Luxury retail media networks are the newest line item on the premium brand media plan, and they are being bought with far less scrutiny than they deserve. The pitch is genuinely compelling: a department store or luxury e-commerce platform offers deterministic, purchase-verified first-party data on shoppers who have spent five figures with them, and lets a brand target that audience on-site and off. For a category that has spent two decades struggling to find affluent consumers at scale, that sounds like the end of the search.
It is not. It is a new inventory source with a specific and narrow set of jobs it does well, priced at a premium that is sometimes justified and frequently is not. This guide covers what luxury retail media networks actually deliver, the CPM and ROAS ranges we see in market, how to structure a first test, and the three measurement failures that make retail media look better on a dashboard than it performs in the P&L.
What a Luxury Retail Media Network Actually Sells
A retail media network monetizes three distinct assets, and they are not equally valuable.
The surface. Sponsored product placements, on-site display, category takeovers, and search results on the retailer's own property. This is the retailer's owned inventory. It is finite, it is the highest-margin product they sell — onsite formats carry 70–90% gross margins for the retailer because they already own the pixel — and it is where the shopper is closest to a transaction.
The data. The retailer's purchase history, browse behavior, and identity graph, activated against inventory elsewhere. This is the "offsite" business: the retailer's segments pushed into a DSP and run against programmatic display, CTV, or online video. Retailer margins here run 20–40% because they are buying third-party media and reselling it.
The measurement. Closed-loop sales attribution tying exposure to a transaction inside the retailer's own environment. This is the part most brands are actually paying for, and it is the part with the most fine print.
Saks launched the first significant luxury-category network in this model, drawing on roughly 435 million annual site visits and initially offering sponsored product and on-site display to brand partners including Chanel Beauty, Dolce & Gabbana, and David Yurman. The consolidation of Saks Fifth Avenue, Neiman Marcus, and Bergdorf Goodman under a single corporate parent created something new in the category: a genuinely large, genuinely affluent, single-source shopper graph. Nordstrom, Net-a-Porter, and a growing set of category specialists in watches, jewelry, and travel have followed.
Where Luxury Retail Media Differs From Mass Retail Media
Buyers who cut their teeth on Amazon Ads or Walmart Connect tend to import assumptions that do not survive contact with the luxury environment.
Basket frequency is an order of magnitude lower. A grocery network sees the same household 40+ times a year. A luxury department store may see a high-value customer four to eight times annually. Every model that depends on repeat exposure and rapid conversion signal — bid optimization, incrementality readouts, budget pacing — runs on far thinner data. Expect longer learning periods and wider confidence intervals.
Scale is the binding constraint, not targeting precision. The addressable on-site audience for a single luxury retailer's high-spend tier is often in the low millions or lower. That is enough for a meaningful test and rarely enough to carry a national brand campaign. Brands that shift meaningful budget into on-site luxury retail media hit frequency ceilings inside of three weeks.
The halo is real and largely unmeasured. Luxury purchases migrate between channels constantly — discovered at the retailer, purchased at the brand boutique, or the reverse. Retail media's closed-loop measurement captures only the transaction that happens inside the retailer. For a brand with meaningful DTC or boutique revenue, that is a systematic undercount, and it is the single strongest argument for buying retail media that most brands never make.
Retail Media CPM and ROAS Benchmarks
These are the working ranges we underwrite against in 2026. Luxury networks generally price at the top of each band or above it, on the argument that the audience justifies the premium.
| Format | Typical CPM Range | Typical ROAS Range | Best Use |
|---|---|---|---|
| Onsite sponsored product | $8–$25 (often CPC-priced) | 2.5x–6.0x | Capturing existing category demand |
| Onsite display / category takeover | $5–$25 | 1.8x–4.0x | Brand presence at the point of consideration |
| Offsite programmatic display | $25–$60 | 1.5x–3.5x | Extending retailer segments to open web |
| Offsite CTV / online video | $28–$65 | 2.0x–4.5x | Upper-funnel reach against verified spenders |
| In-store digital screens | $10–$50 | Not reliably measurable | Flagship and event support |
Two things to hold onto. First, the ROAS figures are almost always retailer-attributed, not incremental — more on that below. Second, offsite retail media CPMs of $25–$60 mean a brand is paying a $15–$45 data premium over what the same impression costs through a standard premium programmatic buy. That premium is defensible when the segment is genuinely unavailable elsewhere. It is not defensible when the retailer is selling a "luxury shopper" segment that is functionally a $200K+ household income model available from three data providers at a fraction of the cost.
Average click-through rates on offsite retail media units run 0.15%–0.4%, which is broadly in line with standard programmatic display and should not be treated as evidence of superior audience quality.
