Measuring promotion effectiveness: did the deal work?

Did that two-week price cut sell more product, or did it just sell the same product cheaper? Measuring promotion effectiveness comes down to comparing sales during the deal against a baseline, an estimate of what would have sold anyway. The comparison is simple arithmetic. Building an honest baseline is the actual work.

Promotion lift is the standard name for the result: sales during the promotion minus the baseline. Computed carelessly, lift flatters almost every deal. Computed carefully, it tells you which promotions earn their cost and which ones quietly hand margin away.

What counts as a clean baseline?

The baseline is your estimate of normal. The usual starting point is an average of recent weeks with no promotion running, for the same item in the same set of stores.

Three things contaminate baselines. Other promotions: if the pre-period included a different deal, normal is overstated. Seasonality: baselining a summer cold-drink deal against spring weeks builds growth into the estimate that the deal didn't cause. Store churn: if the stores in the pre-period aren't the stores in the promo period, you're comparing populations, not weeks. Hold the store set constant.

There's a stronger design available when enough stores are in the data: a control group. Compare the promoted stores against similar stores where the deal never ran, over the same weeks. The control group absorbs whatever else was happening in the period, weather, paydays, a slow stretch on the block, and leaves the promotion's own effect standing out more cleanly.

Measure units and dollars separately

Lift in units tells you the deal moved product. Lift in dollars tells you it paid. The two can disagree.

Say an item normally sells 100 units a week at $2.00, and during a promotion at $1.50 it sells 180. Units are up 80 a week, and dollars went from $200 to $270. Now suppose the discount had gone to $1.00 and sold 150 units: units still up, but dollars down to $150. Same happy unit story, opposite revenue story. Run both columns every time.

Did the deal steal from the weeks around it?

The two classic leaks are pantry loading and cannibalization. Pantry loading is shoppers stocking up at the deal price, which borrows sales from the weeks after. Watch for a dip below baseline in the two to four weeks following the promotion, and subtract it from the lift.

Cannibalization is the deal pulling volume from the brand's own neighbors, another flavor or another size. Check the sibling items during the promo window before crediting the deal with new demand. Lift that survives both checks is net lift, and net lift is the number to weigh against the promotion's cost.

A five-step promotion read

1. Build the baseline from clean, recent, promotion-free weeks, with the store set held constant.
2. Measure promo-period sales against it, in units and in dollars.
3. Check the following weeks for a post-promotion dip and subtract it.
4. Check sibling items for cannibalization and subtract that too.
5. Compare what's left against what the deal cost, and write the verdict down where the next planner will find it.

In independent retail there's one extra wrinkle: execution varies store to store, and a deal that was funded isn't always a deal that reached the shelf. Scan data settles that question, because it records the price shoppers actually paid. Channel-level context, like the NRS Insights monthly same-store sales report, also helps you tell a promotion's effect apart from a month when the whole channel moved.