For CPG and consumer brands in 2026, retail media is no longer an experimental line item — it is among the fastest-growing and most strategically important pieces of the media plan. Retailers and commerce platforms discovered that their first-party shopper data is a high-margin advertising asset, brands discovered closed-loop sales attribution they could not get anywhere else, and the deprecation of third-party cookies made the deterministic purchase data underneath these networks more valuable than almost any other targeting signal. The practical question for a brand is no longer whether to invest in retail media, but how to allocate across the networks that matter — and three stand out: Criteo, Amazon DSP, and Instacart.
This guide compares those three on the dimensions that drive allocation decisions: how each one works and the model it represents, ecosystem reach and inventory, ad formats and where each wins, attribution quality, AOV ranges, and category fit. It closes with a 30-day plan to build a defensible multi-network allocation. The throughline is that these three are not interchangeable — they are an open network, a closed ecosystem, and a category specialist, and the right strategy uses each for what it does best while measuring all three honestly rather than chasing whichever dashboard prints the highest ROAS.
Retail media networks advertise to shoppers who are already in or near a purchasing environment, and they attribute the resulting sales to themselves through closed-loop data. That combination produces spectacular reported ROAS — and systematically over-credits the networks for demand-harvesting they did not create. A shopper already buying your category on Instacart, already shopping your brand on Amazon, or already in-market across Criteo's retailers would often have purchased regardless; the retail media ad merely got the last-touch credit. Brands that allocate toward whichever network reports the highest ROAS end up over-funding capture and starving the upper-funnel demand creation that feeds it, then plateau on total sales while retail media ROAS stays flatteringly high. The fix is to judge networks on incremental contribution and reconcile claimed sales against total sales, not to chase dashboard ROAS.
Why retail media network choice matters in 2026
Retail media — advertising built on retailers' and commerce platforms' first-party shopper data and inventory — moved from a niche tactic to a central pillar of CPG media plans over 2023-2026, driven by three structural forces that also make the choice between networks consequential.
First, cookie deprecation revalued first-party purchase data. As third-party cookies lost most of their addressability, the deterministic, logged-in purchase data that retailers and commerce platforms own became one of the most valuable targeting and measurement assets in advertising. Retail media networks are the productization of that data. The networks with the deepest, cleanest purchase data — Amazon's commerce graph, Instacart's grocery basket data, the aggregated retailer data Criteo connects — gained a durable advantage precisely as open-web targeting degraded.
Second, retailers built retail media into a major, high-margin business. Retailers realized that advertising against their own shopper data is far more profitable than thin retail margins, and invested accordingly, expanding formats, inventory, and self-serve tooling. This turned retail media from a few sponsored-product placements into full networks with on- and off-property reach, audience extension, and sophisticated measurement — and it created real differentiation between networks in capability, reach, and data.
Third, brands gained closed-loop sales attribution. For decades, CPG brands advertised largely on faith, unable to tie a specific ad to a specific sale. Retail media changed that: because the ad and the purchase happen within the same first-party environment, brands can attribute sales to ads with a precision that open-web and even social advertising cannot match. This closed-loop attribution is the killer feature of retail media — and, as the insight above warns, also its biggest measurement trap, because closed-loop credit over-states incrementality.
The result is that retail media is both essential and easy to misuse. The networks are genuinely powerful conversion engines, but their power is concentrated at the bottom of the funnel, and their self-reported ROAS flatters that role. Choosing and allocating across Criteo, Amazon DSP, and Instacart well means understanding what each does, matching it to your category, and measuring all three against incremental contribution rather than dashboard ROAS. For the Criteo-versus-Amazon cut specifically, our Criteo retail media vs Amazon Ads comparison goes deeper on those two.
Criteo: the open retail media commerce network
Criteo's model is the open commerce media network — a single platform that connects brands to many retailers' inventory and shopper data, rather than tying a brand to one retailer's walled garden.
