Retail Media Networks: $50B Market, $35B Wasted:
8 segments identify which retail media networks drive genuine incrementality.
"Modeled results show 70% of retail media spend behaves like incrementality theater, with Amazon and Walmart clearing the “credible” bar—but only barely (62 and 58 on a 50-point baseline)."
The research suggests a fundamental decoupling between trust and transaction. While Gen Z consumers report record-low levels of institutional brand trust, their purchase behavior remains robust, driven by a new architecture of peer-to-peer verification.
"If there’s no counterfactual, it’s not incrementality—it’s a story."
"We’re paying for reach we can’t dedupe, and then acting surprised when lift disappears."
"NTB is useful, but only after you tell me exactly what ‘new’ means."
"Offsite is where the math gets magical—especially CTV."
"Promo makes everything look good. It’s the easiest way to fake performance."
"We don’t need another dashboard. We need a test we can rerun next quarter."
"Operational friction is the silent budget killer—teams shift spend to whoever is easiest to report on."
Analytical Exhibits
10 data-driven deep dives into signal architecture.
Where the $35B “incrementality theater” waste hides
Share of total RMN spend attributed to each modeled failure mode (sums to 70%).
"The biggest leakage isn’t CPMs—it’s measurement design: 22% of spend is exposed to no credible counterfactual, and 18% is dominated by existing-buyer capture."
Modeled waste breakdown (share of total RMN spend)
Raw Data Matrix
| Failure mode | % of spend | $ (B) |
|---|---|---|
| No holdout/geo testing | 22% | $11.0B |
| Existing-buyer capture | 18% | $9.0B |
| Promo confounding | 13% | $6.5B |
| Weak matchbacks | 9% | $4.5B |
| Overlap duplicates | 5% | $2.5B |
| Overhead mis-scoped | 3% | $1.5B |
Waste is defined as incremental ROAS < 1.0x OR lift not defensible due to confounds (promo, overlap, lack of counterfactual).
Which networks earn trust vs. which networks get spend
Trust reflects confidence in incrementality claims; usage reflects current buying penetration.
"Amazon and Walmart lead, but even the #1 network scores only 62/100 on incrementality trust—barely above “slightly credible.”"
Incrementality trust vs current usage penetration
Raw Data Matrix
| Network | Trust | Usage | Trust-Usage gap |
|---|---|---|---|
| Amazon Ads | 62 | 78 | -16 |
| Walmart Connect | 58 | 61 | -3 |
| Target Roundel | 55 | 43 | +12 |
| Kroger PM | 52 | 34 | +18 |
| Instacart Ads | 49 | 29 | +20 |
| Walgreens/Pharmacy | 44 | 18 | +26 |
Usage is modeled as % of buyers allocating any meaningful budget (>$250k annualized) to the network.
What buyers will pay for: proof signals that unlock budget
Multi-select; % selecting each as a requirement for larger commitments.
"Proof beats dashboards: 61% will pay more (fees or CPM) for always-on holdouts, while only 29% consider third-party lift studies sufficient on their own."
Incrementality proof requirements (multi-select)
Raw Data Matrix
| Proof feature | % requiring it | Avg fee/CPM premium tolerated |
|---|---|---|
| Always-on holdout | 61% | +9.5% |
| SKU-level matchback | 54% | +7.2% |
| Transparent NTB | 49% | +5.1% |
| Log-level access | 41% | +6.4% |
Premium tolerated is modeled as additional fee load or effective CPM increase accepted if proof reduces reallocation risk.
Onsite vs offsite: where incrementality actually shows up
Index (0–100) where 50=category average performance on each dimension.
"Onsite wins on measurement clarity and cost efficiency, but offsite is where lift collapses fastest when identity/matchback quality dips."
Onsite vs offsite performance (index)
Raw Data Matrix
| Dimension | Onsite index | Offsite index | Gap |
|---|---|---|---|
| Measurement clarity | 63 | 34 | 29 |
| Cost efficiency | 58 | 39 | 19 |
| Scale | 44 | 61 | -17 |
Index is normalized across respondents; volatility reflects modeled variance by identity resolution + overlap.
