Self-Service vs Managed Retail Media Platforms: How Retailers and Marketplaces Choose in 2026

Last updated: October 2026. Reviewed by Kunal Damgude, Growth and Product Marketing Manager.

A self-service retail media platform gives advertisers a console to plan, launch and report on their own campaigns, while a managed retail media service puts the retailer's or a vendor's team in charge of running those campaigns for them. Most retailers and marketplaces need both, so the platform worth choosing is the one that runs both models on one auction and one set of reports.

Osmos is a white-label retail media operating system for retailers and marketplaces. Its modular apps for onsite, offsite and in-store ads sit over your existing stack, so you can break your monetisation ceiling faster without rebuilding it.

The retail media platform has grown from a sponsored search slot into a full network. For a Head of Retail Media in 2026 the decision is narrower and practical: which advertisers get a console, which get a team, and which platform can carry both without splitting the reporting, the billing or the auction.

Which retail media platforms support both self-serve and managed service, and which fit marketplaces or retailers?

The platforms a retailer or marketplace can build its ad network on in 2026 include Osmos, Mirakl Ads, Topsort, Kevel, Criteo, CitrusAd powered by Epsilon and Moloco, and they differ mainly in which business model they were built for, which channels they cover, and how much of the service layer they run for you. Marketplaces need a platform that turns thousands of sellers into advertisers; retailers need one that wins and keeps a smaller set of large brand budgets; a growing number of businesses are both at once.

Vendor details checked against each vendor's own pages and filings on 12 October 2026.

Osmos's Brand Self-Serve runs brand-facing ad buying on top of the retailer's existing ad server, while Ad Ops as a Service puts trained Osmos specialists inside the program on bids, pacing and creative QA, so self-serve and managed accounts share one platform.

Among the marketplace-led platforms, Mirakl describes its offer as "flexible, self-service technology with hands-on expert support", and its Jumia launch brought Mirakl Ads to more than 70,000 sellers. Topsort's core platform covers ad serving, auctions, self-serve and measurement, and it names marketplace seller monetization as one of its use cases alongside retailer networks.

On the retailer side, Kevel tells buyers its platform "is API-based so everything is completely flexible", which suits a retailer that wants to own the product roadmap of its network. Criteo reported 20% growth in Retail Media contribution ex-TAC across its underlying client base in Q2 2026, excluding the effect of scope changes at two retail media clients.

Epsilon calls CitrusAd "a self-serve retail media platform originally built to power onsite monetisation inside retailer websites and apps" and works with retailers across grocery, electrical, home improvement and marketplace environments. Moloco runs targeting, bidding and optimization on the retailer's data while advertisers manage campaigns inside the retailer's own platform, and the retailer keeps control of governance and creative review.

None of these vendors publishes a price list or take rate for retailers; every one quotes per deployment, so commercial comparison starts at the request-for-proposal stage. Skai, Pacvue and similar tools are bought by brands and agencies to manage their spend across networks, and a retailer does not build its network on them. Walmart Connect, Target's Roundel and Kroger Precision Marketing are networks owned by the retailers that run them. They are useful benchmarks for what a scaled network does, and no other retailer can license them.

Every platform in the table supports onsite placements, and most extend offsite. What separates them is whether onsite, offsite and in-store share a campaign, a wallet and a report, or sit in separate tools that the retailer's team has to stitch together by hand each month.

What is the difference between a self-service and a managed retail media platform?

A self-service retail media platform hands the campaign to the advertiser, and a managed service keeps the campaign with the retailer's team or a vendor's team acting for it. Google's Search Ads 360 documentation states the self-serve half of that split in one line: "With a self-service model, you can set up, optimise and report on your campaigns on your own" (Google Search Ads 360 Help). A managed service moves the same work back across the table, together with responsibility for the numbers the brand later audits.

Self-service retail media is a model in which advertisers and sellers log in to a console the retailer provides and plan, launch, optimize and report on their own campaigns, paying for media and sometimes a platform fee. Osmos runs this side through Brand Self-Serve, where advertisers plan, launch and report through one portal while the retailer's team sees the same campaign status they do.

