What do retail media buyers wish they had known before buying? In Alium Research's verified buyer interviews, the central lesson is that purchasing retail media is easier than proving its incremental value and running campaigns across fragmented retailer systems. Buyers describe eight recurring challenges: independently verifying what a network reports, collecting and standardizing each retailer's data, working across both cross-retailer platforms and native consoles, getting value from Amazon Marketing Cloud without SQL expertise, choosing between software and a managed service, avoiding overlapping tools, learning each network's costs and rules, and — for retailers running their own networks — deciding whether to build their ad technology or share revenue with a partner. These problems carry costs beyond the advertising spend itself. For what retail media is, and why fragmentation is the brand's problem and the retailer's, see what is retail media?; for which tool draws which complaint, the top complaints about retail media platforms.
The eight lessons
A network's own incrementality study is the network grading its own homework
Can brands trust retail media networks' incrementality and ROAS figures? In these interviews, buyers question whether the sales and lift a network reports are genuinely incremental or would have happened anyway. The concern is that the network both sells the advertising and reports its performance, so buyers ask about attribution windows, compare results across networks and look for measurement that does not depend entirely on the retailer's own reporting. Their skepticism does not establish that the reported results are wrong. A consumer-health company says the only incrementality studies it has run “have been with retail media network partners. Again, very suspect because they're sort of grading their own homework.” A consumer-health manufacturer puts the same doubt across networks: “Walmart will tell you it's incremental or Target will tell you it's incremental, but for us as a whole, is it actually incremental?” A multi-brand consumer-goods company asks “What's the attribution window? How do I really prove that?” And a beverage-alcohol wholesaler running retail media for its suppliers says “I feel like ROAS is even being phased out as a really strong metric, and it's more about incrementality, category share, and more total GMV versus just ad attributed.” A global food manufacturer wants planning and measurement “not relying on the retailers to provide that.” Incrementality comes up without prompting in about a third of the retail media buyer interviews, a similar share to attribution buyers in this research. The implication — ours, not a buyer's — is to ask every network what its attribution window is and whether lift is measured against a holdout before comparing returns across networks.
Named in this context: Walmart · Target · Amazon
Measurement breaks on the data before it breaks on the method
Why is retail media measurement difficult across retailers? Buyers describe different data access, reporting formats and attribution methods on each network. Before they can compare performance, teams have to extract each retailer's data, standardize it and reconcile how sales are credited, so the work starts with usable data rather than a measurement model. A global packaged-food manufacturer's retail media lead says “really the big challenge is not necessarily how to measure the data,” but “It's how we actually get the data into these tools and how we clean it so that it's usable.” A global food manufacturer says “all this data is still very siloed because you can receive your retail media data in different formats from different providers,” and a global snacks manufacturer adds that “every single retailer is attributing sales differently.” The retailers hold on to it: “everyone's trying to strong arm and gatekeep their data because that's their biggest asset,” says one consumer-health manufacturer, and so “a lot of people end up building their own Power BI dashboard.” Even a platform's numbers need checking — a global food and beverage company says “There's been times when we've seen significant differences in the numbers in the UI versus what we're seeing in Skai, so we're consistently having to do data QAs.”
Named in this context: Skai · Power BI
One login across retailers does not retire the native consoles
Do platforms such as Pacvue and Skai replace retailers' native advertising consoles? Not entirely, according to these buyers. Cross-retailer platforms centralize campaign management, reporting and automation, but some retailer-specific features are available only in the native consoles, and buyers describe gaps in supported retailers, data and API coverage, particularly on Amazon. Much of that depends on what each retailer exposes through its integrations, so a platform reduces the number of systems a team uses without eliminating them. A supplement brand running Amazon, Walmart, Target and Kroger natively says “It is unavoidably many log-ins to different platforms, while having one log-in on everything will be very helpful and nice” — and, choosing between Pacvue and Skai, still plans to keep Amazon's DSP in its native console, because “there are still functionalities that APIs would not cover, and hence they are not available in Pacvue or Skai.” A global food and beverage company says “Some of the third-party platforms haven't really gotten up to speed in terms of capabilities or incorporating a lot of that information, so the Amazon UI is still a big focal point.” Coverage is the other gap: a global snacks manufacturer's complaint about Pacvue is a “Limited number of retailers that the data is available for.” A specialty retailer that runs its own network sums up the market: “There are many companies that claim they do all of that, but there's no silver bullet here.” Our recommendation, not a practice buyers describe: test the platform's coverage for your top retailers, and the features you use most, during the trial.
