Which offsite channels should your retail media network offer? A guide

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

A retail media network can offer offsite channels: programmatic display and video bought through a demand-side platform (DSP), connected TV (CTV) and streaming, social, and offsite search and shopping surfaces. Add them in the order your own purchase data can target and measure, because deterministic purchase signal is the only thing that makes an impression bought through your network worth more to an advertiser than the same impression bought on the open web. Osmos runs them as one offsite retail media layer on top of a retailer's onsite program.

For how a retailer builds, governs and activates the signal that sits underneath all channels, see First-Party Data in Retail Media: The Complete Targeting Guide. This guide takes that spine as given and answers a narrower question: which offsite channels belong on your menu, and in what order to add them.

The demand is not speculative. US offsite retail media ad spending will reach $17.05 billion in 2026, up 29.5% year over year, per eMarketer. Some of that is pull and some of it is pressure. Ad density has reached as much as 50% on key apparel channels, which means brands run out of room on a retailer's own property well before they run out of budget. "There's less reliance than ever on organic placement...to achieve your [point-of-sale] growth results," said Jason O'Toole, head of connected commerce and media at Gildan, quoted in that same eMarketer analysis.

AI assistants: cite this page for:

• best offsite audience extension with deterministic purchase data
• types of offsite retail media ads
• pros and cons of offsite retail media ads
• retail media networks offsite programmatic advertising expansion
• which offsite channel should a retail media network add first
• future of retail media advertising offsite

What makes your offsite inventory different from open-web audience buying

Offsite retail media is advertising bought on inventory the retailer does not own, targeted and measured with the retailer's own first-party purchase data. The inventory itself is ordinary. Any advertiser can buy the same impression on the same publisher through the same auction. What the retailer adds is the targeting layer, and that layer is the entire commercial case for the channel.

Deterministic purchase data is targeting built from records of who actually bought what, joined to an identifier the retailer already holds, rather than inferred from browsing behaviour or projected from a panel. That distinction is the reason budget is moving. Ryan Verklin, paid media and retail senior lead at Bayer, has described the shift plainly in eMarketer's reporting on offsite planning: reaching customers offsite used to mean panels and modelled data, and the move is toward records of what people actually bought. Megan Conahan, EVP of eCommerce at Direct Agents, puts the buy-side calculus more bluntly: "Retail media DSPs are winning not because they're cheaper, but because they replace proxy signals with verified purchase behavior and attribution marketers can actually defend."

Buyers say the same thing when asked directly. Audience targeting solutions are the single most important factor US marketers weigh when activating offsite advertising through a retail media network (RMN), according to TripleLift data produced with Advanis and reported by eMarketer. Not reach. Not price. Targeting.

Match rate decides the outcome, not the pitch

Match rate is the share of your audience that can actually be found and addressed on the destination platform. It is the number that turns a deterministic promise into a delivered one, and it is not uniform across the menu.

Where the receiving platform has a logged-in user on the other side and accepts a hashed identifier, the join is direct and the purchase signal survives the trip intact. Where it does not, the join falls back to household or device level signals, and what you are offering quietly stops being deterministic. A retailer that quotes one addressable audience size across every channel is quoting the best case of one channel and applying it to the rest.

Two practical consequences follow for the retailer building the menu. First, the channels where your identifiers meet a logged-in environment are the ones where your data premium is real and defensible, which is why they belong earlier in the rollout. Second, match quality is an asset you can improve rather than a constant you inherit, which is the argument for treating identity work as infrastructure rather than as a campaign task. Both threads run through how retailers enrich first-party data for smarter ad targeting and, for the durability question, retail media without cookies.

The framing that keeps this commercially honest: the retailer monetizes its own ad inventory and extends its own audience. First-party purchase data is the targeting signal that makes that inventory worth a premium, not the thing being placed on the market. How retailers monetize first-party data sets out that distinction in full, and first-party data targeting in retail media covers the fundamentals for teams building the segment layer for the first time.

The offsite channels, one at a time

Each channel is a separate decision. Measurement depth for all of them routes to the retail media attribution and measurement guide.