Onsite vs. Offsite: Choosing the Right Entry Point
The most common structuring mistake is buying both at once, in equal weight, and being unable to tell which one worked.
Start onsite if the brand already sells through the retailer and the objective is share of category. Onsite is the closest thing to a controlled environment retail media offers: the audience is in-market, the competitive set is visible, and conversion is measured natively. Budget floors are usually modest — six-figure annual commitments are common, and several networks will test at less.
Start offsite if the objective is reaching the retailer's high-spend customers with a brand message they will not encounter on a product page. This is where luxury retail media gets genuinely interesting, because a verified $25,000-annual-spend customer is a far better proxy for high-net-worth targeting than any modeled wealth segment. It is also where the measurement discipline has to be strongest, because the retailer's attribution window is now being applied to media that ran nowhere near their site.
Buy both only when the brand has enough volume to run them as separate, independently measured lines with distinct budgets, distinct creative, and a holdout structure that can separate them.
The Three Measurement Failures
1. Retailer-attributed ROAS is not incremental ROAS
Every retail media network reports on its own attribution model, typically a 14-day view-through and 14- or 30-day click-through window inside its own environment. That model will credit the network for purchases from customers who were going to buy regardless — and in luxury, where brand demand is pre-existing and strong, that share is large. Across the retail media tests we have run for premium clients, retailer-attributed ROAS has typically overstated true incremental lift by a factor of 1.8x to 3.5x. A reported 4.2x ROAS is often a 1.4x–2.3x incremental reality.
The fix is not complicated, but it must be negotiated before the insertion order is signed: a geographic or audience holdout, agreed in advance, with the retailer suppressing the holdout group. Networks that will not support a holdout are telling you something.
2. The comparison set is wrong
Brands routinely compare retail media ROAS against their branded search ROAS or their overall blended return, both of which are contaminated by existing demand. The correct comparison is against the incremental return of the next-best alternative use of that dollar — usually premium CTV or a direct publisher deal against the same audience. When we run that comparison honestly, offsite luxury retail media wins roughly a third of the time, and it wins on audience verification rather than on cost.
3. Cross-channel purchase leakage runs one direction
Retail media undercounts because purchases that migrate to the brand's own channels are invisible to the retailer. This cuts in the brand's favor and is worth quantifying — a matched-market test or a clean-room analysis comparing exposed and unexposed cohorts across both the retailer's and the brand's transaction data will size it. Brands that do this work often find the retailer's numbers were conservative on total revenue impact even while being generous on incrementality.
How to Structure a First Luxury Retail Media Test
A defensible first test takes one quarter and follows six steps.
- Define the audience you cannot otherwise buy. Write down the specific segment — "customers with $15,000+ trailing-twelve-month spend in fine jewelry" — and ask your data partners whether they can approximate it. If they can within 20%, the retail media premium is probably not worth paying.
- Negotiate measurement before rate. Holdout support, raw log-level or clean-room data access, agreed attribution windows, and the right to run third-party verification. Rate concessions are easy to get later; measurement rights are not.
- Cap the test at 8–12% of quarterly working media. Enough to read, small enough that a null result is survivable.
- Run one format, one audience, one creative concept. Multivariate tests on thin luxury data volumes produce noise.
- Set a minimum detectable effect in advance. With low basket frequency, many luxury retail media tests are statistically incapable of detecting the lift they are being asked to prove. Calculate this first; if the test cannot detect a 15% lift at 90% confidence, redesign it or do not run it.
- Read at 6 weeks and 12 weeks. Luxury consideration cycles routinely exceed 30 days, and a four-week readout will systematically understate performance.
Negotiating the Contract: What to Ask For
Rate cards are the least interesting part of a luxury retail media agreement. These six terms determine whether the investment is measurable.
- Holdout support in writing. The retailer must agree to suppress a defined audience or geographic group from targeting for the duration of the test, and to report on that group's purchase behavior alongside the exposed group. This is the entire basis of an incrementality read.
- Clean-room or log-level data access. Aggregate dashboards cannot answer whether the exposed cohort differed from the control on pre-period spend. Access to exposure and transaction data in a privacy-safe environment can. Several luxury networks now support this; ask before signing, because retrofitting it mid-flight rarely happens.
- Attribution window control. Default windows are set to flatter the network. For a category with a 30-to-90-day consideration cycle, a 14-day view-through window is arbitrary in both directions. Negotiate the right to see results at multiple windows rather than accepting one.
- Separate onsite and offsite reporting and pricing. Bundles obscure which asset produced the result. Insist on line-item separation even when the commercial agreement is bundled.
- Supply path disclosure on offsite. Ask which DSP and which SSPs the retailer uses for offsite activation, and what the total take rate is between your dollar and the publisher. If the answer is unavailable, the offsite product is a black box priced at a premium.