A managed retail media service is a model in which the retailer, or a vendor working for the retailer, runs the advertiser's campaigns end to end, usually for a service fee or a minimum commitment on top of media. Osmos runs this side through Ad Ops as a Service, with Osmos specialists working on delivery, ROAS and wallet share inside the retailer's program.

Most networks run the two models side by side. Osmos's rule of thumb is that self-serve is where advertiser count grows and managed service is where account value grows, and a network that scales keeps both open. Staffing and scale decide where each advertiser starts: a retailer that cannot staff ad operations starts managed and moves the long tail to self-serve as the program matures, while a marketplace with thousands of sellers opens self-serve first because no team could serve that many accounts by hand.

Self-service advertising infrastructure, the phrase buyers and analysts use for the console, the auction and the billing behind it, has to cover the whole advertiser account. A console that works for a retailer needs onboarding an advertiser can finish without a call, exception handling for rejected creative and out-of-stock products, wallets and credit, and reports that match the ones the managed desk sends.

Why does platform choice differ for a marketplace and a retailer?

Platform choice differs because a marketplace earns ad revenue from many sellers competing for placement on its own pages, while a retailer earns it from a smaller set of large brand budgets activated against its shopper data, and the two need different tools. Marketplaces need multi-seller self-serve onboarding, seller wallets and auctions built for thousands of small bidders; retailers need identity, closed-loop measurement and onboarding that suits large brand teams.

Marketplace retail media is seller-funded advertising on marketplace-owned inventory, where the marketplace offers placements in its search results, category pages and product pages to the third-party sellers listed on it. The economics are covered in Osmos's guide to why marketplaces earn from retail media; for platform choice, the requirement that matters most is converting sellers into advertisers at scale. The size of the prize shows in India, where advertising contributed 31% of Flipkart's total revenue in FY25, with ad income up 27% to Rs 6,317 crore, according to Storyboard18 (September 2025).

Retailer-direct retail media is brand-funded advertising on a retailer's own sites, apps, stores and offsite extensions, targeted with the retailer's first-party customer data. The advertiser set is smaller and each account is larger: consumer goods, beauty, electronics and lifestyle brands that already buy media at scale and want proof that it moved sales. The longer path from a single program to a full network is described in Osmos's piece on retailers becoming media networks, and retailers running several store brands face an extra layer of choices covered in the multi-banner retail media network guide.

US retail media ad spend will reach $71.09 billion in 2026, up 17.8% from $60.32 billion in 2025, according to a December 2025 EMARKETER forecast reported in its FAQ on retail media networks (January 2026). More than 200 retail media networks now operate globally, per the RMIQ 2025 Retail Media Market Guide. With that many networks chasing the same brand budgets, a network that cannot onboard its advertisers efficiently starts at a disadvantage.

Commerce media is the wider category that holds both models: advertising run by any company that owns a transaction, from retailers and marketplaces to delivery apps, travel firms and payment providers. Osmos's explainer on what commerce media is sets out where retail media sits inside it, which matters for platform choice because the same platform increasingly has to serve a grocery chain's marketplace, its stores and its delivery app at once.

Seller wallet management, covering prepayments, credit lines, refunds and currency, decides more deals than most rows in the table, because it separates platforms built for marketplaces from retailer tools adapted to them; Osmos's Wallets & Billing handles top-ups, credit lines, invoices and rule-based incentives on one ledger. Multi-tenant architecture matters because each seller's data, billing and reporting must be isolated from every other seller's. Retail media adtech for both retail and marketplace environments therefore has to run multi-tenant seller accounts and single-tenant brand campaigns side by side, which is how Osmos is built to serve retailers and marketplaces on one platform.

Hybrid businesses are now common: a grocery chain with a third-party marketplace, a fashion retailer with concession brands, a delivery app with both restaurants and consumer goods brands advertising. For them the platform has to meet the requirements in both columns of the feature table. The food delivery retail media guide shows how that mix plays out when restaurants and brands bid for the same app surfaces.

How do managed services support commerce media campaigns?