The core value proposition is breadth without operational overhead. Instead of negotiating, integrating, and managing separate relationships with dozens of individual retailer media networks, a brand can access multiple retailers' on-site inventory and commerce audiences through Criteo's platform. This is genuinely valuable for CPG brands sold across grocery, mass, drug, and specialty retailers, because it lets them advertise on the digital shelves of many retailers where their products sit, with consolidated buying, targeting, and reporting.
Criteo's capabilities span retail media on participating retailers' sites — sponsored products and on-site display close to the point of purchase — plus commerce audiences usable across the open web, drawing on Criteo's long history in commerce-data-driven retargeting. This gives brands both on-retailer capture and off-retailer reach built on commerce intent signals, all through one network.
The trade-offs follow from the open model. Because Criteo aggregates across many retailers, the depth of data and attribution quality on any given campaign depends on what each participating retailer shares — it is not the single, unified, deep dataset that a closed ecosystem like Amazon owns end to end. Reach and category fit also vary with the retailer mix in a campaign: a campaign weighted to grocery retailers behaves like grocery, one weighted to electronics behaves accordingly. Criteo's strength is therefore breadth and convenience across the fragmented retailer landscape; its constraint is that it does not own a single deep commerce ecosystem the way Amazon does.
A practical strength of Criteo's open model is operational efficiency for brands managing many retailer relationships. CPG brands often sell across dozens of retail banners, each with its own emerging media network, each demanding separate onboarding, creative specs, billing, and reporting. Managing that directly is a heavy operational burden that scales badly. Criteo's value is collapsing much of that into one platform, one set of creative workflows, and one reporting view, which frees a lean retail media team to focus on strategy rather than vendor administration. For a brand without the headcount to manage a dozen retailer networks individually, this consolidation is itself a major part of the value proposition, independent of any single campaign's performance.
The other consideration with Criteo is that its capabilities and the retailers available in its network vary by market and category, so a brand should validate that the specific retailers where its products sell are accessible through the network before assuming full coverage. Where the relevant retailers are present, Criteo delivers genuine multi-retailer reach; where a key retailer runs its own exclusive network outside Criteo, that retailer must still be handled separately. The honest framing is that Criteo covers a large and growing share of the fragmented retail landscape but not all of it, so it functions as the efficient backbone for multi-retailer reach rather than a guarantee of universal coverage.
For a CPG brand whose products live across many retailers and which wants reach beyond Amazon and a single grocery specialist, Criteo is the network that delivers that multi-retailer breadth through one operational relationship — which is exactly its role in a balanced retail media allocation.
Amazon DSP: the closed commerce ecosystem at scale
Amazon DSP is Amazon's demand-side platform — a closed but enormous ecosystem that lets brands buy display, video, and audio inventory programmatically, powered by Amazon's first-party commerce data, both on Amazon and across the web.
The defining asset is the data and scale. Amazon owns one of the richest purchase datasets in existence — what hundreds of millions of shoppers searched, viewed, and bought — and Amazon DSP lets brands target audiences built from those deterministic commerce signals. This is purchase-intent targeting of a quality open-web and even social platforms cannot match, applied across a vast inventory footprint: Amazon's own properties (the store, Prime Video, Fire devices, Twitch, and more) plus extensive off-Amazon inventory through the DSP. For a brand that wants the largest, deepest commerce audience available, Amazon DSP is the anchor.
Amazon DSP sits above the self-serve Sponsored ads suite (Sponsored Products, Brands, and Display, the last of which we cover in our Amazon Sponsored Display vs Google Discovery guide). Where Sponsored ads are bounded to Amazon's store and self-serve, the DSP extends programmatic reach off Amazon, supports richer audiences and creative, and is the tool for brands wanting to use Amazon's data to reach shoppers across the web, not just on Amazon. It also enables sophisticated retargeting — reaching shoppers who viewed but did not buy, cross-selling to past purchasers, and conquesting competitor audiences — all on deterministic Amazon data.