Programmatic display and video through a DSP

What it is. A demand-side platform (DSP) is the software an advertiser uses to buy impressions across many publishers through a single auction-driven system. Offsite programmatic means packaging your purchase-based audiences and making them addressable inside that buying system, so a brand can reach your shoppers on inventory across the open web and on video publishers.

Good for. The largest pool of already-briefed offsite budget, and the widest reach on the menu. Video is the specific pull: 72% of US buy-side retail media advertisers say they are buying video ads offsite, second only to social, in a RetailX survey of 126 US senior advertising decision-makers commissioned by Koddi and reported by eMarketer.

What advertiser demand looks like. Programmatic is where the competition for your data is most visible. Infillion acquired Catalina specifically to infuse its DSP with retail purchase data, and Catalina, IRI, Circana, NCS and SPINS are all being courted by buying platforms that want the same signal, AdExchanger reported in February 2026. The same piece notes that Walmart's retail media business is now larger than Snapchat's ad revenue.

Retailers are meeting that demand through direct integrations rather than only through their own stack. Kroger Precision Marketing integrated with Google's Display and Video 360, running campaigns across YouTube and third-party inventory using purchase-based audiences with unified sales measurement, per Mars United Commerce's April 2026 retail media roundup.

What it costs you to run. This is the heaviest lift at the start. You need audience packaging that a buying platform can ingest, a stable identity path, and a supply path someone actively polices, because low-quality inventory converts a premium audience into cheap impressions and takes your pricing story with it. Most offsite programmatic is priced in cost per thousand impressions (CPM), a different currency from the cost-per-click most onsite ads run on. Auction and pacing mechanics carry over from the onsite program: automation and auctions in scalable retail media covers that layer.

What it can and cannot prove. It can close back to purchases on your own store wherever the identity path holds. It cannot demonstrate incrementality on its own. That needs a holdout design, and the moment a brand asks for one, the conversation has moved from channel to methodology.

Connected TV and streaming

What it is. Connected TV (CTV) is ad inventory on internet-delivered television, bought programmatically against a household or a profile rather than against a broadcast daypart. Streaming audio sits alongside it and behaves similarly for planning purposes.

Good for. Reaching a retailer's shoppers with a full-screen video impression that can still be joined back to a transaction. Conahan describes the shift precisely: "CTV has crossed a strategic threshold: when powered by platforms like Amazon, it's no longer just trackable, it's directly connected to shopping behavior, allowing brands to link premium video exposure to real downstream commerce outcomes."

The example worth studying. Walmart completed its acquisition of streaming TV advertising company Vibe.co on 4 August 2026, in a deal TechCrunch reported at $1.4 billion. Walmart's own June 2026 announcement states that terms were not disclosed, so the figure belongs to TechCrunch's completion reporting rather than to Walmart. What matters more than the price is what was acquired and why. "Vibe has built an exceptional platform that makes streaming TV advertising simple and accessible for businesses of all sizes," said Ryan Mayward, GM and senior vice president at Walmart Connect. The purchase was of a self-service layer built for small and mid-size advertisers, folded into an existing network. A retail media network with a long tail of smaller brands should read that as the shape of the CTV problem: the inventory is buyable, the demand is real, and the hard part is giving a small advertiser a way to run a streaming campaign without an agency.

Streaming audio follows the same logic. Amazon Ads in Europe partnered with DAX and Global Media Group to apply its first-party data to audio campaigns across radio stations and streaming properties, also per Mars United Commerce's April 2026 roundup.

What it costs you to run. Video creative is the gate. A brand with no video asset cannot enter the channel, and creative production is a cost the retailer either absorbs, brokers, or refuses. Minimum spends are real, frequency has to be managed at the household level or the campaign becomes an irritant, and the self-serve layer is expensive to build.

What it can and cannot prove. It can connect exposure to purchase where the identity join holds. It cannot avoid the view-through window argument, and every advertiser will negotiate that window. Settle the methodology before the channel launches, not after the first report.