Managed services support commerce media campaigns by putting an ad operations team, either the retailer's own or the platform vendor's, in charge of planning, building, bidding, optimizing and reporting on campaigns for advertisers who hold budget but have no team to run it. For the retailer a managed service turns idle budgets into live campaigns quickly, and its headcount grows with every account it serves. A self-serve console has the opposite profile: it is slower to turn budgets into campaigns and cheaper to scale.

An April 2025 IAB analysis found that "the shift from managed service to self-serve platforms is lowering operational costs for retailers but increasing complexity for brands" (IAB, April 2025). For the retailer, that makes the decision a matter of which advertisers get which model and what each one costs to run.

Managed service providers simplify commerce media campaigns for the brand. The service team handles campaign configuration for the brand. The team also brings format knowledge the brand often lacks, since a brand team that knows search ads rarely knows how to traffic an in-store screen campaign or an offsite audience. And the provider reconciles the results into one report, in metric definitions the brand can audit. Osmos offers that layer as a paid service through Ad Ops as a Service, alongside the self-serve tools, so a retailer can decide account by account which side of the line each advertiser sits on.

Retail indirect managed services and retail direct managed services are procurement terms for outsourcing a retailer's own back-office categories, such as facilities, IT or logistics spend.

What is a managed service in commerce media, and what does it actually do?

A managed service in commerce media is a service model in which the network runs the advertiser's campaign end to end. Google's Search Ads 360 documentation puts it from the network's side of the table: "With a managed service model, the retailer sets up, optimises and reports on campaigns for the brand." Offering it means taking the campaign onto the retailer's own books, so that the retailer's team, or a vendor's team acting for it, owns the plan, the build, the in-flight decisions and the numbers.

Osmos breaks a managed retail media service into separate workstreams, because each one carries a different cost to the retailer:

1. Planning and packaging. Translating an advertiser's objective into an inventory plan across onsite search and category placements, offsite audience extension and in-store screens where the retailer has them. This is the workstream that most often decides whether a budget renews, and Osmos's guide to retail media ad packages brands pay a premium for covers how the package itself is built.
2. Campaign build and creative. Product Ads are assembled automatically from the catalog listing, so listing quality is the only lever there. Display, sponsored brand, video and offsite units are the formats where a managed team builds and traffics creative on the advertiser's behalf; the guide to retail media ad formats sets out which format takes which asset.
3. Bid and budget management. Pacing, floor management and daypart decisions inside the auction. Retailers running automated bidding delegate more of this to the system, and Osmos's guide to automated bidding and auctions covers where that line sits.
4. Mid-flight optimization. Moving spend between placements, products and keywords while a campaign is live. This is the workstream advertisers most often believe they are buying, and the one that most often justifies the fee.
5. Reporting and review. Reports on a set cadence, in metric definitions that survive the advertiser's own audit. The IAB and MRC Retail Media Measurement Guidelines (January 2024) are the reference point for impression counting, viewability, invalid traffic, attribution windows and reporting.

Why advertisers ask for a managed service at all is measurable. In IAB buyer research, 69% of retail media ad buyers cited the need to resolve complexity in the buying process and 60% wanted better collaboration with their networks, while only 12% were "very satisfied" with the relationship (IAB, September 2023). The staffing picture behind those numbers shows in where the money sits. A Forrester survey reported by Digiday (March 2025) found that 36% of US retail media spend comes from trade marketing budgets and 26% from shopper marketing budgets, so the person holding the budget often sits in a commercial team, away from the media team and its campaign specialists.

Those trade and shopper marketing budget holders are the accounts a managed service exists for. They measure success in sales and distribution, and they will spend with the network that makes the campaign simple and the report clear.

Which advertisers should a retailer put on managed service, and which on self-serve?

A retailer should put its largest and most complex advertisers on managed service and run everyone else on self-serve, then move accounts across the line as their spend and skills change. A workable starting rule is to give a managed seat to the small group of advertisers that carry most of the spend, widen that group as demand deepens, and review the line whenever the cost of serving a managed account approaches what a hire would cost. The retailer should document the line and review it each quarter as advertiser spend changes.