Named in this context: Pacvue · Skai · Amazon DSP
Amazon hands brands the deepest data — and the steepest skills bill
Do brands need SQL to use Amazon Marketing Cloud? Buyers describe it as the richest retailer data they work with, and say getting full value from it takes SQL or a partner who handles the queries. Teams without the skill rely on platforms such as CommerceIQ or Pacvue to make the data accessible, or on specialists to build audiences and analyze results — access to the data is not the same as being able to use it. A personal-care manufacturer says “It's an extremely powerful platform. The challenges with Amazon Marketing Cloud are that it requires a very technical person to get the most out of it,” and “you need to program it in SQL queries.” A global food and beverage company agrees that “some knowledge of SQL is definitely a huge help in terms of making sure that you're putting in the right information and creating the right audiences,” and leans on CommerceIQ for “a lot of audience creation capabilities that we on the team don't have the knowledge or experience to handle on our own.” Brands without the skill buy it: a baby-products brand values that Pacvue will “pull AMC data and make it more accessible without SQL querying.” Others take the data home — “It is an extra step to bring the data into our own environment to be visualized,” says a consumer-electronics manufacturer — and a global consumer-goods company describes “a lot of data engineering required to do all of this stuff.”
Named in this context: Amazon Marketing Cloud · CommerceIQ · Pacvue
Software is not service — decide who does the work before you pick the tool
Should brands use a retail media management platform or an agency? The answer depends on who will actually run the campaigns. Platforms such as Pacvue, Skai and CommerceIQ provide technology for managing retail media, but access to the software does not necessarily include the people needed to run campaigns, optimize performance or interpret results. Buyers describe under-used capabilities when their own teams lack the time or skills, and agency-run programs where they cannot separate the software's value from the people operating it. A large beer company says “I thought that there would be more managed services with CommerceIQ, and there were not,” and concluded “they are truly offering a software platform.” A mid-size consumer-products company running Skai itself says “I'd rather have a third party run our campaigns or more automated, because it just takes a lot of time.” The reverse problem shows up when an agency runs the tool: an online apparel retailer says “I can't tell if it's the tool or if it's the team.” Even with the platform in-house, a large packaged-food manufacturer says of Pacvue that “there are a lot of capabilities that we're not fully leveraging. I'm not sure if that's more on us or on Pacvue.” Brands that chose an agency for Amazon paid for expertise, and a household-products manufacturer ties it to results: “We also tie them to performance in how we pay them.” Our recommendation, not a practice buyers describe, is to decide who will build, run and check the campaigns before choosing the tool.
Named in this context: CommerceIQ · Skai · Pacvue · Flywheel Digital
The tools overlap, and large brands end up paying twice
Why do brands use several retail media and digital-shelf platforms? In these interviews, large brands run overlapping tools because no single platform covers every workflow: one identifies performance gaps on the digital shelf, another acts on them or runs the advertising. The overlap adds cost and complexity, but buyers sometimes keep several platforms because each does something the others do not. A large beer company says of Profitero and CommerceIQ “in some ways, what they offer is duplicative, but they also each have things that the other doesn't. So you need both to have the full suite,” and that “it seems like a lot of people are double dipping.” A global personal-care manufacturer uses CommerceIQ “to fill the gaps that we have with Profitero identifying performance gaps but then not automatically correcting them.” A global packaged-food manufacturer counts “six or seven different providers today and workflows requiring multiple different teams,” and a consumer-health manufacturer that cut back asks “why do we need both or two or three some cases?” Below a certain size the platforms are more than a program needs: an online apparel retailer calls Skai “a powerful tool when you have a lot of campaigns to manage. However, we don't have that many.”