Grocery RMNs are trusted for baskets—penalized for complexity
Index (0–100) where 50=category average; Grocery=Kroger/Instacart-style; GM=Target/Walmart-style.
"Grocery wins on data freshness and basket attribution, but loses on cost inflation and creative flexibility—creating an “incrementality yes, operational no” ceiling."
Grocery vs general merchandise RMNs (index)
Raw Data Matrix
| Theme | Grocery RMNs | GM RMNs |
|---|---|---|
| Best for | Stock-up, baskets, loyalty targeting | Broad assortment, national scale, creative packages |
| Main constraint | Ops + cost inflation | Incrementality skepticism + overlap |
| Modeled incremental ROAS (median) | 1.31x | 1.25x |
Cost inflation score is inverted (lower score indicates higher perceived inflation and auction volatility).
Formats most associated with real incrementality
Net agreement: “This format typically drives incremental sales in our business.”
"Sponsored Products is the only format with majority confidence (57%). Offsite extensions (display/CTV) trail by 26–30 points due to matchback and overlap concerns."
Net incrementality confidence by format
Raw Data Matrix
| Format | Reported ROAS (x) | Incremental ROAS (x) |
|---|---|---|
| Sponsored Products | 3.6x | 1.7x |
| Onsite display | 3.1x | 1.3x |
| Offsite display | 2.8x | 1.1x |
| CTV extensions | 2.5x | 0.9x |
Net agree = % agree minus % disagree; neutral responses excluded from net calculation.
Where budgets will grow (and why): trust is becoming a gate
Usage here reflects % planning to increase spend on the network in the next 12 months.
"Target and Kroger are “trust-rich” relative to their growth intent—suggesting operational simplification, not credibility, is their bottleneck."
Incrementality trust vs planned spend increase penetration
Raw Data Matrix
| Network | Planned increase | No change | Planned decrease |
|---|---|---|---|
| Amazon Ads | 46% | 44% | 10% |
| Walmart Connect | 39% | 48% | 13% |
| Target Roundel | 31% | 55% | 14% |
| Kroger PM | 27% | 58% | 15% |
Planned decrease includes respondents expecting budget reductions >5% on that network.
The real blockers to incrementality aren’t creative—they’re data rights and dedupe
Multi-select; % selecting each as a top-3 friction point.
"Deduped reach and log-level access are now bigger blockers than CPMs—meaning “better reporting” won’t fix the problem without identity and overlap solutions."
Top friction points blocking scale (top-3 selection)
Raw Data Matrix
| Complexity driver | Avg monthly hours | Avg incremental cost |
|---|---|---|
| Multi-portal reporting + normalization | 18.4 hrs | $6,900 |
| Manual promo/media reconciliation | 12.7 hrs | $4,800 |
| Creative approvals & spec differences | 9.6 hrs | $3,500 |
Ops overhead excludes managed-service fees and is modeled for mid-market spenders ($2M–$10M/year RMN).
Capability gap: what networks offer vs what buyers now require
% of respondents saying the capability is (A) commonly available today vs (B) required to scale budgets.
"The single largest gap is always-on holdouts (28% available vs 72% required). Incrementality-based billing is the next frontier (12% available vs 38% required)."
Current availability vs buyer requirement
Raw Data Matrix
| Capability | Gap (pts) | Budget at risk (12 mo) |
|---|---|---|
| Always-on holdouts | 44 | $3.1B |
| Deduped reach | 39 | $2.4B |
| Log-level access | 29 | $1.8B |
| Incrementality billing | 26 | $1.2B |
Budget at risk is modeled as dollars likely to be paused, shifted, or re-scoped absent the capability.
If $35B gets cleaned up, where does it go?
Distribution of reclaimed dollars among likely destinations (sums to 100%).
"Reclaimed budget concentrates into “credible + scalable” (Amazon/Walmart = 43%), but 21% flows to non-RMN alternatives when networks can’t prove incrementality fast enough."