Enterprise advertisers use managed commerce media services the way they use any other delegated buying relationship. They hold the budget, hand over a brief, and expect the network's team to plan, build, optimize and report against it. The retailer has to decide which advertisers earn that seat and what that seat costs to staff. The same logic answers which retail media platform is best for enterprise companies on the retailer side: the one that lets a large network serve its top brands by hand and its long tail by software without running two systems.

Brands with modest spend on a single retailer "often do fine with an agency or a generalist performance marketer handling it as one channel among several," according to a 2026 retail media hiring guide from eCommerce Placement (July 2026), which names the trigger for in-house specialists as spend concentrated across two or more retailers. From the retailer's side, an advertiser who has crossed that line will either hire or ask for a managed seat, and the advertisers who have not crossed it are the self-serve base.

Managed programmatic demand and self-serve ad ops sit at the two ends of the same spectrum. A managed programmatic desk buys on the advertiser's behalf across the network's inventory and offsite partners; self-serve ad ops means the advertiser's own team, or its agency, works the console. The self-service DSP versus managed DSP question that brands ask about offsite buying follows the same logic, and a retailer running offsite needs to support both, because agencies with trading desks want the console and consumer goods brands without one want the service.

The hardest group to serve is the middle: torso brands with real budgets and no retail media specialist. They are too small for a dedicated account manager and too unfamiliar with the formats to spend well alone. Osmos's Ad Sales service exists for exactly that group, with an Osmos team that handles prospect lists, onboarding and first-campaign activation for the torso and tail advertisers a retailer's own sales team has no capacity to work.

Activation of lapsed advertisers shows what the service layer adds. Osmos helped Superbalist, South Africa's top online fashion and lifestyle retailer, reactivate 151 advertisers within 20 days, and 70% of its GMV-driving brands became active advertisers on the media network. Superbalist had launched its retail media program with Osmos, but advertiser engagement stayed low and ad revenue stalled through leadership and point-of-contact changes. The Osmos team agreed a priority brand list with the retailer and later widened it for category coverage and GMV potential, ran virtual training sessions and on-ground alignment meetings with internal teams and advertisers, and set up a dedicated incentive program for advertisers. The priority list, the training and the incentive program produced the result together.

The split between managed and self-serve also depends on the channel, and in-store rarely arrives self-serve first, because screen inventory, scheduling and proof of play are unfamiliar to most brand teams, and in-store is planned and priced by store, day and time. The mechanics of that surface are set out in Osmos's guide to how retailers monetize in-store traffic, and where the channel is heading is covered in the piece on the future of in-store retail media monetization.

What should a retailer look for in a commerce media platform with managed services?

A retailer should look for a commerce media platform that runs self-serve and managed accounts in one auction, one inventory pool, one billing system and one set of reports, so the best commerce media platform with managed services for advertisers is, from the retailer's seat, the one that never forces a choice between the two. A platform that runs self-serve and managed accounts as one hybrid program needs five things:

1. One auction and one inventory pool. Self-serve and managed demand should compete in the same auction for the same inventory. Two stacks mean two yield curves and no single view of fill; Osmos's comparison of one stack against several vendors sets out the cost of the split.
2. Service tiering inside the platform. Service level should be a property of the account, so an advertiser can move from managed to self-serve without being migrated between systems and losing its campaign history.
3. Reporting parity. The same metric definitions across both surfaces, per the IAB and MRC guidelines, or managed and self-serve accounts will report different numbers for the same placement.
4. Wallet, billing and credit across both. Prepaid wallets suit long-tail sellers and invoiced credit suits large brands. A platform that supports only one payment model quietly caps which advertisers a retailer can take.
5. Multi-tenant onboarding for marketplaces. Marketplace demand is long tail by construction, thousands of sellers each spending modestly, so a self-serve console is part of the structure of a marketplace network.

Osmos's Orchestration layer is built around those requirements: Brand Self-Serve, Wallets & Billing, Inventory Studio and BYO Media Channels run setup, approvals, inventory and billing as one workflow, so the same team can serve advertisers well beyond its top twenty.