The trade-offs are the walled-garden ones. Amazon DSP is Amazon-centric: its data, attribution, and optimization all orbit Amazon's ecosystem, so its closed-loop attribution is deep within Amazon but more modeled off it, and it naturally favors brands with an Amazon sales presence. Access historically required managed service or a partner, though self-serve has broadened, making it more suited to mid-to-large brands or those working through an agency. And, like all retail media, its reported ROAS over-credits the bottom-funnel capture it excels at.
A point brands often underappreciate is that Amazon DSP's value compounds with an Amazon retail presence but is not strictly limited to it. A brand selling heavily on Amazon gets the fullest closed-loop benefit — ads, audiences, and purchases all in one loop. But even brands selling primarily through other channels can use Amazon's commerce audiences via the DSP to reach shoppers off Amazon, borrowing the data without the sales necessarily landing on Amazon. The attribution is weaker in that case, since the conversion happens outside Amazon's loop, but the targeting quality of Amazon's purchase-based audiences can still outperform open-web alternatives. This makes Amazon DSP relevant to a wider set of brands than the "only if you sell on Amazon" intuition suggests, though the brands that extract the most are those with meaningful Amazon sales.
The creative and audience sophistication of the DSP is also a step up from self-serve Sponsored ads. Brands can build layered audiences — combining purchase history, browsing behavior, lifestyle signals, and their own first-party data uploaded into Amazon's clean-room environment — and serve display, video, and audio creative across Amazon's properties and the open web. For sophisticated CPG advertisers, this audience and creative depth is a major part of the appeal, enabling strategies like sequenced messaging, suppression of recent purchasers, and cross-sell to complementary-category buyers that the simpler Sponsored formats cannot support. The trade-off is complexity: the DSP rewards skilled operation and punishes set-and-forget management, which is part of why managed service and agency partnership remain common.
For breadth, scale, and the deepest commerce data, Amazon DSP is the center of gravity in most large CPG retail media programs — the network the others complement rather than replace.
Instacart: the grocery-and-CPG specialist
Instacart's retail media network is the specialist of the three — a network built for online grocery and the CPG categories that fill grocery baskets, sitting precisely at the point of grocery purchase.
The defining strength is intent and category alignment. Instacart's shoppers are actively buying groceries — food, beverage, household, and personal care — when they encounter ads, which makes it an extraordinarily high-intent environment for the CPG brands in those categories. A sponsored product placement for a snack, a beverage, or a cleaning product reaches a shopper at the exact moment they are filling a basket with exactly those kinds of items. This point-of-purchase, in-category context is why Instacart converts so efficiently for grocery CPG and why it has become a must-have for brands in those categories.
Instacart's formats center on sponsored products that surface brands in grocery search and browse, plus display and other formats that influence basket composition and brand choice at the digital grocery shelf. The closed-loop attribution is clean and category-specific: because the ad and the grocery purchase happen on Instacart, brands get precise sales attribution for grocery conversions, with visibility into basket behavior and repeat purchase that is especially valuable for consumable CPG.
The trade-offs are scope. Instacart is grocery-and-CPG specialized — its reach and relevance are concentrated in those categories and in the online-grocery context, not the broad commerce span of Amazon or the multi-retailer breadth of Criteo. For a grocery CPG brand, that specialization is the whole point; for a brand outside food, beverage, and household, Instacart is a smaller or non-fit. Its AOV reflects grocery baskets — multiple items per order — which is ideal for brands wanting to influence basket composition and capture repeat grocery purchases, but is a different commerce profile from Amazon's wide AOV range or Criteo's retailer-dependent mix.
One underrated capability is Instacart's value for new-product launches and trial in grocery CPG. Because the platform sits at the exact moment of basket-building and can target by category, competitor, and purchase behavior, it is an efficient way to put a new SKU in front of shoppers already buying the adjacent category — exactly the audience most likely to try it. Sponsored product placements and sampling-style mechanics can drive trial that, for a consumable, converts into repeat purchase and lifetime value, which the platform's repeat-purchase visibility lets a brand actually measure. For a CPG brand launching into grocery, this combination of in-category targeting and measurable repeat behavior is hard to replicate on any other network, and it is one of the strongest arguments for a meaningful Instacart line in a grocery launch plan.