Reallocation destinations for reclaimed RMN waste
Raw Data Matrix
| Destination | % of reclaimed | $ (B) |
|---|---|---|
| Amazon Ads | 24% | $8.4B |
| Walmart Connect | 19% | $6.7B |
| Target Roundel | 14% | $4.9B |
| Non-RMN digital | 13% | $4.6B |
| Kroger PM | 12% | $4.2B |
| Instacart Ads | 10% | $3.5B |
| In-store + DOOH | 8% | $2.8B |
Reclaimed dollars represent spend shifted from weak/non-incremental placements into channels with higher modeled counterfactual lift and lower overlap.
Cross-Tabulation Intelligence
Segment behavior + belief indices (5–95; higher=more true for segment)
| Agree: “~70% of RMN spend is incrementality theater” | Requires holdout/geo test before scaling | Uses cross-retailer deduping today | Willing to pay for log-level access | Prefers Amazon as anchor network | Actively reallocating trade dollars into media | Trusts retailer-reported NTB as decision-ready | |
|---|---|---|---|---|---|---|---|
| Incrementality Purists (14%%) | 82 | 88 | 41 | 76 | 54 | 47 | 22 |
| Retail Media Pragmatists (18%%) | 69 | 61 | 33 | 49 | 58 | 52 | 41 |
| Platform Loyalists (Amazon-first) (12%%) | 48 | 34 | 21 | 31 | 86 | 38 | 63 |
| Omnichannel Growth Hackers (16%%) | 72 | 57 | 44 | 58 | 62 | 66 | 46 |
| Budget-Guard CFO Shadow (10%%) | 77 | 63 | 18 | 29 | 49 | 34 | 28 |
| New-to-RMN Experimenters (13%%) | 58 | 42 | 17 | 36 | 55 | 59 | 52 |
| Agency Efficiency Seekers (9%%) | 64 | 46 | 27 | 41 | 57 | 45 | 55 |
| Trade-to-Media Reallocators (8%%) | 74 | 52 | 24 | 33 | 51 | 91 | 37 |
Trust Architecture Funnel
Trust-to-budget funnel for retail media incrementality (modeled)
Demographic Variance Analysis
Variance Explorer: Demographic Stress Test
"Brand Distrust 73% → 78% ▲ (High reliance on peer verification in lower income brackets)"
SES here proxies seniority/political insulation. ~$50K–$100K equivalent (junior practitioners): high anxiety, low power; they keep spending where it’s easiest to execute and safest to explain. ~$150K–$250K (directors): most threatened—own performance narratives; they selectively test and quietly narrow RMNs. $300K+ (VP/SVP): more willing to force holdouts, but only when finance pressure rises or when they can trade concessions in JBPs. This demographic slice exhibits high sensitivity to Organizational seniority (SES proxy) because it determines who can survive telling a retailer ‘no’ and who can force counterfactual standards.. The peer multiplier effect is most pronounced here, suggesting a tactical shift toward community-led verification rather than broad brand messaging.
Segment Profiles
Retail Media Pragmatists
Omnichannel Growth Hackers
Incrementality Purists
New-to-RMN Experimenters
Platform Loyalists (Amazon-first)
Budget-Guard CFO Shadow
Persona Theater
MAYA, THE TEST-FIRST COMMERCE LEAD
"Owns a $18M commerce media budget at a national CPG brand; refuses to scale anything without a counterfactual and promo separation."
"Her biggest frustration is that “NTB” means different things across networks; she discounts dashboards by ~30% unless audited."
"Offer always-on geo/holdout tooling with pre-registered test plans and a standardized NTB glossary; target adoption: move her segment’s holdout usage from 34→55 within 2 quarters."
JORDAN, THE SPEED-OBSESSED OMNI PLANNER
"Agency-side, manages 6 clients; will trade precision for speed if learning velocity stays high and overlap is visible."
"He treats offsite as a growth lever only if deduped reach exists; otherwise he caps frequency aggressively and shifts budget back onsite."
"Ship overlap reporting + frequency controls; KPI: reduce ‘duplicate reach’ friction from 52% to 42% among this segment."