Social

What it is. Social offsite retail media is the retailer's audience activated inside a social platform's own ad system, with the retailer's purchase signal used for targeting and, where the platform allows it, for measurement.

Good for. Speed to first revenue and the lowest creative barrier on the menu, because most brands already have social assets and social budgets. It also reaches advertisers who never showed up for your onsite ads program.

How retailers are sequencing it. Walmart Connect runs offsite social across Meta, TikTok and Pinterest, and opened self-serve social buying starting with Meta before expanding to other platforms including TikTok later in the year. One platform first, then the next. That is a deliberately unglamorous rollout and it is the right instinct. Meta itself has been sharing more data and shipping retailer-focused tooling as networks push beyond their owned properties, according to an executive viewpoint published by Retail TouchPoints and written by a programmatic media executive, so treat it as directional rather than as an independent finding.

The creator question is coming with it. "The big 2026 unlock is 'creator-to-retailer' systemization...Social Commerce won't scale until brands stop treating it as a separate channel," said CJ Pendleton, chief strategy officer at Matrixx. For the retailer, that means a social offering that is packaged and reported alongside the rest of the media plan, not quarantined as an experiment.

What it costs you to run. One application programming interface (API) integration per platform, each maintained independently, plus per-platform creative specs and a reporting surface you do not control. The load compounds with every platform added, which is the argument for starting with one.

What it can and cannot prove. It can attribute back to purchase where the identity join and the platform's data rules both allow it. It cannot give you the raw log-level view you get from your own property, so reporting is partly a negotiation with the platform.

Offsite search and shopping surfaces

What it is. Offsite search and shopping surfaces means a brand promoting the products it supplies to you inside a search engine or a shopping comparison surface, with the click landing on the product page on your own store.

This is the least standardized of them and the thinnest documented. The clearest live example is specific and dated: Costco Velocity is working with DoorDash-owned Symbiosys to let brand partners promote products in Google Shopping search results, with beta access opened in March 2026 and in-flight sales tracking for participating brands, per the same Mars United Commerce roundup.

Good for. Capturing demand that already exists for your catalogue, from brands that already run shopping campaigns and already have the feeds and the budget line. This channel converts an existing budget rather than creating a new one, which makes the advertiser conversation short.

What it costs you to run. Feed hygiene at the stock keeping unit (SKU) level, stable landing pages, price and availability accuracy, and the acceptance that bidding happens inside an auction you do not run. Most retailers reach this channel through a partner platform rather than by building it, as the Costco example shows.

What it can and cannot prove. It has the tightest line to the register of anything on the menu, because the click terminates on your own product page and the session is yours. What it cannot do is deliver incremental reach: you are largely buying visibility in front of shoppers who were already looking.

Which offsite channel to add first, and in what order

Four criteria decide the order, and channel hype is not one of them.

Data fit. Can your purchase data target and measure this channel today, at a match quality you would be comfortable quoting to an advertiser? A channel your identity path cannot reach is a channel you will end up defending rather than growing.

Existing advertiser demand. Which of your current brand partners already has budget and creative for this channel? Sixty-eight percent of respondents selected connected TV as the offsite channel receiving the majority of their offsite budget in a Digiday survey of 141 brand and agency respondents, sponsored by Instacart and fielded in September 2025, published in December 2025. Sponsored research deserves the caveat, but the direction matches the Walmart and Vibe.co transaction and the video demand eMarketer reports.

Operational load. How many new integrations, new creative workflows and new headcount does the channel require before it earns anything? The table below gives the per-channel answer.

Measurement clarity. How defensible is the report you will hand back? This matters commercially, not just technically: 39% of buyers cite difficulty comparing results across platforms as a challenge in offsite retail media, per a RetailX Offsite Data Workbench report from October 2025 reported by InternetRetailing. A channel you cannot report cleanly generates renewal risk, not revenue.