Named in this context: Profitero · CommerceIQ · Stackline · Skai
Each network's rules cost money until you learn them
What unexpected costs do retail media buyers run into? Beyond media spend and software subscriptions, buyers describe network-specific minimum bids, DSP fees, agency requirements and budget controls, which cost money when a brand scales spend without knowing how each network's system works. A consumer-health manufacturer's digital commerce lead found one grocery network's minimum bid only by sitting in on an agency's audit of the account — the floor applies, in its words, as “a minimum no matter what,” and “That could generate a lot of waste if I'm not aware of that.” A direct-to-consumer hardware brand says connecting its streaming ads to its sponsored ads means paying for Amazon's DSP — “And that's a huge fee. And I have to have another agency run it.” A consumer-health brand found that Amazon's managed services had been “converted into self-service. So now you have to go to the partners.” And a mid-size consumer-products company's complaint about Skai is budget control: “the fact that I can't just put in a monthly budget and we never spend over that budget is ridiculous to me.” Our recommendation, not a practice buyers describe: audit each network's mechanics before scaling spend on it.
Named in this context: Amazon DSP · Instacart · Skai
For retailers, revenue share buys speed; building buys margin
Should retailers build their own retail media technology or share revenue with a technology partner? Retailers running networks describe a trade-off between launching quickly and keeping more of the margin and control. Revenue-share partners reduce the up-front work, while a more customized stack needs engineering time and money; the choice depends on how fast the retailer needs to launch and how much it wants to own. A national grocery retailer weighing ad servers says Criteo's model “is a revenue share, of course, and would be more expensive from a revenue sharing perspective than say a Kevel would be, but would require less upfront investment,” while “Kevel requires a significant amount of technical integration and engineering work, which is obviously expensive and takes time” — and “In retail media, it's extremely important to move as quickly as possible.” A specialty retailer running its own network says “you'll find a lot of networks, especially at the beginning, cobbled together with six, maybe seven different agencies and ad tech partners. And over time those companies start to insource those capabilities.” A national retailer building onsite and in-store media agrees that the stack is “going to have a bunch of different companies kind of put together unless you build it in house.” What a network costs to build and how long it takes to pay back is on what is retail media?
Named in this context: Criteo · Kevel · Instacart Carrot Ads
How buyers rate the tools
Among the retail media tools with publishable ratings in these interviews, Pacvue and Stackline each rate around 8, Skai and CommerceIQ around 7, and Profitero around 6.5. Most of these tools are rated by fewer than twenty-five buyers, so their averages are rounded. Amazon DSP and Criteo carry numbers, but most of the buyers rating them use them for general programmatic advertising or retargeting rather than retail media.
| Tool | Rating | What buyers describe |
|---|---|---|
| Pacvue | around 8 | One login, bulk actions and rule-based automation across retailers; its gap is the retailers its data covers, and some under-use what they bought. |
| Stackline | around 8 | Category-level market share with custom segments; buyers call it expensive and narrow in retailer coverage. |
| Skai | around 7 | Centralized reporting with longer look-back windows than the native consoles; its numbers sometimes diverge from them, and its budget controls frustrate. |
| CommerceIQ | around 7 | Trusted data, Amazon operations and audience building in Amazon Marketing Cloud; buyers describe it as software rather than a managed service, and weaker outside the US. |
| Amazon DSP | around 7 | Essential on Amazon for brands that use it; most of its raters here use it as a general programmatic platform, and its fees often come with an agency. |
| Criteo | 6.6 | Mostly rated as a retargeting tool; retail media users value one tool across many retailers and note less transparency than Amazon's own data. |
| Profitero | around 6.5 | One place to see the digital shelf against competitors; someone still has to find and fix the gaps by hand. |
| Flywheel Digital | not rated | Amazon media expertise from a managed service; too few ratings to publish a number. |
| Amazon Marketing Cloud | not rated | The richest retailer data buyers describe, and built for someone who writes SQL; too few ratings to publish a number. |
The stories behind the lessons
The counter-current: for many brands the platform layer works
Not every buyer is fighting their tools. A beverage-alcohol wholesaler runs its grocery and delivery networks from Pacvue and says “We have been very happy with it thus far,” adding that “they seem to be releasing a lot of new tools that have been helpful.” The packaged-food manufacturer that under-uses Pacvue chose it because “Pacvue really stood out from a usability perspective and being able to integrate with our agency partners.” A global beverage company says “the strength of CommerceIQ came with the accuracy of the data they are providing, which is critical in this area.” And for Amazon itself, a global packaged-goods company calls Amazon's DSP “the most important platform you need to use if you want to succeed on Amazon in terms of retail media.” The lessons above are about what it takes to get there, not a case against the tools.