Leading commerce media platforms are supporting advertisers in 2026 by widening self-serve access, and the networks that have scaled run both models. Walmart Connect launched a self-serve onsite display platform in July 2026 and kept the alternative open in the same announcement: "For advertisers who prefer additional support, Walmart Connect's managed-service offering remains available" (Retail Insider, July 2026). Instacart extended its Ads Manager to retailers in May 2026, so a retailer can create and manage basket-level offers itself. Criteo opened full self-service access to its GO platform in March 2026, aimed at small and growth-stage advertisers. Topsort extended self-service to in-store screens in May 2026, with its chief executive, Regina Ye, explaining that "retail media teams are under pressure to scale new surfaces without increasing operational complexity" (Path to Purchase Institute, May 2026).

EMARKETER's analysts set out the pressure on every network below the top two in their 2026 commerce media predictions: growing networks must keep up with retail media giants like Amazon and Walmart, and technology is one of the routes the analysts name (EMARKETER, January 2026). A white-label platform offers another route: the retailer licenses the technology and keeps its brand, its data and its advertiser relationships.

The commerce media platforms that provide the best personalization run relevance ranking and targeting on the retailer's own first-party data, with the retailer setting the rules. Osmos's Product Ads ship relevance ranking and retailer-set bid floors as part of the ad engine. The trade-offs between targeting approaches are covered in Osmos's guides to contextual versus behavioral targeting and AI-powered ad serving and behavioral targeting, and the data layer underneath in the first-party data targeting guide.

Agentic self-service advertising portals for retail media networks are the newest item on the checklist. The idea is that software agents plan, build and tune campaigns for advertisers who would otherwise need a managed seat, which would let a retailer serve the long tail with managed-quality campaigns at self-serve cost. Osmos's Sofie AI Agents are seven specialist agents for media planning, creative, performance, reporting and inventory, and the live page lists all seven as coming soon. The system design behind that shift is explained in Osmos's essay on multi-agentic AI orchestration in retail media. When a vendor demonstrates agents, ask which ones are live today and which are announced.

The retail media measurement platform worth investing in measures every campaign on the network the same way, managed or self-serve, in definitions the advertiser can audit; Osmos's attribution and measurement guide covers the methods in depth.

What does a self-serve advertiser portal cost a retailer to run?

A self-serve advertiser portal costs a retailer the platform itself, the onboarding and support desk behind it, and the demand generation that fills it, and the support desk is the cost that most often surprises retailers. Vendors do not publish portal pricing or take rates; licence fees, revenue shares and combinations of the two are quoted per deployment, which is why the cost of self-serve portal software for retail media advertisers has to be built from the retailer's own operating plan.

1. Platform. A licence fee, a revenue share or a combination, depending on the vendor and the deployment model. Ask for the total at three different advertiser counts, because pricing that looks cheap at fifty advertisers can look different at five thousand.
2. Onboarding and support. This is where most of the cost sits. A console moves campaign building to the advertiser, and the advertiser still needs someone to explain why a campaign did not spend, why a listing was rejected and how to read the report. In Osmos's Brand Self-Serve, out-of-stock products, rejected creative and pacing misses surface as exceptions, the system routes routine ones automatically, and people handle the rest, which is where the support cost is won or lost.
3. Demand generation. A portal earns only when advertisers use it. Activating the long tail is a recruitment and education program, and its cost belongs in the portal's own business case.

Managed service is priced differently: a service fee or a share of media, usually with a minimum, so in the early months its cost tracks revenue. That is why many networks launch managed and add self-serve once the advertiser count makes the support desk worth staffing. The full business case, including why margin is the number to plan around, is set out in Osmos's piece on the retail media business case.

Managed service or self-serve for CTV and offsite with a lean team?

A lean team should run onsite on self-serve first and keep offsite and CTV managed until it has a dedicated programmatic desk, because offsite and CTV need audience construction, frequency management and incrementality reads that a listing campaign never does. Onsite Product Ads are the closest thing retail media has to a self-operating format: the ad is assembled from the catalog and the auction does the allocation. Offsite and CTV need more hands, and the guide to offsite channels for retail media networks sets out what each channel demands.

CTV is growing fast inside retail media. EMARKETER's Sara Lebow wrote in March 2025 that "Retail media CTV ad spend is growing about three times faster than retail media search is" (EMARKETER). Brand budgets are following that growth into CTV. A retailer's forecast for a new CTV line should come from its own audience size, advertiser roster and sales capacity.

