What Snapchat x Dick's Clean-Room Attribution Costs You

On July 14, 2026, Snap Inc., DICK'S Sporting Goods, and LiveRamp flipped on closed-loop clean-room attribution: matching a Snapchat ad exposure directly to a verified DICK'S purchase, in-store or online, no pixel required. The headline number everyone is repeating is an adidas sneaker pilot that reportedly returned over $12 ROAS. That matters because $12 is doing a lot of work it probably can't support on its own, and if you run paid media anywhere, the real story here isn't the number, it's what a measurement upgrade like this actually costs and who it's actually built for. Expect more of these announcements through the back half of 2026 as every walled garden races to prove its attribution works, so it's worth knowing now what to watch before a vendor sells you the wrong tier of it. For the underlying spend math, see our breakdown of where retail media budgets are actually going, and for how a fractional CFO evaluates a measurement-vendor pitch like this one, keep reading.

What happened

LiveRamp, Snap Inc., and DICK'S Sporting Goods (through its DICK'S Media retail media network) announced general availability of a closed-loop clean-room integration for all US advertisers on July 14, 2026. The architecture has three parts: Snap contributes anonymized ad-exposure data, DICK'S contributes verified purchase data (in-store and online), and LiveRamp resolves identity between the two datasets using its identity graph, roughly 700 million consumer profiles sourced from 150 data providers, without either party seeing the other's raw records.

The proof point everyone is quoting is a pilot with adidas, run around the Anthony Edwards 2 and Harden Volume 10 sneaker launches, that reportedly returned "over $12 ROAS." That's nearly double the Q1 2025 cross-platform retail-media average of 6.1x reported by Skai. Neither Snap nor LiveRamp has published the campaign's spend, impression volume, or flighting, so there's no way to independently verify the $12 figure against a baseline.

Claim	Status
Announcement date: July 14, 2026	Confirmed
Parties: Snap Inc., DICK'S Sporting Goods, LiveRamp	Confirmed
Adidas pilot products (Anthony Edwards 2, Harden Volume 10)	Confirmed
Reported ROAS: $12+	Confirmed as reported, not incremental
Q1 2025 cross-platform retail-media ROAS average: 6.1x	Confirmed (Skai)
LiveRamp identity graph: ~700M consumers, 150 data providers	Self-disclosed
iROAS = 30-60% of reported ROAS	Industry rule of thumb, not party-confirmed
Publicis to acquire LiveRamp at $38.50/share, ~$2.17B enterprise value	Confirmed (announced May 17-18, 2026)

Why this matters for your business

Here's the nuance that matters more than the headline number itself: reported ROAS and incremental ROAS are not the same thing, and the gap between them is usually large.

Reported ROAS credits every purchase a platform can tie to an ad exposure, including the customers who would have bought the sneakers anyway because they already loved adidas or already shop at DICK'S. Incremental ROAS isolates the sales the ad actually caused, typically measured against a holdout group that saw no ad at all. The rule of thumb cited in coverage of this exact deal is that true incremental lift runs 30-60% of reported ROAS. Apply that to the adidas pilot's $12 and the real number is likely somewhere between $3.60 and $7.20.

This isn't a knock on this specific deal, it's the pattern with every retail-media ROAS claim you'll ever be pitched. When we've dug into a brand's channel mix, we've seen a channel get labeled "unproductive" and cut, only to watch a completely different channel's conversions drop right alongside it, because the customer saw the ad on the channel that got cut and then converted somewhere else that got the credit. Without a system that ties exposure to the actual sale (which is exactly what a clean room claims to fix), you end up cutting the channel that was quietly doing the work and keeping the one that was just standing where the customer happened to convert.

The stakes on getting this wrong are not abstract. A good contribution margin in DTC runs around 20% in our experience, and every dollar of acquisition cost above what that margin supports is a loss on the first order. A 30-60% haircut between reported and incremental ROAS isn't academic when you're deciding whether a channel is actually profitable or just well-attributed.

What this infrastructure actually costs

Clean-room measurement is not free, and the cost curve is steep. The average enterprise data clean-room setup runs $879,000, and 62% of clean-room users spend at least $200,000 a year, with 23% spending over $500,000. For a mid-market brand ($50M-$500M in revenue), a realistic Year 1 range is $75,000-$250,000 in vendor and cloud costs alone, before internal data-engineering time.

Vendor	Pricing model	Mid-market annual cost	Best fit
LiveRamp Data Collaboration Platform	Consumption / token subscription	$75K-$500K (entry SKU ~$120K/yr)	Cross-cloud, multi-retailer measurement
AWS Clean Rooms	Usage-based ($4/CRPU-hour + per-record fees)	$5K-$100K+	AWS-centric technical teams
Snowflake Data Clean Rooms	Standard compute credits (~$2-4/credit)	$5K-$50K (compute only)	Brands already on Snowflake
InfoSum	Tiered subscription	$68K+ (Bronze tier)	Privacy-sensitive / EU use cases
Walled-garden free tiers (Amazon Marketing Cloud, Google ADH, Meta Advanced Analytics)	Free with ad spend	$0	Single-platform measurement only

The math only starts favoring a neutral, paid platform above roughly $2 million in annual media spend with multi-retailer or multi-platform complexity. Below $500,000 a year, server-side conversion tracking and GA4 do the job. Between $500,000 and $2 million, the smarter move is testing the clean rooms already bundled inside the platforms you're paying for, before you sign a contract for a neutral one.

