How to Measure Incrementality in aMulti-Retailer Media Mix

‍

Here's a fun math problem. Add up the attributed sales from every retail media network you run on. Now compare that to how much your brand actually grew.

For a lot of brands, the networks collectively claim more credit than the business actually earned. Sometimes way more.

That's not fraud. It's how attribution works. Each network measures its own campaigns, with its own rules, inside its own walls, and each one gets to count the shopper who saw ads in three places. Nobody's lying. Everybody's just counting the same sale.

If you're running media across multiple retailers, the question that matters isn't "what did each network report?" It's "what did all of this actually add?" That's incrementality.

‍

Why multi-retailer makes it harder

‍

Measuring incrementality on one network is tricky enough. Across several, a few extra problems pile up.

Different methodologies. Each network defines attribution windows, view-through credit and new-to-brand differently.

Overlapping audiences. The same shopper can see your ads on a grocer's site, a delivery app and a marketplace in the same week.

Cross-retailer spillover. An ad on one retailer's network can drive a purchase at a different retailer, which the first network can't see at all.

Off-site media. Many networks now sell placements across social, streaming and the open web, which blurs the lines even more.

So you need a measurement approach that sits above any single network.

‍

The incrementality toolkit

‍

There's no one perfect method. Most brands use a combination.

Holdout tests within each network. Randomized test-versus-control studies inside a single network tell you the real lift of that network's campaigns on its own sales. Great for calibrating, but they can miss spillover to other retailers.

Geo-based tests. Turn media up or down in some markets and compare total brand sales against similar markets. Because you measure total sales, not just one retailer's, you capture spillover. Takes planning and enough comparable markets.

Media mix modeling (MMM). Statistical models that estimate how each channel contributes to total sales over time, while controlling for price, distribution, seasonality and other factors. MMM gives the big-picture view across retailers and channels. It needs a solid history of data and works best for strategic budget decisions, not weekly optimization.

Clean room analysis. Some retailers allow brands to match exposure and purchase data in privacy-safe environments. Useful for deeper cuts, but usually limited to one retailer's walls at a time.

The smart setup uses MMM for the big budget split, geo and holdout tests to check and calibrate it, and network reports for day-to-day optimization.

‍

Build a calibration habit

‍

Once you have some lift results, compare them to what each network's attribution reported. You'll usually find that some networks' reports run closer to reality than others.

Use that gap as a calibration factor. If one network's measured incremental sales consistently come in at a fraction of its attributed sales, discount its in-platform reports accordingly when making optimization decisions. Over time, you get a more honest, comparable view of every network without running a test every week.

‍

Focus on total brand outcomes

‍

The ultimate scoreboard is total brand sales and share, not the sum of network dashboards.

Track a simple question every quarter: as total retail media spend went up or down, what happened to total brand sales, accounting for price, distribution and promotions? If spend doubled and sales didn't move, no amount of strong network ROAS makes that a win.

‍

Practical tips for getting started

‍

Start with your biggest network and run a proper holdout test. Use that as your first calibration point. Plan one geo test a year on a meaningful chunk of your budget. Agree on internal definitions for incremental sales and new-to-brand, so you're not comparing apples and pineapples. If you have enough data history, invest in an MMM, even a lightweight one. And share results with your retail partners. The good ones want to know what actually works too.

‍

A quick scenario

‍

A personal care brand runs retail media across five networks. Added together, attributed sales look fantastic. Total brand sales are up only modestly.

The team runs a geo test, pausing most retail media in a set of matched markets for six weeks. Total sales in the paused markets dip, but by less than attributed sales would suggest. Network-level holdouts show two networks driving strong incremental lift and two largely capturing shoppers who would have bought anyway. The brand shifts budget toward the high-lift networks and cuts back on branded keyword spend elsewhere. Total spend stays flat. Total sales grow faster.

‍

The bottom line

‍

In a multi-retailer world, every network is a narrator with its own version of the story. Incrementality is how you find out what actually happened. Combine tests, models and calibration, judge everything against total brand growth, and let the real lift decide the budget.

Five Eighty helps brands measure true incrementality across retail media networks and channels. If your network reports add up to more than your actual growth, let's find out what's really working.