Promotional Baseline: The Counterfactual That Drives Every ROI Calculation

Baseline volume is the volume you would have sold in a given period if no promotion had run. It is a counterfactual, a version of events that never happened, and you have to reconstruct it from data and assumptions. If you cannot defend the baseline, you cannot defend any number that sits on top of it.

The small difference between two large numbers

Every trade promotion calculation depends on the baseline. Incremental volume, the extra volume the promotion actually earned, is what you sold minus what the baseline says you would have sold anyway. Because it is the small gap between two large numbers, a small error in the baseline turns into a big error in the incremental figure. A 5,000‑unit baseline against 8,000 actual units gives 3,000 incremental. Raise the baseline to 5,500 and incremental volume falls to 2,500. A 10% change in the baseline has changed the incremental figure, the extra volume the promotion gets judged on, by 17%. When the uplift is a third of the baseline or less, so a promoted week sells about 1.33 times a normal week or less, that same 10% error distorts incremental by 30% or more.

Why the number turns political

Baseline estimation is the most debated topic in trade promotion analytics, because so much money rides on it. Sales teams favor low baselines, which make their promotions look more incremental. Finance wants accurate baselines, because the true return depends on them. This course plans on manufacturer and retailer baselines sitting 15 to 25% apart, a starting assumption rather than a measured finding. A gap that size is enough for the two sides to disagree on whether a promotion worked.

Three methods dominate baseline estimation, and they sit on a ladder of sophistication. Each one produces a different number from the same sales history, which is exactly why the choice matters.

Moving average (the simplest)

Transparent and easy to defend in a joint review. It ignores trend and seasonality, so it fits flat, non‑seasonal categories best.

Adjusted moving average

• Seasonality index: what this period of the year normally sells, divided by what an average period sells. Above 1.0 and the period runs hot on its own, below 1.0 and it runs cold.
• Trend factor: the brand's recent growth rate, applied to the weeks ahead so the baseline rises or falls along with the business instead of staying flat.

Regression‑based

The most accurate and the hardest to explain. It isolates the promotional effect from everything else moving at the same time.

Across every method, the baseline is always an estimate, never a fact. The method you pick is a strategic decision, not a technical one.

A juice brand sells 4,000 units per week on average. During a 2‑week promotion, actual sales jump to 7,500 units per week. Run the same actual sales through three baseline methods and you get three different stories.

Scenario: one promotion, three baselines

The incremental math

• Simple moving average, the average of recent weeks with no promotion in them, puts the baseline at 4,000/wk. The promotion then looks like 3,500 extra units a week, 7,000 across the two weeks. A strong promotion.
• Seasonality‑adjusted, the same average scaled up because summer is peak juice season and some of that volume was coming anyway, puts the baseline at 4,600/wk. The promotion now adds 2,900 a week, 5,800 in total. A good promotion.
• Trend‑adjusted, which adds the category's 4% a year of growth on top of the season, puts the baseline at 4,800/wk. The promotion adds 2,700 a week, 5,400 in total. A moderate promotion.

Same actual sales, three methods, incremental estimates spread across 30%, and three ROI calculations that would tell the board three different things. This is why the method has to be agreed before the result is in dispute.

Baseline arguments get political because so much money depends on the number. The return you report on trade spend, your claim to be the retailer's category captain (the one supplier the buyer lets advise on the whole category), and next year's promotional calendar all follow from it. A few working rules keep the fight productive.

Expect each side to pull the baseline its way

A manufacturer arguing for a lower baseline shows more incremental volume and better ROI, which justifies continued trade spend. A retailer arguing for a higher baseline claims a smaller uplift, which pressures the manufacturer to discount deeper or invest more. Neither side is lying. They are reading the same data through different incentives.

Agree on the method once, in joint business planning

Experienced practitioners lock an agreed baseline method with key retail partners up front. The specific method matters less than both parties using the same one every time. The companies that sell shared retail sales data, the ones you and the retailer both already subscribe to, publish their own baseline estimates, and a number neither side produced makes a useful neutral reference. Their models carry their own assumptions, so treat those baselines as a third opinion rather than the truth. Two quick checks catch a bad one before you lean on it. First, look at your promoted weeks: a bought baseline should not jump when the deal runs, and if it does, the model has counted part of the promotion as baseline and will understate every uplift. Second, look at your quiet weeks: the baseline should land close to what you actually sold, sometimes a little above and sometimes a little below. If it sits above your actual sales week after week, or below, it is biased, and every promotion you measure against it is tilted the same way.

Consistent beats perfect

Use one baseline methodology across every promotion so the results compare. An imperfect but consistent baseline drives better decisions than a theoretically perfect one that shifts with each analysis.