Want your own stack read the same way?
Get mine →What this means for your retail media program
Across these interviews, choosing retail media technology is as much an operating and measurement decision as a software purchase. Brands have to decide how they will verify incremental results, bring retailer data together, divide the work between their own team and an agency, and avoid paying for overlapping capabilities. Retailers building their own networks face a separate decision: faster launch through a technology partner, or more long-term control over their stack and economics. Based on these buyer experiences, our recommendations — not practices buyers describe — start with measurement. Establish what counts as incremental and whose data will verify it before comparing networks on the returns they report. Map how each retailer's data will be collected, cleaned and reconciled before choosing a reporting platform. Test cross-retailer coverage and native-console limits, decide who will operate the campaigns, and budget for SQL expertise or a partner if Amazon Marketing Cloud is part of the plan. Finally, review the software you already pay for before adding a platform that overlaps it. For what retail media is, see what is retail media?
Common questions
What do retail media buyers wish they had known before buying?
Eight things recur in these interviews. A network's own incrementality study is the network grading its own homework. Measurement breaks on getting each retailer's data in and cleaned before it breaks on the method. One login across retailers does not retire the native consoles. Amazon Marketing Cloud is the richest data and needs SQL or a partner. Software is not service, so decide who does the work first. The tools overlap, and buyers pay twice. Each network's rules cost money until you learn them. And for retailers, revenue share buys speed while building buys margin.
Can you trust retail media incrementality and ROAS numbers?
Buyers treat them as the network's own numbers. Several describe a retailer's incrementality study as the retailer grading its own homework, ask what the attribution window is, and are moving their headline measure away from ROAS toward incrementality, category share and total sales measured on their own side. About a third of buyers discussing retail media raise incrementality without being asked about measurement — about as often as attribution buyers, and more often than CDP or email buyers.
Pacvue vs Skai: what do buyers say?
They are the two cross-retailer campaign platforms brands in these interviews weigh against each other, and buyers choosing between them say it comes down to functionality more than price. Pacvue is praised for one login, bulk actions and rule-based automation, with retailer data coverage as its gap; Skai for centralized reporting with longer look-back windows than the native consoles, with numbers that sometimes diverge from them. Pacvue rates around 8/10 and Skai around 7/10.
Do you need SQL to use Amazon Marketing Cloud?
To get the most out of it, buyers say yes, or a partner who has it. They describe Amazon Marketing Cloud as the richest retailer data they work with and as a tool built for someone who can write SQL queries. Brands without that skill pay a platform or an agency to front it, and some take the data into their own environment to visualize it.
Should you use a retail media platform or an agency?
Buyers say the two are different purchases. Some expected a managed service and bought software that their own team had to run; others hired an agency and found they could not tell whether the tool or the team was underperforming. Brands that chose an agency paid for Amazon expertise, and one ties the agency's fees to performance. Our recommendation, not a practice buyers describe, is to decide who will do the work before choosing the tool.
Your stack, read by the buyers who've run it
Peer-powered: built from hundreds of verified buyer interviews, and fitted to the tools you already run. You get a shortlist for your stack and your next steps.
Get my stack readFree · no password · your answers are anonymized before they inform research like this.