The vendor behind the curtain: LiveRamp's own numbers

Before any brand builds a dependency on this specific stack, it's worth knowing two things about LiveRamp itself. First, it just crossed into real profitability. FY2026 (ended March 31, 2026): revenue $812.9 million, gross margin 70.7%, and operating margin 10.3%, up from 0.7% in FY2025 and 1.7% in FY2024, after an operating loss as recently as FY2023 (-21.1%).

LiveRamp fiscal year (ended March 31)	Revenue	Gross margin	Operating margin
FY2023	$596.6M	71.5%	-21.1%
FY2024	$659.7M	72.8%	1.7%
FY2025	$745.6M	71.0%	0.7%
FY2026	$812.9M	70.7%	10.3%

For context, The Trade Desk, a mature, profitable ad-tech infrastructure business, runs a 20.3% operating margin. LiveRamp's numbers say clean-room infrastructure is still an earlier-stage, lower-margin business than programmatic buying, not a mature cash cow with years of proven economics behind it.

Second, LiveRamp is mid-acquisition. Publicis Groupe announced a deal to buy LiveRamp on May 17-18, 2026, at $38.50 a share (a roughly 30% premium), an enterprise value of about $2.17 billion, with close expected before the end of 2026, subject to regulatory and shareholder approval. That's not disqualifying on its own, plenty of good products change ownership, but it means any brand building a long clean-room roadmap on LiveRamp specifically is also underwriting an acquisition integration it has no control over.

DICK'S own numbers explain why a retailer builds a media network in the first place. DICK'S Sporting Goods' FY2025 (ended January 31, 2026) gross margin compressed to 32.9% from 35.9% the prior year, and operating margin fell to 6.4% from 11.0%, largely from the lower-margin Foot Locker acquisition. Retail media, ad revenue sold against DICK'S own audience of 45 million-plus marketable customers and 200-plus brand partners, is a high-margin lever retailers are pulling to offset that kind of core-margin pressure. It's the same reason nearly every major retailer now runs one: Kroger Precision Marketing, Target's Roundel, Albertsons Media Collective, and Best Buy Ads all disclose named retail media businesses in their most recent annual filings, though the field is lopsided: Amazon and Walmart already control 87.7% of the roughly $69-71 billion in US retail media spend forecast for 2026 (Amazon 79.7%, Walmart 8.0%), leaving DICK'S Media and every other network fighting over what's left.

What to do if you're not at $2M+ in retail-media spend

Most DTC and mid-market ecommerce brands are not the target buyer for this specific product, and that's fine. The move isn't to skip measurement upgrades, it's to sequence them correctly.

Start with what's already free. Amazon Marketing Cloud has been free with Amazon ad spend since September 2025. Google Ads Data Hub and Meta's Advanced Analytics (in beta) offer similar single-platform matching at no incremental cost. If you haven't checked whether your current media spend already gives you access to one of these, that's the first hour of work here, not a $250,000 contract.

Second, treat any clean room, free or paid, as a verification layer on top of incrementality tests and holdouts, never as a replacement for them. The operators who get the most out of measurement upgrades aren't the ones chasing the newest attribution tool, they're the ones running basic holdout tests on their top two or three channels every quarter and using whatever attribution data they have (clean room or otherwise) to sanity-check the result, not to generate it. We've seen brands lose real money to exactly this kind of measurement blind spot: one operator's CPMs quietly quadrupled, from roughly 100 to 400, and nobody noticed until someone happened to check the dashboard, by which point a channel that should have delivered around 20 conversions had delivered seven. A $879,000 clean room wouldn't have caught that. A basic weekly efficiency check would have.

The operator decision here was never "should I build a clean room." It's "which parts of this measurement upgrade are already free inside the platforms I'm already paying for, and which parts are being sold to me as a solution to a problem I don't have yet."

What to watch next

Three things to do this week, not this quarter:

• Audit your free tier first. Check whether Amazon Marketing Cloud, Google Ads Data Hub, or Meta Advanced Analytics already give you cross-exposure matching on the platforms you're spending on, before anyone pitches you a paid clean room.
• Ask "reported or incremental" every time a vendor says ROAS. If a rep can't answer without checking, assume it's reported, and haircut it by 40-70% before you make a budget decision off it. For a working framework on the acquisition-cost side of that math, see our guide to reducing ecommerce CAC.
• Don't reallocate budget off a single-platform ROAS claim without a holdout. One pilot with one brand on one retailer, adidas at DICK'S in this case, is a case study, not a benchmark for your category. Our look at how Australian buyers are measuring retail media ROI covers the same measurement gap from a different market, and the pattern holds.