Between self-serve and managed sits a co-managed model. A 2026 comparison of CTV platform models by Jamloop (July 2026) defines it directly: "Co-managed CTV combines direct platform access with strategic and operational support from the technology partner." Fully self-serve platforms, in the same comparison, "give your internal team or agency trading desk direct access to the campaign management interface." Test that against the retailer's own bench: without dedicated programmatic staff, a self-serve CTV console mostly moves the work to a brand team that usually lacks them too.

Which model fits depends on the retailer's desk. Keep CTV and offsite fully managed while the program is early and there is no programmatic desk. Move to self-serve once that desk exists and wants speed, control and transparent fees. Use co-managed in between, when the retailer wants control without carrying the whole execution risk through ramp-up. Price belongs in the same decision: the Jamloop comparison cites enterprise DSP benchmarks showing effective programmatic markups of 23% to 29% on top of raw media spend. That figure covers programmatic buying in general, so treat it as the order of magnitude when pricing a managed offsite package or setting the bar an in-house desk has to beat.

Osmos's offsite retail media offers both ends of that range. Its Extended Media Network runs Google, Meta and TikTok on the retailer's own seat with channel-level margin for the retailer, and its Guaranteed Outcomes app has Osmos run tail and torso advertisers across Meta, Google and TikTok, managed to outcomes, so the retailer earns the ad revenue without adding ad ops headcount. Retailers moving from onsite to offsite for the first time will find the sequencing in Osmos's guide to expanding retail media beyond onsite into offsite.

Advertisers who cannot reconcile an offsite campaign do not renew it, so measurement has to be in place before offsite scales. Fragmentation across networks makes it "challenging for brands to achieve standardized measurement and interoperability across networks," as the same April 2025 IAB analysis puts it. IAB Europe's Commerce (incl. Retail) Media Measurement Standards V2, released in January 2026 with a transition window to the end of July 2026, give both sides a shared vocabulary (IAB Europe, January 2026). Retailers extending into store screens have a reference in the IAB in-store retail media guidelines.

How long does it take to set up and launch a commerce media campaign?

Setting up a commerce media campaign takes from the same day for a self-serve Product Ads campaign to several weeks for a managed offsite or CTV campaign, once the retailer's platform is live; standing up the platform itself is a separate and longer clock. A retailer comparing which commerce media platforms are easiest to implement should measure platform implementation time separately from campaign setup time.

Platform standup time depends on how much of the existing stack the platform can sit on: the catalog feed, the ad server if one exists, identity, checkout events and billing. Vendors quote their own targets, and the useful comparison is between those targets and the retailer's own integration list. Moloco states that a retailer can go live in as little as six weeks; Mirakl says a migration to Mirakl Ads takes "weeks to months with zero revenue loss during the switchover." Osmos's modular apps are designed to start one at a time over the retailer's existing stack, so a network can launch with onsite Product Ads and add offsite, in-store and the service layer as demand arrives. Ask every vendor for a named customer who went live on a comparable stack, and for the date their first campaign served.

Campaign launch time depends on format and service model. A self-serve Product Ads campaign can go live the same day, because the ad is assembled from the catalog and the approval path is short. A managed offsite or CTV campaign runs longer: creative production, audience build, insertion order and quality assurance all sit in front of the first impression. In-store campaigns add booking, scheduling and proof of play.

Retailers should publish a launch commitment per model and per format and hold to it, because time to first impression is one of the few things an advertiser can compare across networks before committing budget.

How should a marketplace or a retailer decide when and how to launch?

A marketplace should launch when it has enough seller depth and search traffic to make seller-funded placements competitive, and a retailer should launch when it has enough loyalty data and category depth to make a brand proposition compelling; in both cases the right first step is a thin, well-run program. Whether to build that stack in-house is a separate question, covered in Osmos's build vs buy analysis and, for marketplaces specifically, in the piece on how marketplaces build their own ad platform.