The flip side is that Instacart's economics and relevance live or die on grocery basket dynamics. A brand whose products do not naturally enter a grocery basket — durables, apparel, electronics — gets little from the platform, and even within CPG, the value concentrates in categories with frequent repeat purchase where lifetime value justifies acquisition cost. Brands should therefore weight Instacart by how central grocery is to their sales and how repeat-driven their category is, treating it as a high-conviction bet for grocery-staple CPG and a marginal or non-fit elsewhere. The specialization that makes it powerful for the right brands is exactly what makes it irrelevant for the wrong ones.
For grocery, food, beverage, and household CPG, Instacart is the network with the highest-intent, most category-aligned audience — the specialist that earns a heavy weight in those brands' allocations and little to none outside them.
Ecosystem reach and inventory compared
The three networks differ fundamentally in the shape of their reach and inventory, which shapes what each is good for.
Criteo's reach is wide and shallow by design — it spans many retailers' inventory, giving a brand presence across the fragmented retail landscape, but the data depth and attribution on any one campaign depend on the participating retailers. Its value is being everywhere your products are sold, through one platform.
Amazon DSP's reach is vast and deep but Amazon-anchored — the largest commerce audience and inventory footprint, on Amazon's many properties and across the open web via the DSP, all built on the single deepest commerce dataset. Its value is scale and data quality, with the caveat that everything orbits Amazon.
Instacart's reach is narrow and deep — concentrated in online grocery, with the highest intent and cleanest category data within that scope. Its value is precision and conversion within grocery CPG, with little relevance outside it.
The practical implication is that these reaches are complementary, not competitive. A grocery CPG brand reaches its category's highest-intent shoppers on Instacart, builds scale and uses the richest data on Amazon DSP, and extends across other retailers where it is sold via Criteo. Trying to make any one network do all three jobs leaves reach and efficiency on the table. The allocation question is not which network has the most reach in the abstract, but which combination of reaches matches where your category's shoppers actually buy.
The brands that get retail media right stop asking which network is best and start asking which network is best for which job. Instacart is unbeatable for grocery-basket conversion; Amazon DSP is unmatched for scale and commerce data; Criteo is the efficient way to be present across the long tail of retailers. The brands that struggle pick one network on the strength of its ROAS dashboard and try to run their whole retail media program through it — and either miss reach (if they pick a specialist) or pay for breadth they cannot use (if they pick the broadest). Match network to job, and the allocation almost designs itself.
Ad formats and where each network wins
The networks share a common format vocabulary — sponsored products, display, off-property extension — but each network's formats win in different contexts.
Sponsored products are the workhorse bottom-funnel format on all three, surfacing a brand's products in search and browse close to the point of purchase. They convert hardest because they meet the shopper at the moment of choice. Instacart's sponsored products win for grocery, surfacing brands as shoppers fill baskets; Amazon's win for the breadth of Amazon search; Criteo's win across whichever retailers a brand is sold on. For most brands, sponsored products are where retail media spend should start, because they deliver the cleanest closed-loop conversion data.
On-property display influences brand choice and basket composition within the network's environment — banners, featured placements, and category takeovers on the retailer's or platform's properties. Instacart display shapes grocery basket decisions; Amazon display spans its properties; Criteo display runs on participating retailers' sites. These formats sit slightly higher than sponsored products, building consideration at the digital shelf rather than capturing the exact-moment search.
Off-property and audience extension is where the networks extend reach beyond their core. Amazon DSP is strongest here, using Amazon's commerce data to reach shoppers across the open web with display, video, and audio. Criteo's commerce audiences likewise extend across the open web on commerce-intent signals. Instacart's extension is narrower, reflecting its specialist scope. These formats do the demand-creation and reach work that the sponsored-product capture layer cannot, and they are essential to a retail media program that wants to grow the category, not just harvest it.
