Why a good group ROAS can hide struggling locations

Because an average can't tell you where the money is working. A group-wide ROAS puts every location's spend and every location's revenue into one ratio, so a strong restaurant can cover for a weak one and the number still looks healthy.

Picture the monthly marketing review. The marketing director puts up one slide: return on ad spend (ROAS) for the group, 3.0x, right on target. The owner nods and the meeting moves on to the holiday calendar. Nobody asks about the two restaurants underneath that number, because the slide gives no one a reason to.

That is a reasonable outcome. A group does need one answer to whether the budget is paying for itself. The trouble is that the same answer can't tell the marketing director what to do next.

What can a healthy group ROAS be hiding?

Here is a made-up pair. Say a group spends $5,000 a month on each of two restaurants, $10,000 in all, and the revenue tied to paid checks comes to $30,000. That is 3.0x. Underneath it:

• Downtown: $5,000 in, $24,000 back, 4.8x. It looks like the star. But it is close to full on Fridays and Saturdays anyway, and most of its results land on those nights, so a good share of the spend may be reaching guests who would have come regardless.
• Riverside: $5,000 in, $6,000 back, 1.2x. The campaign that works downtown barely registers here, and the average has been quietly covering for it.

Neither restaurant deserves "hold steady." Downtown's number needs a question put to it: what would those Fridays have looked like without the spend? Riverside needs a different campaign. These are two different problems, and a group average encourages one response to both.

Compare how the review sounds if the slide shows both. The owner hears something different: "Overall we're on target, but most of the return comes from one restaurant, and we're not sure how much of that we caused. The other needs a different campaign." That is a conversation ownership can act on. The same thing happens one level up, too. A brand that looks strong across several locations can be hiding a campaign that only works near one of them.

Why not just rank the locations against each other?

Because they aren't the same business. Capacity, opening hours, local demand, average check and how many guests leave a trail a report can follow all differ from one location to the next, so a ranking by revenue mostly measures size. The question is not "Which restaurant sold the most?" It is "Where is marketing helping, where is it underperforming, and what else might explain the difference?"

A fairer comparison sets each location against its own usual week. It also checks that the revenue behind each location comes from the register and not from the platforms' reports, which is the gap we wrote about in why Google and Meta ROAS don't match your register.

What should the group report show instead?

It should keep the levels connected. Start with the group view for total spend and results, then move down through brand, location, campaign and service period, with the same definitions at every level. Otherwise the group report is only a stack of different reports from the platforms, dressed in the same format.

Ownership needs to see the overall return. The people running each brand need to see what changed locally. Both should be working from the same numbers.

A monthly review built that way tends to produce three different actions instead of one:

• Keep a campaign running where check-backed results are strong and the gain looks real.
• Investigate a location where engagement is healthy but sales aren't following.
• Improve measurement where guests are plainly coming in but too few visits connect to checks.

The restauWant app's portfolio view is built this way: the group, each brand and each location in one place, reading from the same checks, with a general manager seeing their own restaurant and ownership seeing all of them.

You can start without it. Take last month's group ROAS and split it by location: spend and check-backed revenue for each. If you can't split it, that is your first finding. If you can, ask each location one question: would we repeat this spend next month?

Next step: Read One campaign, eight locations: how to find out where it actually worked for the step-by-step version of reading a campaign one location at a time.