ROAS for in-store media: calculating it honestly

ROAS, or return on ad spend, is attributed sales divided by what the advertising cost. Retail media ROAS applies that formula to campaigns on retail networks, including in-store screens. The arithmetic is simple. The hard part is the numerator: which sales count as attributed, and how you know the campaign caused them. Two honest analysts can get very different ROAS figures from the same campaign depending on that choice.

A ROAS figure is only as trustworthy as the method behind it.

How is ROAS calculated?

Divide the sales credited to the campaign by the campaign's cost. If a campaign cost $10,000 and $30,000 in sales are credited to it, ROAS is 3, often written as 3 to 1 or $3 per $1 spent.

That's the formula. Everything interesting happens in deciding which $30,000 to count.

One common approach counts all sales of the advertised products in the advertised stores during the flight. That's easy to calculate and almost always too generous, because many of those sales would have happened without the campaign.

What's the difference between attributed and incremental?

Attributed sales are the sales a method assigns to the campaign. Incremental sales are the sales the campaign actually caused, meaning sales that wouldn't have happened otherwise.

Here's a hypothetical to show the gap. A campaign costs $10,000 and runs in 200 stores. Those stores sell $50,000 of the advertised product during the flight. Counting all of it gives a ROAS of 5.

Now add a comparison. Two hundred similar stores that didn't run the campaign sold $44,000 of the same product over the same weeks. The difference, $6,000, is a better estimate of what the campaign added. Incremental ROAS is 0.6.

Same campaign, same data, ROAS of 5 or 0.6 depending on the method. Neither number is fabricated. Only one describes what the advertising did.

What can't a ROAS figure tell you?

Profit. ROAS uses sales, not margin. A ROAS of 2 might be profitable for a high-margin product and a loss for a low-margin one. Pair ROAS with margin before judging a campaign.

Long-term effects. Most ROAS calculations cover the flight and maybe a few weeks after. Effects on brand recognition, or repeat purchases months later, fall outside the window.

Precision. An incremental estimate comes with uncertainty. Store sales vary week to week, and the difference between test and comparison groups can move just by chance. A ROAS stated as a single number hides that. A range is more honest.

Comparability. ROAS figures from different networks, calculated by different methods, can't be compared directly. A network that counts all sales in advertised stores will always look better than one that measures against controls.

Broader approaches such as marketing mix modeling estimate contributions across many channels over longer periods, and they can complement campaign-level ROAS.

How should ROAS be reported honestly?

State the method. Say exactly which sales were counted, over which period, in which stores, compared with what.

Prefer incremental measures. Use comparison stores that were similar before the campaign. Store-level scan data, like the data analyzed by NRS Insights, is the kind of source that can support that comparison.

Show a range. If the incremental estimate is uncertain, say so, and give the range.

Include margin, or at least note it. A sales-based ROAS with no margin context invites misreading.

And don't compare across methods. If you're weighing networks, ask each to calculate ROAS the same way, or calculate it yourself from comparable data. To talk through how a campaign could be measured, contact the NRS Digital Media team, or see the NRS Digital Media site.