Video and emerging formats — including Amazon's streaming and Prime Video inventory through the DSP — add upper-funnel brand-building within the commerce ecosystem, blending awareness with commerce data. These suit larger brands and launches.
The strategic pattern is to layer: sponsored products for capture and clean attribution, on-property display for shelf-level consideration, and off-property extension plus video for reach and demand creation. Each network contributes its strongest formats to the layer it does best — Instacart and Amazon sponsored products for capture, Amazon DSP and Criteo extension for reach — and the program balances capture against creation rather than over-indexing on the high-ROAS capture formats alone.
Attribution, AOV, and measurement quality
Attribution quality is retail media's headline selling point and its subtlest trap, and it differs in important ways across the three networks.
The shared strength is closed-loop attribution: because ad and purchase occur within the same first-party environment, all three networks can tie ads to actual sales far better than open-web or social advertising. This is real and valuable — it gives CPG brands sales-level measurement they historically lacked. But the completeness and the bias differ by network.
Amazon DSP offers deep closed-loop attribution within Amazon's ecosystem — precise on-Amazon sales tracking — plus modeled measurement for off-Amazon inventory. The depth is unmatched within Amazon; the limitation is that it is Amazon-centric, and off-Amazon attribution is more inferential. Instacart provides clean, category-specific closed-loop attribution for grocery purchases on its platform, with valuable basket and repeat-purchase visibility, bounded to the grocery context. Criteo provides sales attribution across its retailer network, with quality depending on what each participating retailer shares — broad but more variable than a single owned ecosystem.
The shared trap is over-credit. Closed-loop attribution within a walled garden over-states incrementality, because the network advertises to shoppers already in a buying environment and claims the resulting sales — many of which would have happened anyway. Each network's reported ROAS therefore flatters its bottom-funnel role. The discipline that addresses this is reconciliation and incrementality: sum each network's claimed sales and compare against total brand sales (the sum will exceed reality, revealing the over-credit), and run holdout tests to estimate true incremental contribution. This is the same logic our data-driven vs last-click attribution guide applies across paid media generally, and it applies acutely to retail media's flattering closed-loop numbers.
On AOV, the networks reflect their underlying commerce. Instacart skews to grocery-basket AOVs — multiple consumable items per order — ideal for influencing basket composition and capturing repeat purchases. Amazon DSP spans a vast AOV range from cheap consumables to high-ticket durables, fitting nearly any category but demanding targeting discipline. Criteo's AOV tracks the retailer mix in a given campaign. The practical move is to match network to your category's natural basket and price point, and to measure not just ROAS but new-to-brand contribution and incrementality, since a network that brings genuinely new customers or incremental sales is worth more than its raw ROAS suggests — and one that merely re-monetizes existing demand is worth less.
30-day plan to allocate retail media budget
The HowTo schema above is the day-by-day; here is the strategic framing for the four weeks.
Week 1 — Map fit and launch capture. Before allocating, map your category to each network's strength — grocery toward Instacart, breadth and scale toward Amazon DSP, multi-retailer reach toward Criteo — and define a distinct objective per network rather than one blended ROAS goal. Then launch the highest-intent capture formats first: sponsored products and on-platform display on all three, targeted to your categories and competitors. This establishes a baseline of clean, closed-loop sales data per network before you add costlier reach formats, and it gets the fastest-converting spend working immediately.
Week 2 — Add reach and stand up measurement. Layer in off-platform and audience-extension formats — Amazon DSP off-Amazon, Criteo commerce audiences, Instacart extension — kept in separate campaigns so capture and reach never blur in reporting. Then build the measurement that matters: connect each network's reporting and reconcile claimed sales against total sales, because the sum of three walled gardens' claims will exceed your real sales and expose the over-credit. Set up an incrementality holdout on your largest spender.
Week 3 — Adjust for incrementality and decide. Convert each network's flattering reported ROAS into an incrementality-adjusted figure using the holdout read, and compare adjusted ROAS, AOV, and new-to-brand contribution across the three. Expect the highest-reported-ROAS network to be doing the most harvesting, and a lower-reported network to be contributing more incremental sales. Set the allocation on adjusted contribution and category fit — Instacart heavy for grocery CPG, Amazon DSP for scale, Criteo for multi-retailer reach — while holding a floor on the reach-and-creation formats so you do not over-fund pure capture.