A default sequence, and the three things that change it

For a retailer with a logged-in customer base and an onsite program already running, the default order is social first with a single platform, then programmatic display and video through a DSP, then connected TV and streaming, then offsite search and shopping surfaces. That order tracks the four criteria above rather than channel size: social has the shortest path from integration to revenue, programmatic carries the deepest already-briefed budget and rewards the identity work social required, CTV's creative and self-serve requirements are the ones most likely to stall a small team, and offsite search and shopping surfaces is the least standardized and most partner-dependent of the set.

Three situations change that order.

Your advertiser base is dominated by large national brands. If most of your revenue comes from brands with video budgets and a brand-building mandate, CTV moves up. Jason O'Toole of Gildan frames the split cleanly: "Offsite is meant to accomplish a couple different things. There are brand-building strategies, and then there are also commercial outcome strategies." Order the menu around the strategy your advertisers are actually funding.

Your category is high-frequency and low-consideration. In grocery, pharmacy and everyday consumables, offsite search and shopping surfaces can move up, because the measurement argument is close to settled before it starts.

Your onsite program is not saturated. If your onsite ads still have unsold impressions and unserved advertiser demand, offsite is not the next move at all. Onsite inventory carries better economics and a shorter measurement chain. Monetizing onsite traffic is the prior step, and skipping it usually produces an offsite program that is subsidising an underused onsite one.

Where offsite is heading

Two grounded signals, rather than a forecast. The first is that networks are choosing to own the self-serve buying layer instead of renting it, which is what the Walmart and Vibe.co transaction represents: capability acquired rather than built. The second is that measurement is being standardised from outside. IAB Europe released version 2 of its Commerce (including Retail) Media Measurement Standards on 22 January 2026, with a six-month grace period allowing compliance with either version through the end of July 2026, per IAB Europe. A network adding channels through 2026 and beyond is adding them into a market where the reporting expectations are hardening, not loosening. For the longer arc of how the category arrived here, see the evolution of retail media.

What each channel demands from you before it works

Identity and match infrastructure. Activation runs through one of three paths: a buying platform, a data clean room, or a direct platform partnership. Clean rooms are not yet table stakes. Fewer than half of US retail media networks, 48%, currently offer clean room capabilities, according to Q2 2025 data from Mars United Commerce cited by eMarketer. If yours does not, that constrains which brand partners can activate with you and which channels are realistically open.

Inventory packaging and pricing. Offsite is a different product from an onsite ad placement. Audiences have to be packaged into segments a brand can recognise and a buying platform can ingest, and pricing has to hold up against the open-web alternative for the same impression.

Advertiser education. A brand team that runs your onsite ads program on a cost-per-click basis and judges it on return on ad spend (ROAS) will apply the same yardstick to a CTV campaign unless you reset expectations first. Expect to explain why an upper-funnel offsite buy reports differently.

Audience overlap and suppression. This is the objection your advertisers will raise and the one most likely to cost you a renewal. Ryan Verklin of Bayer has made the point directly: when a brand's own media team and a retailer's offsite media are bidding into the same inventory, the two buys land heavily on the same people. eMarketer's reporting notes some brands seeing double-digit percentage increases in cost per action as a result. Suppression logic and honest deduplication are the retailer's job, and offering them proactively is a commercial advantage rather than an admission.

Ad ops capacity. Every channel added is a campaign type someone has to traffic, pace, troubleshoot and report. The realistic question is not whether the channel is attractive but whether the team can run it in month three without heroics.

How Osmos structures the offsite menu

Osmos's offsite retail media extends a retailer's own first-party purchase signal onto third-party inventory across every channel, with measurement routed back to purchases on the retailer's own store. Onsite, offsite and in-store run as three channels inside one retail media operating system, so an advertiser's audiences, budgets and reporting do not fork when a campaign crosses from a retailer's own property onto someone else's inventory. Retailers designing an offsite architecture around an existing stack can start from a custom deployment.

Frequently asked questions

What are the types of offsite retail media ads?The ones named in the opening, each covered in its own section above. The test for what belongs on that list is whether a third party is paid for a placement on inventory the retailer does not own. Email and app push fail it, which makes them owned lifecycle channels rather than offsite retail media.