Sources and methodology

Financial figures come straight from SEC filings. LiveRamp, Snap, DICK'S Sporting Goods, and The Trade Desk margin and revenue figures are pulled directly from each company's 10-K/10-Q filings on SEC EDGAR, not from press summaries or secondary reporting.

Clean-room cost figures are synthesized from vendor pricing and analyst surveys, not one price list. The $879,000 average enterprise setup cost and the 62%-spend-$200K+ figure come from IAB and Funnel.io survey data; Forrester's Total Economic Impact study on LiveRamp supplies the multi-year cost tiers. The $2 million annual-spend threshold is directional guidance synthesized from that same guide, not a vendor-published breakeven number, so treat it as an estimate, not a hard line.

The reported-to-incremental ROAS ratio is an industry rule of thumb, not a disclosed number. Neither Snap nor LiveRamp has published the adidas pilot's spend, impression volume, or flighting; the 30-60% incremental-to-reported conversion is cited across dated trade coverage of the deal, not confirmed by either party.

US retail media market-size figures come from eMarketer's ongoing forecast series, which is periodically revised. Two vintages of the 2026 figure ($69.33B and $71.09B) are both in circulation; we show the range rather than pick one.

Two active legal and regulatory threads involving LiveRamp are worth knowing about, though not detailed here. A putative class action (Riganian v. LiveRamp Holdings, filed January 2025, N.D. Cal.) and data-protection complaints from the Open Rights Group to the UK ICO and French CNIL (filed February 2024) were both unresolved as of this writing. Verify current status independently before treating either as settled.

Where we describe patterns we've seen with ecommerce operators, those observations are anonymized from our own client conversations, never from public reporting on this specific deal, and no client is named or identifiable.

Frequently asked questions

what is a retail media clean room?

A data clean room is a neutral, privacy-safe environment where two companies (say, a platform like Snapchat and a retailer like DICK'S) can match their data against each other without either side seeing the other's raw records. In this deal, LiveRamp runs the match: Snap contributes ad-exposure data, DICK'S contributes purchase data, and LiveRamp resolves identity between them using its own identity graph so no pixel or cookie has to follow the customer around.

how much does a data clean room actually cost?

It depends heavily on the vendor and scale. Free options exist inside the platforms you already use (Amazon Marketing Cloud, Google Ads Data Hub, Meta Advanced Analytics). Usage-based options like AWS Clean Rooms or Snowflake Data Clean Rooms run roughly $5,000-$100,000 a year. A neutral, cross-retailer platform like LiveRamp runs $75,000-$250,000 for a mid-market deployment, and a full multi-clean-room enterprise setup averages $879,000, with some Forrester-modeled deployments running past $1.25 million by year three.

is the $12 roas from the snapchat and dick's deal real?

The number itself is real, but it's not the whole story. It's reported ROAS from a single adidas sneaker pilot, and neither Snap nor LiveRamp has disclosed the spend, impression count, or flighting behind it. Reported ROAS almost always overstates the sales a channel actually caused, because it credits purchases that would have happened anyway. Treat $12 as a marketing number, not an audited result.

what's the difference between reported roas and incremental roas?

Reported ROAS counts every purchase a platform can attribute to an ad exposure, including customers who would have bought anyway. Incremental ROAS (iROAS) isolates the sales the ad actually caused, usually measured with a holdout test. The industry rule of thumb cited around this deal is that iROAS runs 30-60% of reported ROAS, so a $12 reported number likely reflects $3.60-$7.20 in real incremental return.

do i need a clean room if i'm not a huge retailer?

Almost certainly not yet. The economics only start working above roughly $2 million a year in retail-media spend with multi-retailer or multi-platform complexity, and below $500,000 a year, server-side conversion tracking and GA4 are enough by most analysts' and vendors' own admission. Below $2M, the smarter move is testing the free clean rooms already bundled into the ad platforms you're paying for.

how much ad spend do i need before a clean room is worth it?

Roughly $2 million a year in retail-media or cross-platform spend, per the vendor-cost and analyst data behind this deal. Between $500,000 and $2 million, test what Amazon Marketing Cloud, Google Ads Data Hub, or Meta Advanced Analytics already give you for free before you pay for a neutral platform. Under $500,000, a paid clean room is very likely overkill.

is liveramp a safe vendor to build on right now?

LiveRamp is a going concern and just posted its first real double-digit operating margin (10.3% in FY2026), but it's also mid-acquisition: Publicis Groupe announced a deal to buy LiveRamp on May 17-18, 2026 for $38.50 a share, roughly a $2.17 billion enterprise value, with close expected before the end of 2026. That's not a red flag by itself, but any brand building a long-term dependency on LiveRamp's specific stack right now is betting on how that integration plays out, not just on the product as it exists today.