Week 4 — Institutionalize. Document the split, predicted incremental sales, and assumptions so the allocation is defensible to finance, then codify a monthly review (incrementality-adjusted ROAS, claimed-versus-total reconciliation, new-to-brand per network) and a quarterly incrementality re-test rotating which network you hold out. Align the retail media calendar to your category's seasonal and promotional cycles, and write the playbook so the multi-network program runs as one coordinated system rather than three disconnected campaigns.
The long-term posture is to treat Criteo, Amazon DSP, and Instacart as a complementary portfolio — open network, closed ecosystem, and category specialist — each carrying the job it does best, all measured against incremental contribution rather than self-reported ROAS. Retail media is the conversion layer closest to the transaction, but it converts demand that upper-funnel channels create; fund the capture and the creation deliberately, measure honestly across the walled gardens, and the portfolio compounds rather than cannibalizes. For the cross-channel view of how this fits with demand creation, our budget allocation framework covers the same capture-versus-creation discipline applied to Meta and Google.
If you also run Google and Microsoft Search to capture and create the demand that feeds your retail media conversion layer, and want that side of your stack optimized by AI so your team can focus on retail media strategy and merchandising, SteerAds runs a free 14-day audit on your accounts.
Sources
Official and third-party sources consulted for this guide:
-
criteo.com
— Criteo Commerce Media network, formats, and retailer partnerships -
advertising.amazon.com
— Amazon DSP documentation, audiences, and attribution -
instacart.com/company/ads
— Instacart Ads formats and closed-loop measurement -
emarketer.com
— retail media spend, network share, and CPG trends 2025-2026 -
iab.com
— retail media measurement standards and attribution guidance
Related reading: AI Creative with Veo 3, Runway & Flux for Google Ads 2026 · Answer Engine Optimization (AEO) for SaaS Vendors 2026 · CTV / Connected TV Ads: SMB Buyer's Guide 2026 · DV360 Setup Checklist: First 90 Days 2026 · GA4 Explorations: Cohort Analysis for Paid Acquisition 2026 · GTM Server Container on Cloud Run: Setup & Cost 2026
FAQ
What is a retail media network, and why did it become so important by 2026?
A retail media network is an advertising business built on a retailer's or commerce platform's first-party shopper data and inventory — letting brands advertise to shoppers on and around the retailer's properties, attributed to actual sales. It became central by 2026 for three reasons: third-party cookie deprecation made retailers' deterministic, logged-in purchase data far more valuable than open-web behavioral data; retailers discovered retail media as a high-margin revenue stream and invested heavily in it; and brands gained closed-loop sales attribution they could not get elsewhere. Criteo, Amazon DSP, and Instacart represent three distinct models of this — an open multi-retailer network, a closed at-scale ecosystem, and a category specialist — which is why comparing them is the practical question for any CPG or consumer brand planning retail media spend.
How do Criteo, Amazon DSP, and Instacart differ at the highest level?
Criteo is an open commerce media network that connects brands to many retailers' inventory and data through one platform, giving multi-retailer reach without managing each retailer separately. Amazon DSP is Amazon's demand-side platform, a closed but enormous ecosystem with unmatched first-party purchase data, on- and off-Amazon inventory, and deep commerce signals, best for brands selling on or adjacent to Amazon. Instacart is a grocery-and-CPG specialist, a retail media network for online grocery with high purchase intent and category depth in food, beverage, and household. In short: Criteo for breadth across retailers, Amazon DSP for scale and depth in the largest commerce ecosystem, Instacart for grocery and CPG precision. Most large CPG brands use all three for their distinct strengths.
Which retail media network is best for a CPG brand?