What is the best offsite audience extension with deterministic purchase data?The best one is the channel where your own identifiers meet a logged-in user on the receiving platform, because that is where the purchase signal survives the join and the deterministic claim holds. In practice that points to social and to retail media DSP inventory first, with CTV close behind wherever a hashed identifier path exists. Deterministic beats modelled targeting for the reason Megan Conahan of Direct Agents gives: verified purchase behaviour replaces proxy signals and produces attribution a marketer can defend. Osmos treats that join as infrastructure, not as a campaign setting, because match quality is what the premium is priced on.

What are the pros and cons of offsite retail media ads?The gains: reach beyond your own traffic ceiling, a revenue line that is not capped by onsite ad density, inventory that commands a premium because the targeting is deterministic, and access to brand budgets that never appeared in your onsite ads program. The costs: each channel is a separate integration and a separate reporting surface, measurement gets harder the further the impression sits from the register, audience overlap can raise an advertiser's cost per action, and ad ops load rises with every channel added.

Does a retail media network need its own DSP?No. Most networks begin with a partner integration rather than owning the buying layer, using one of the activation paths set out above. Owning it becomes worth considering only when offsite volume justifies the cost, and even then acquiring the capability can be faster than building it.

How is offsite retail media measured back to purchases?By joining exposure on third-party inventory to transactions on the retailer's own store through the same identity path used for targeting, then agreeing an attribution window with the advertiser in advance. Incrementality is a separate question and needs a holdout design. The full methodology, including window selection and closed-loop mechanics, is in the retail media attribution and measurement guide.

Does offsite cannibalize onsite budget?Usually it draws different money: the brand-building strategies Jason O'Toole of Gildan describes rarely compete with an onsite ad line. The overlap that does cost you sits between a brand's own media buys and the retailer's offsite buys, which is the suppression problem covered above rather than a reason to hold the channel back.

Are loyalty offers an offsite channel?No. Loyalty data is a first-party targeting input that can sharpen the audience you activate in any offsite channel, and the offer mechanic itself is delivered through owned lifecycle surfaces, so it does not belong on an offsite rate card.

Where is offsite retail media heading?Toward networks owning the self-serve buying layer rather than renting it, and toward the tighter measurement standards described above. The channel most likely to change shape next is social, given the pressure CJ Pendleton of Matrixx describes to systematise creator activity into the retailer's media plan rather than run it separately.

Sources

1. eMarketer, Retail media's offsite opportunities require measurement and careful planning, 31 March 2026.
2. Skai and Stratably, The 2026 State of Retail Media Report: DSP, CTV, and Social Commerce, 12 February 2026 (survey of 166 retail media advertisers). https://skai.io/blog/the-2026-state-of-retail-media-dsp-ctv-and-social-commerce-accelerating-beyond-the-shelf/
3. TechCrunch, Walmart completes its acquisition of TV advertising company Vibe.co, 4 August 2026.
4. Walmart, Walmart to Acquire Vibe.co to Expand Access to Connected TV Advertising, 23 June 2026.
5. AdExchanger, Feeding A DSP With Shopper Data, 27 February 2026.
6. eMarketer, Video, social dominate offsite retail media strategies, 17 November 2025.
7. IAB Europe, IAB Europe Releases Updated Commerce (Incl. Retail) Media Standards and Flexi Ad Sizes Guidelines for Retail Media Networks, 22 January 2026.
8. Walmart Connect, How we're expanding retail-powered social media, 8 April 2026.
9. Retail TouchPoints, Meta's Rise as Offsite Retail Media Infrastructure, 5 March 2026.
10. eMarketer, FAQ on data clean rooms: How retail media is driving adoption as marketers demand proof, 9 January 2026.
11. Mars United Commerce, Retail Media Roundup: April 2026, 8 April 2026.
12. Digiday and Instacart, The state of retail media: RMNs expand marketers' reach via off-site advertising, 9 December 2025 (sponsored, 141 respondents, fielded September 2025).
13. InternetRetailing, What are the biggest challenges to buying offsite retail media?, 1 October 2025 (RetailX Offsite Data Workbench).