For a marketplace, Osmos's launch sequence runs in this order:

1. Start with seller activation tools. Self-serve onboarding and a workable seller wallet matter more than an elaborate bidding interface when sellers are the demand.
2. Pick an auction that keeps small sellers bidding. Second-price and similar designs limit overpaying and keep small-seller participation viable on Product Ads.
3. Instrument measurement from day one. Sellers renew on results they can see in their own seller dashboards.
4. Pace seller spend. Wallet-based prepayments and credit lines protect both seller liquidity and the marketplace.
5. Plan formats around the category mix. Fashion marketplaces need shoppable video, story ads and editorial placements; consumer goods marketplaces need Product Ads and display; B2B marketplaces lean on category-page placements.

Post-launch performance expectations for marketplaces are covered in Osmos's analysis of retail media ROI for marketplaces.

For a retailer, Osmos's launch sequence runs in a different order:

1. Data audit. Map loyalty, point-of-sale, online behavior and app usage data, and assess identity coverage.
2. A unified shopper record. Stand up the customer data layer that targeting and measurement will run on. This is where many retailer programs stall.
3. Onsite inventory. Define and instrument Product Ads slots, display placements and category-page real estate on the retailer's own pages.
4. Launch with a short advertiser list. Start with three to ten endemic brands in strong categories.
5. Measurement as a launch deliverable. Incrementality and closed-loop reporting should be ready for the first renewal conversation.

Osmos lists the selection mistakes that marketplaces and retailers share in the order a platform decision meets them:

1. Choosing a marketplace-optimized platform as a retailer, or the reverse. Marketplace platforms can lack first-party data activation and brand portal depth; retailer platforms can lack seller self-serve, wallet management and multi-tenant architecture. Procurement rarely surfaces the mismatch, and it shows about six months in.
2. Over-engineering before proving demand. Elaborate stacks commissioned before endemic advertisers have committed are the most common waste pattern. Getting a thin program live and reading real demand signals compresses that risk.
3. Underestimating data control. Some arrangements place the retailer's customer data inside structures the retailer does not fully control. White-label deployments keep the retailer as the data controller and the brand in front of advertisers.
4. Supporting one pricing model. Marketplace sellers respond to cost-per-click Product Ads; brands buying display or CTV expect cost per thousand impressions; premium tenancy is a fixed fee; in-store is priced by store, day and time. A platform that supports only one model forces inventory choices that limit revenue.
5. Ignoring measurement fragmentation. A network that launches without a measurement plan compatible with the IAB and MRC guidelines and IAB Europe's V2 standards sets up its own renewal problem.
6. Treating retail media as a side project. The largest marketplaces now run advertising as a core revenue line. Networks that resource it with a dedicated team, platform and measurement capture the upside.

Vertical mix changes the weighting of every step, and Osmos's sector deep dives show how grocery, pharmacy, fashion and other categories balance them. To see how Osmos would run self-serve and managed accounts on one stack for a specific network, request a demo.

Frequently asked questions

Should a retailer start with self-service or a managed retail media service?

A retailer should start with the model it can staff and bill for on launch day. Managed service needs a team that plans and runs campaigns, either the retailer's own or a vendor's such as Osmos's Ad Ops as a Service, plus invoiced credit for large brands. Self-serve needs onboarding an advertiser can finish without a call, prepaid wallets and reports an advertiser can read alone, which Osmos covers with Brand Self-Serve and Wallets & Billing.

Can a marketplace run self-serve for sellers and managed service for top advertisers on one platform?

Yes, provided the platform treats service level as a setting on the account, so both run in one system. Sellers onboard through the self-serve console and pay from prepaid wallets; the largest brands get a managed team and invoiced credit; both compete in the same auction and appear in the same reports. Osmos runs both on one stack, with Brand Self-Serve for the long tail and Ad Ops as a Service for the accounts that want a team.

How can marketplace sellers run ads on a retailer's platform?

Sellers run ads on a marketplace's platform through a self-serve console tied to their seller account: they log in with their existing seller credentials, fund a wallet or draw on credit, pick products to promote, set bids and budgets, and see results in a seller dashboard. The retailer controls the rules underneath, including which categories open to advertising, bid floors, creative approval for display units and how spend is billed against seller payouts. The retail media tools marketplace sellers need are therefore mostly the marketplace's own: onboarding, wallets, product selection and reporting.