Most large CPG brands use a combination, but the weighting depends on category and where the brand sells. For grocery, food, beverage, and household CPG, Instacart is uniquely strong because its audience is shopping those exact categories with high intent, and its sponsored product and display formats sit right at the point of grocery purchase. For brands with significant Amazon sales or wanting the largest commerce audience and richest data, Amazon DSP is the anchor. For brands wanting to reach shoppers across many retailers — including grocery, mass, and specialty — without building separate relationships with each, Criteo provides that breadth. A typical large CPG allocation anchors on Amazon DSP for scale, adds Instacart for grocery-specific category depth, and uses Criteo for multi-retailer reach beyond those two.
How does attribution differ across the three networks?
All three offer better attribution than open-web advertising because they tie ads to first-party shopper data and, often, to actual purchases — but they differ in completeness. Amazon DSP offers deep closed-loop attribution within Amazon's ecosystem plus modeled off-Amazon measurement, strong but Amazon-centric. Instacart provides closed-loop sales attribution for grocery purchases on its platform, clean and category-specific. Criteo provides sales attribution across its retailer network, with quality depending on the data each participating retailer shares. The common caveat is that closed-loop attribution within a walled garden tends to over-credit the network for sales that would have happened anyway, so brands running multiple networks should reconcile each network's claimed sales against total sales and, ideally, run incrementality tests rather than summing dashboards.
What AOV ranges and category fits should I expect on each?
AOV and category fit are shaped by each network's underlying commerce. Instacart skews to grocery-basket AOVs — multiple items per order across food, beverage, and household — making it ideal for CPG brands wanting to influence basket composition and capture repeat grocery purchases. Amazon DSP spans a vast AOV range from low-cost consumables to high-ticket electronics and durables, reflecting Amazon's breadth, so it fits nearly any category but requires targeting discipline. Criteo's AOV and category fit vary by the retailers in your campaign — a campaign weighted to grocery retailers behaves like grocery, one weighted to electronics or fashion retailers behaves accordingly. The practical implication is to match the network to your category's natural basket and price point: Instacart for grocery baskets, Amazon DSP for breadth, Criteo for whichever retailer mix fits your category.
Do I need a large budget to use retail media networks?
It varies by network and access route. Instacart and Criteo offer self-serve and managed options that accommodate a range of budgets, with sponsored product formats accessible to smaller brands and the more advanced display and off-platform formats suiting larger spend. Amazon DSP historically required either a managed-service minimum or working through a partner, making it more accessible to mid-to-large brands or those using an agency, though self-serve access has broadened. The honest framing is that meaningful retail media programs reward scale — first-party data and closed-loop attribution shine when there is enough volume to measure — so while smaller brands can start with sponsored products on a single network, the full multi-network strategy described here is built for brands with real category presence and budget to allocate across networks.
How does retail media fit alongside my Google and Meta advertising?
Retail media complements rather than replaces upper-funnel demand creation on Google and Meta. Google and Meta build awareness and demand and reach shoppers before they are in a buying environment; retail media networks capture and convert that demand at or near the point of purchase, with closed-loop sales attribution. The strongest CPG strategies treat them as a funnel: Google and Meta create and warm demand, retail media converts it on the shelf — physical or digital. The risk is double-counting, since each platform claims credit for sales, so brands should measure incrementality across the whole stack rather than allocating on each platform's self-reported ROAS. Retail media is the conversion layer closest to the transaction; demand creation still has to feed it.
What is the biggest mistake brands make with retail media networks in 2026?
The biggest mistake is allocating by each network's self-reported ROAS, which structurally over-credits the bottom-funnel capture these networks excel at. Because retail media advertises to shoppers already in or near a purchase, and attributes the resulting sales to itself, its reported ROAS looks spectacular — but much of it is harvesting demand that already existed rather than creating incremental sales. Brands that pour budget toward whatever network reports the highest ROAS end up over-funding capture and starving the demand creation that feeds it, then wonder why total sales plateau even as retail media ROAS stays high. The fix is the same as across all paid media: judge networks on incremental contribution, reconcile claimed sales against total sales, and fund demand creation as deliberately as capture.