iROI vs. ROAS vs. MER: Which Number Should Run Your Budget?

Every retail media team runs on a number. For most, it's ROAS. For a growing crowd, especially teams influenced by DTC thinking, it's MER, the marketing efficiency ratio. And for a smaller group that includes some of the largest CPG advertisers in the world, it's iROI.

These get treated as interchangeable flavors of the same idea. They aren't. Each one answers a different question, and only one of them answers the question a budget decision actually asks.

ROAS: what did my ads touch?

Return on ad spend divides attributed revenue by ad spend. A campaign spends $100K, the platform attributes $500K in sales, ROAS is 5.0.

The strength is granularity and speed. You get a number per campaign, per keyword, per day, and you can act on it immediately. The weakness is the word "attributed." ROAS counts every sale an ad touches, including the enormous share that would have happened anyway: loyal repeat buyers, shoppers already searching your brand name, people three steps from checkout when the ad appeared. Branded search campaigns post spectacular ROAS for exactly this reason. They intercept demand that already existed.

ROAS answers "what did my ads touch?" It cannot answer "what did my ads cause?", and optimizing to it systematically shifts budget toward campaigns that are good at claiming credit rather than creating sales.

MER: is marketing efficient overall?

The marketing efficiency ratio divides total revenue by total marketing spend. No attribution at all. If the business did $10M and marketing spent $1M, MER is 10.

People reach for MER precisely because they've stopped trusting attribution, and the instinct is sound. MER can't be gamed by attribution windows or last-click logic, and as a monthly health check on whether marketing costs are scaling sanely against revenue, it does its job.

But notice what it gave up to get that honesty: all resolution. MER is one number for the entire business. It moves when pricing moves, when distribution changes, when a competitor stumbles, when the category grows. It cannot tell you whether Amazon DSP is outperforming Walmart Connect, whether campaign 47 deserves more budget than campaign 12, or whether marketing caused the revenue at all. Total revenue over total spend includes every sale you would have made with the marketing turned off.

MER answers "is the overall ratio healthy?" It is a thermometer. You cannot steer with a thermometer.

iROI: what did my marketing cause, and where should the next dollar go?

Incremental return on investment starts from total sales and decomposes them: the organic baseline you'd have sold anyway, the effects of pricing, promotion, and the digital shelf, and finally the residual lift that advertising actually caused. That causal lift, measured against spend, is iROI. (See more about our methodology here.)

It keeps what's good about the other two and fixes what's broken. Like MER, it's anchored to total real sales rather than platform-attributed ones, so it can't be inflated by aggressive attribution. Like ROAS, it resolves down to the campaign level, so you can act on it. And because it's causal, it supports the one thing neither of the others can: response curves showing where additional spend still buys additional sales and where a campaign has flattened out.

The catch is that iROI is harder to produce. It needs sales data, media data, and merchandising data unified across retailers, and a model doing real work on top. That difficulty is why most teams settle for ROAS. It's also why the teams that don't settle find so much money on the table: brands optimizing to iROI typically pull 15-20% more sales from the same budget, because for the first time the budget is flowing toward campaigns that create demand instead of campaigns that collect credit. Bayer targeted a 10% media efficiency gain on Amazon, kept spend flat, and hit 32%. Church & Dwight saw iROI improve up to 122% across TheraBreath, Batiste, and Arm & Hammer Laundry, again on a fixed budget.

So which number runs the budget?

Keep all three, but give them different jobs.

MER is the board-level sanity check. Watch it monthly; investigate when it drifts. ROAS remains useful as a fast operational signal within a single platform, as long as everyone in the room understands it measures credit, not cause. iROI is the allocation metric, the one that decides where dollars move between campaigns, channels, and retailers, because it is the only one of the three built from causation.

A useful test for any metric you're about to optimize toward: if we had run zero ads, how much of this number survives? For MER, most of it. For ROAS, an uncomfortable amount you can't isolate. For iROI, none, by construction.

The number that survives nothing is the one that should run the budget.