How do professional-grade retail media platforms differ from a standard retailer interface?

A standard retailer interface is a basic form where an advertiser selects products and assigns a budget. A professional-grade retail media platform adds what a network needs to run at scale, including multi-channel campaigns from one brief, pacing across channels, credit lines and incentives, creative review, role-based access for the retailer's sales and ops teams and the advertiser's, and closed-loop reporting in auditable definitions. A network needs a professional-grade platform once its advertiser count passes a few hundred, the point where manual campaign management stops scaling.

Which retail media platforms support both onsite and offsite media?

Most platforms in the comparison table support both onsite and offsite, and the difference is whether they run in one campaign and one report. Osmos covers onsite, offsite and in-store retail media from the same platform, so a single brief can run across channels.

Can mid-sized retailers compete with Amazon and Walmart in retail media?

Yes. Mid-sized retailers win on audiences the giants cannot reach in the same way (category depth, local loyalty, niche shoppers), on faster launches through white-label platforms, and on service: a brand with a modest budget gets more attention and quicker answers from a regional network than from the largest marketplace in the market.

Sources

1. Google, Search Ads 360 Help: About commerce media service models.
2. IAB (Cintia Gabilan), Commerce Media: At the End of the Beginning?, 8 April 2025.
3. IAB, IAB and MRC Release Retail Media Measurement Guidelines, 13 September 2023.
4. IAB and MRC, Retail Media Measurement Guidelines, January 2024.
5. IAB, In-Store Retail Media guidelines.
6. IAB Europe, Retail Media Spotlight: Kicking Off 2026 with New Standards, Smarter Creative and Strong Momentum, 22 January 2026.
7. EMARKETER, FAQ on Retail Media Networks: How Marketers Should Allocate Budgets in 2026, 9 January 2026.
8. EMARKETER, Commerce Media Predictions for 2026, 5 January 2026.
9. EMARKETER (Sara Lebow), Sizing up the CTV opportunity in retail media, 3 March 2025.
10. RMIQ, 2025 Retail Media Market Guide, 19 May 2025.
11. Storyboard18, Flipkart's Ad Income Jumps 27% to Rs 6,317 Crore in FY25, 15 September 2025.
12. Digiday, Marketers and agencies grapple with divisions over who controls retail media spend, 11 March 2025.
13. Retail Insider, Walmart Connect Launches Self-Serve Onsite Display Platform to Expand Retail Media Capabilities, 29 July 2026.
14. Path to Purchase Institute, Topsort Debuts Self-Service In-Store Ads to Scale Retail Media Past Digital Platforms, 4 May 2026.
15. Jamloop, Self-Serve, Managed, or Co-Managed? A 2026 CTV Advertising Platform Comparison, 31 July 2026.
16. eCommerce Placement, Retail Media Hiring: How to Fill Amazon Ads, Walmart Connect and Retail Media Roles, 21 July 2026.
17. Instacart, Instacart Expands Ads Manager to Retailers, Unlocking New Self-Serve Tools to Drive Growth Across the Marketplace, 13 May 2026. company.instacart.com
18. Mirakl, Mirakl 2025 Key Results and Launch of Mirakl Nexus, 26 February 2026 (checked 12 October 2026). mirakl.com
19. Mirakl, Mirakl Ads product page (checked 12 October 2026). mirakl.com/products/mirakl-ads
20. Topsort, homepage and H-E-B case study title (checked 12 October 2026). topsort.com
21. Kevel, homepage and iFood customer story (checked 12 October 2026). kevel.com/customers/ifood
22. Criteo, Criteo Reports Second Quarter 2026 Results, 5 August 2026. criteo.investorroom.com
23. Criteo, Criteo Expands GO with Full Self-Service Access to Its AI-Powered Performance Platform, 31 March 2026. criteo.investorroom.com
24. Epsilon, CitrusAd is now part of the Epsilon retail media offering (checked 12 October 2026). epsilon.com
25. Moloco, Moloco Commerce Media product page (checked 12 October 2026). moloco.com/moloco-commerce-media