PRIYA, THE “GOOD ENOUGH” BRAND MARKETER
"Balanced operator with quarterly targets; wants fewer dashboards and clearer proof thresholds."
"She will pay a ~7% premium for SKU-level matchback because it reduces internal debate time by ~25%."
"Bundle SKU-level matchback + standardized reporting API; KPI: cut reporting normalization hours from 18.4→12.0 per month."
ELI, THE AMAZON-FIRST CATEGORY CAPTAIN
"Runs an Amazon-heavy program where share-of-search is a proxy for success; sees other RMNs as secondary."
"He equates stability with truth; when a new test contradicts ROAS, he assumes the test is wrong."
"Introduce marginal ROAS curves and buyer-suppression testing within Amazon surfaces; KPI: increase incremental ROAS reporting adoption in this segment from 31→45."
CARLA, THE PILOT-HEAVY CHALLENGER BRAND
"Runs a $3M RMN budget across 5+ networks; constantly piloting, frequently under-resourced on measurement."
"She over-weights early wins; without test templates she repeats the same bias mistakes across networks."
"Provide a 14-day pilot kit (holdout defaults, promo flags, overlap checklist); KPI: reduce ‘none’ validation rate from 17% to 12%."
SAM, THE CFO-ADJACENT GATEKEEPER
"Finance partner embedded in marketing; suspicious of platform metrics, pushes for profit-adjusted lift."
"He is most responsive to margin-adjusted incrementality; he will cut offsite first (aligns with 51% cut-first behavior)."
"Implement profit-based incrementality reporting (net of promo + fees); KPI: convert 20% of this segment from ‘decrease/flat’ to ‘increase 1–10%’."
NOAH, THE THROUGHPUT AGENCY OPERATOR
"Optimizes for operational throughput across many RMN portals; prefers standardization over bespoke insights."
"He’ll recommend shifting spend to the networks with the least reporting friction even when trust is equal."
"Offer unified taxonomy + API exports; KPI: reduce ops burden as a top-3 friction from 29% to 22% in 6 months."
DENISE, THE TRADE-TO-MEDIA CONVERTER
"Owns joint business planning; moving trade dollars into RMNs but needs defensible incrementality to survive internal audits."
"She’s the most aggressive reallocator (index 91) but distrusts retailer NTB definitions (37)."
"Create promo/media separation contracts (promo holdouts + clear funding lines); KPI: cut promo confounding as a cited issue from 49% to 41% among trade-heavy programs."
Recommendations
Make holdouts a product, not a project
"Launch always-on geo/holdout experimentation as a default buy option across onsite and offsite. Standardize templates (test length, success thresholds, promo flags) and include pre-registered analysis plans to reduce internal debate time."
Ship overlap and deduped reach as a planning primitive
"Provide cross-retailer overlap reporting (at minimum within a retailer’s onsite+offsite ecosystem; ideally via interoperable clean rooms). Include frequency caps that respect deduped exposure across placements."
Standardize NTB and make it auditable
"Publish consistent NTB definitions (lookback windows, household rules, returns handling) with an audit trail. Provide side-by-side NTB under alternative definitions so finance can reconcile differences without rejecting the metric outright."
Separate promo funding from media measurement (contractually)
"Require promo flags at impression and order level, support promo holdouts, and enforce reporting that shows lift net of discount depth. This directly attacks the 49% promo-confounding driver of non-incrementality."
Rebuild offsite/CTV with stricter truth requirements
"Gate offsite spend behind matchback quality thresholds, deduped reach reporting, and incremental billing pilots. Offsite currently underperforms onsite by 29 points on measurement clarity and drives the largest ROAS illusion gap (CTV 2.5x→0.9x)."
Reduce cognitive load: unify taxonomy + API exports
"Minimize reporting friction via standardized placement taxonomy, stable naming conventions, and API exports that feed BI tools. This targets the $15.2k/month median ops overhead and prevents “ops-driven” budget decisions."
Generate your own Intelligence with the Mavera Platform.
Get Full Access→Join 500+ research teams using synthetic intelligence to generate unique insights.
