Ask anyone who buys retail media what makes it hard and you'll get some version of the same list. There are too many networks. Every one of them works differently. The data sits in separate places that don't talk to each other, and all of it keeps getting more expensive.
None of that is wrong. The trouble is the list makes it sound like four separate problems, each with its own fix: a consolidation plan here, a new tool there, a tougher conversation about CPMs. Look closer and they mostly trace back to one thing. Brands can't compare networks on the same terms, because every network measures its own results its own way.
Every network grades its own homework
Each retail media network reports performance with its own attribution window, its own definition of a conversion and, more and more often, its own lift studies. ROAS already has a basic flaw: it gives ads credit for sales that would have happened anyway. On top of that, it isn't even the same metric from one network to the next. One might credit a sale that happened two weeks after someone saw an ad, while another only counts clicks.
Even when networks run incrementality tests, they're testing themselves. The network sets up the test, picks the window and reports the result. That doesn't make the results wrong. It does mean you can't line them up side by side and trust what the comparison tells you.
Once you see that, the four complaints start to look different.
Fragmentation and data silos are the same complaint
The number of networks isn't really what hurts. Brands have bought across dozens of TV networks and hundreds of publishers for decades. What made that manageable was a shared currency for comparing them, flawed as it was. Retail media never got one, so every new network adds another set of numbers that can't be stacked against the rest.
Data silos are that same gap seen from the data side. Each network's first-party data is valuable, which is exactly why each one guards it. Clean rooms help brands get at some of it. But pulling audience data from five networks into one place still won't tell you which of the five drove sales you wouldn't have gotten anyway. More data in one room is useful. It doesn't answer the allocation question by itself.
Operational complexity is partly just real
This is where the argument only half holds. Different interfaces, workflows and activation processes eat a team's time, and no measurement approach makes a clunky buying platform less clunky.
But look at where the work actually piles up. Most of the hard calls happen inside a single retailer. Joint business plans mean money rarely moves from Target to Walmart mid-year. It moves between campaigns, products and ad types on the same network, and every one of those decisions gets made in that network's interface, against that network's numbers. Without a consistent read on what's incremental, teams keep every campaign running and adjust by feel. Once you can see which campaigns and products actually drive incremental sales, there's a lot less to babysit.
Rising costs turn it into a budget problem
When retail media was cheaper and growing fast, nobody needed to know exactly which network was earning its spend. Higher CPMs changed that. Finance wants answers now, and the standard advice is to move money toward the partners driving incremental growth.
That advice assumes you can tell which partners those are. If each network's numbers come from its own method, the one that looks most efficient might simply be the one with the most generous attribution. And platform ROAS won't sort it out for you. In our own data, ROAS and incremental ROI don't correlate. Ranking networks by ROAS and cutting from the bottom can mean cutting the ones that were actually working.
Fix the comparison first
The way out is one standard for measuring every network, applied by someone who isn't also selling the media. Same definition of an incremental sale across Amazon, Walmart, Instacart, Target, DoorDash and the rest. With that in place, the questions behind the four complaints start getting answers: which networks to keep, where the next dollar goes, whether a CPM increase is worth paying.
Brands that have made the switch see it show up in results.
The four-item list will keep showing up in industry decks, since it describes what the job feels like day to day. Treat it as four problems, though, and you'll end up buying four fixes. Most of it comes back to not being able to compare networks honestly. Start there.
