Same-store sales lift: reading a campaign's effect

Suppose a beverage brand runs six weeks of screen advertising in neighborhood stores, and sales of its drinks rise in those stores during the flight. Encouraging, but not yet an answer. Same-store sales lift is the discipline that turns that observation into a read: compare the same set of stores against their own baseline, so the change you measure comes from changed behavior, not a changed store list.

The concept is simple. The care is in the reading.

What does "same-store" actually control for?

Composition. Any store network changes over time as locations join and leave. If you compare this month's total sales across all stores against last year's total, part of the difference is just the roster. Same-store analysis holds the set of stores fixed, so growth means the same doors sold more, not that there are more doors.

This is why same-store sales is a standard yardstick in retail. NRS Insights publishes a monthly same-store sales report built from point-of-sale scan data across the NRS network for exactly this reason: it isolates real trading changes from network growth.

What baseline should lift be measured against?

A lift number is always relative to an expectation, and you have three honest options. The pre-period baseline compares the flight against the weeks before it. The year-ago baseline compares against the same weeks last year, which absorbs seasonal rhythm. The control baseline compares advertised stores against similar stores that didn't run the campaign during the same weeks.

The control baseline is the strongest, because it experiences the same season, weather, and news cycle as your test stores. Whatever moved everyone shows up in both groups and cancels out, leaving the campaign as the main difference. A year-ago read is a good supplement; a pre-period read alone is the weakest of the three. Whichever you choose, write the choice down before launch, so the comparison can't drift toward whichever baseline flatters the result afterward.

What else can move the number?

Plenty, and this is where campaigns get over-credited. Seasonality is the big one: many categories swing hard across the calendar, which is visible even in aggregate measures like the Census Bureau's national retail sales figures. A price change, a new pack size, a distribution gain, a competitor's out-of-stock, or a trade promotion running in the same weeks can each move sales on their own.

You can't freeze the world during a flight, but you can log it. A simple record of what else changed, kept during the campaign, is worth more than any amount of after-the-fact reconstruction.

Suppose the beverage brand from the opening also gained a second shelf facing in half its advertised stores mid-flight. Its lift read now blends media with merchandising, and only that log lets anyone say how much of the rise belongs to which.

How do you read lift responsibly?

Decide the comparison before launch, not after the results arrive. Read the advertised SKUs specifically, not just the category, since a category can drift for reasons that have nothing to do with you. Check whether the pattern holds across regions and store types rather than leaning on one standout cluster. And treat a single flight as one data point. A result that repeats across two or three flights is a finding; a result that appears once is a lead worth testing again.

Because campaigns on the NRS Digital Media network activate by geography down to zip-code level and by SKU, structuring a clean test-and-baseline read is a planning choice rather than a measurement afterthought.