can generate trial that converts to repeat purchase. That first-time buyer has genuine LTV. But if your household penetration is already 40%+, the same endcap mostly accelerates existing buyers.

Second, the product has elastic consumption. Snack brands, beverages, and impulse-purchase categories see genuine consumption lift from promotions — people actually eat more chips when they have more chips in the house. Products with fixed consumption rates (laundry detergent, vitamins, toothpaste) almost never see true demand creation from promotions.

Third, the promotion introduces a new format, size, or variant. A promo that gets your new 16oz size onto shelves and into trial baskets is fundamentally different from a promo that discounts the 12oz size everyone already buys. The incremental value is distribution and awareness, not price-driven volume.

Connecting Promo Forecasting to Supply Planning

The operational cost of promotional demand volatility goes beyond the promo P&L. Every volume spike creates a ripple through your supply chain:

Production planning needs 8–12 weeks of lead time to build promo inventory. If your demand planning team can’t quantify the expected lift with reasonable accuracy, you either overbuild (carrying cost, expiration risk) or underbuild (stockouts during the promo, retailer chargebacks, damaged relationships).

Warehouse labor scheduling depends on predictable volume. A 4x spike in outbound units means temporary labor, overtime, or missed SLAs. Your 3PL’s surge pricing is typically 15–25% above standard rates. If you’re not modeling that into your promo P&L, you’re understating the true cost.

Cash flow timing shifts when promos pull demand forward. You spend the trade dollars and production costs upfront, get the revenue spike during the promo, then watch the post-promo dip eat into the following period’s cash position. Brands with seasonal concentration and high trade spend regularly create artificial cash crunches by stacking promotions without modeling the downstream effects.

The fix is to treat the promo forecast with the same rigor you apply to your baseline forecast, not to stop promoting. That means modeling the full demand curve — ramp, peak, dip, and recovery — and flowing it through your production schedule, labor plan, and cash flow model.

Where to Start

If you’re running trade promotions without an incrementality framework, don’t try to boil the ocean. Pick your top three SKUs by trade spend and run the baseline + lift + dip analysis on their last four promotions. You’ll have enough data within a week to know whether your promo calendar is building your brand or borrowing from it.

The brands that get this right promote differently rather than less. They shift spend from hero-SKU discounts to new-buyer acquisition. They replace blanket percentage-off with targeted formats that filter out pantry loaders. And they model the full demand curve before approving the spend, not after the quarter closes.

If your demand planning and trade promotion teams are operating independently, that gap is where the margin disappears. CommerceOS connects promotional calendars to demand forecasts so your supply plan reflects the whole picture — the spike, the dip, and the real incrementality underneath.