- Third-party verification rights. Offsite retail media is programmatic media and should carry the same viewability, brand safety, and invalid traffic measurement as the rest of the plan.
Networks that accommodate most of these are serious media businesses. Networks that decline all of them are selling trade dollars with a media label attached, and should be evaluated as trade spend.
Common Mistakes We See
- Treating the retailer as a media partner rather than a trade partner. Retail media budgets frequently come out of trade or co-op funds, which changes the negotiation entirely. Understand which pocket the money is leaving before you agree to a rate.
- Accepting bundled onsite/offsite packages. Networks bundle because onsite is scarce and offsite needs a buyer. Insist on separate pricing and separate reporting.
- Ignoring supply path on the offsite side. Offsite retail media is programmatic media with a data fee attached, and it is subject to the same supply path inefficiencies as any open-web buy. Ask which SSPs the retailer's DSP is buying through and what the total take rate looks like.
- Buying for the deck. Luxury retail media presents beautifully to a board. That is not a media strategy.
Where This Fits in a Premium Media Plan
For most luxury and high-consideration brands, luxury retail media networks earn a place as a 5–15% allocation serving a specific job: reaching verified high-spend customers of a retailer where the brand has distribution, with measurement rights strong enough to prove it worked. It does not replace premium CTV for reach, it does not replace direct publisher relationships for brand environment, and it is not a substitute for building a first-party data asset the brand controls.
Bought with discipline, it is one of the few places in digital media where "affluent" is a verified fact rather than a modeled inference. Bought without discipline, it is an expensive way to buy customers who were already walking toward the register.
Ready to pressure-test whether luxury retail media belongs on your plan? Stillwater Media takes a limited number of new engagements each quarter, and every one starts with an honest read of where your media dollars are actually working. Apply to work with us.
Frequently Asked Questions
What is a luxury retail media network?
A luxury retail media network is an advertising business run by a high-end retailer — such as Saks, Neiman Marcus, Nordstrom, or Net-a-Porter — that lets brands buy ad placements using the retailer's first-party shopper data. It sells three things: onsite placements on the retailer's own site, offsite programmatic and CTV media targeted with the retailer's audience segments, and closed-loop measurement tying ad exposure to purchases made in the retailer's environment. The distinguishing asset is verified high-spend purchase data rather than modeled affluence estimates.
How much do luxury retail media networks cost?
Onsite display typically runs $5–$25 CPM and sponsored product placements are often priced on a CPC basis, while offsite programmatic display runs $25–$60 CPM and offsite CTV runs roughly $28–$65 CPM. Luxury networks generally price at or above the top of these ranges. The practical question is not the absolute CPM but the data premium — offsite retail media often costs $15–$45 more per thousand impressions than the same inventory bought through a standard programmatic path, and that premium is only justified when the audience is genuinely unavailable from other data sources.
Is retail media ROAS the same as incremental ROAS?
No, and the gap is large. Retail media networks report on their own attribution models, typically crediting purchases within a 14-day view-through or 14- to 30-day click-through window inside the retailer's environment, which captures many customers who would have purchased anyway. In luxury categories with strong pre-existing brand demand, retailer-attributed ROAS commonly overstates incremental ROAS by 1.8x to 3.5x. The only reliable correction is a pre-agreed holdout group that the retailer suppresses from targeting.
Should a luxury brand buy onsite or offsite retail media first?
Buy onsite first if the brand already sells through the retailer and the goal is winning share of an in-market category shopper. Buy offsite first if the goal is reaching the retailer's verified high-spend customers with a brand message on CTV or display, which is where the audience data creates advantage the brand cannot replicate. Avoid buying both simultaneously in equal weight, because bundled reporting makes it impossible to determine which channel produced the result.
How much budget should a luxury brand allocate to retail media?
For most luxury and high-consideration brands, a 5–15% allocation of working media is appropriate once the channel has proven incremental. A first test should be capped at 8–12% of quarterly working media — large enough to generate readable signal given low luxury purchase frequency, small enough that a null result does not damage the year. Before committing, calculate the minimum detectable effect; many luxury retail media tests are statistically incapable of proving the lift they are asked to demonstrate.
What makes luxury retail media different from Amazon or Walmart retail media?
Three structural differences matter. Purchase frequency is roughly an order of magnitude lower — a luxury retailer may see a high-value customer four to eight times a year versus 40-plus for a grocer — which lengthens learning periods and widens confidence intervals. Addressable scale is far smaller, so campaigns hit frequency ceilings quickly. And cross-channel purchase leakage runs heavily in one direction, because luxury shoppers routinely discover at the retailer and buy at the brand boutique, making the retailer's closed-loop measurement a systematic undercount of total impact.
