Retail Media Intel

Executive summary

Retail media's center of gravity shifted decisively in-store this week, but plumbing lags ambition. IAB/IAB U.K./SMG surveyed 281 professionals and found in-store is the top incremental growth driver yet the hardest to connect: only 14% always plan from a single brief or get unified reporting, and 46% of brands require measurable ROI to increase in-store spend. Placer.ai's Kroger/Sam's Club/CVS playbooks show why: trade-area income, trip frequency, visit duration and dwell vary sharply by banner and market. Kroger is expanding digital endcaps for discovery, Sam's Club is leaning into in-club screens + audio on growing unique visitors, while Target's Roundel, Walmart and Dollar General keep scaling. Vistar Media with Omnicom/Annalect (n=7,200) proved a no-cost lever — top/bottom high-contrast store locators with pin icons lift visit and purchase intent in pDOOH — and South Korea's CU is testing synthetic shoppers + digital twins to predict signage performance pre-build. Commerce is being rewritten by AI agents abstracting discovery and checkout away from retailer sites. Google now calls itself a commerce 'matchmaker, not a retailer' with Universal Commerce Protocol for agent-compatible shopping and Universal Cart live in the U.S. on Gemini and Search — discovery and transaction stay inside Google, fulfillment stays with retailers. The e-Conomy SEA 2026 report explains why this matters: 50% of SEA consumers use AI to discover products, 69% to compare, 19% start on standalone AI apps, though 90%+ still verify via search/reviews/social. Google is testing Gemini-powered Business Agent in Search ads in India, EMARKETER launched an AI Visibility Index (Apple led at 20% of ChatGPT/Gemini mentions in September), Harper named ex-Akamai Lelah Manz CEO to push millisecond in-memory runtimes for AI agents with Fortune 50 retail adoption, and Block unveiled Dessa local full-duplex voice AI with obvious Square ordering implications. Agencies are already productizing the shift: Brandtech cites Salomon +83% eCommerce revenue via AI-led Google Shopping optimization. Infrastructure money and CTV-to-checkout plumbing confirm a platform war between independent transparent infra and walled-garden AI curation. Kevel raised $52-53M (total >$100M from Fulcrum, dunnhumby Ventures, Iberis, ORIX) to scale API-first commerce media with transparent AI decisioning for clients like Chewy and PayPal. Criteo countered with a 3-year Flaconi extension pairing onsite with C-Max Offsite Services. Google launched Gemini-curated DV360 Marketplace bundles (50 packages per line item) + LinkedIn CTV + Gemini Search Ads 360 Home linking CTV to checkout. Netflix expanded Kokai with The Trade Desk for November: first/third-party targeting with 4M+ segments from 300+ commerce/behavioral providers plus planned in-DSP measurement. Coactive claims scene-level AI can unlock 80% of streaming inventory now untapped as ad-supported nears half of premium subs — just as Projectivy Launcher (2M+ downloads), CleanYourTV and others let users strip smart-TV OEM home-screen ads, threatening OEM audience value with India CTV at 207M viewers. The cross-domain flywheel is live commerce + entertainment IP + fast fulfillment + global-local data. QVC is building a post-Chapter 11 (August) holiday flywheel across 60,000 hours of TV and 200 hours/week of TikTok Live, with TikTok Shop driving nearly 1M new customers in 2025 via distinct hosts/sets/hooks. Walmart's Muppet Christmas Carol 360 Who Knew? campaign with Walton Goggins/Stephanie Beatriz spans broadcast/streaming/social/creators while plugging Express and 30-minute delivery. MercadoLibre heads into early-November Q3 after its fastest revenue growth in nearly four years on logistics/credit/tech, with focus on marketplace GMV, Mercado Pago TPV, advertising vs traffic, and credit quality. In Africa, Shoprite/Checkers, Pick n Pay and Takealot are building loyalty/first-party purchase-data ecosystems with PAPSS payments as enabler; QSRs (87% face higher food costs, traffic down) are piloting app/kiosk/menu-board media networks where franchisee rev-share is the scale gate; Wesfarmers' OneReach ran Airbnb x Bunnings non-endemic through December; and Quebec Bill 29 (effective Oct 5, 2026) forces 3- to 6-year durability disclosures near price, now an optional Merchant Center warranty attribute for Shopping ads — feeds are becoming compliance infrastructure.

Market dynamics

Consolidation is bifurcating into owned monetization stacks vs aggregated buying points. Retailers want differentiated networks they control — hence Kevel's raise and Retail Media Cloud funding, Flaconi owning onsite + buying offsite via Criteo, and African and LatAm players tying loyalty + payments + ads (Takealot, Mercado Pago) — while buyers want one place to buy commerce audiences across CTV/display/search: DV360 Gemini bundles, Netflix-TTD Kokai with 300+ data providers, Omnicom Q3 core organic +7.2% to $6.01B without channel disclosure hiding where commerce sits. Talent circulation (The Trade Desk <> Walmart DSP <> Albertsons Media Collective) signals onsite/offsite convergence is now a personnel flow, not just a product roadmap. The tech shift is from touch-and-search to AI-orchestrated and simulated. Opinion pieces argue the next interactive store won't wait to be touched — operational signals + AI + retail media will decide what screens show, requiring governance for competing demands on inventory — while CU's digital twins move planning from costly pilots to pre-build prediction, and Coactive/Gemini move CTV from title-level to scene-level/AI-ranked curation. Google's Target ROAS tolerance (start at 10% degradation for Smart Bidding Exploration/promotion mode) makes the trade explicit: strict efficiency guarantees are being sacrificed for exploration volume, forcing advertisers to budget ROAS slippage for growth as AI ranking dictates which inventory and queries get priority. Budget flows follow proof, not inventory. Brands prioritize in-store digital, RMNs still favor onsite, and IAB's 46% ROI gate plus ANA Masters' ditch-vanity-metrics push mean outcome-based measurement is the unlock for golden-quarter spend where IGD sees same vehicles/higher stakes: HFSS compliance, front-of-store, digital screens, QR, loyalty-linked incentives and gifting amid clutter/value-seeking. Regulatory pressure is now product data pressure: Quebec's 3-6 year warranty attribute, HFSS display rules for Christmas, and privacy constraints on responsive stores mean regional durability, health and consent rules land directly in Merchant Center attributes, in-store content rules, and AI-agent feed requirements like UCP compliance.

Key developments

Strategic implications

Builders should treat agent-ready product data as the primary surface, not retailer PDPs. That means UCP-compliant structured feeds, Universal Cart checkout compatibility, warranty/durability and fulfillment attributes (Quebec 3-6 year disclosures, 30-minute/Express delivery flags), and AI-visibility tracking modeled on EMARKETER's ChatGPT/Gemini index — if 50% discover and 69% compare via AI in SEA while 90%+ still verify, winning requires optimizing for AI recommendation and for the verification click on search/reviews/social. Pair this with low-latency edge runtimes like Harper's for agent/voice (Block Dessa) ordering, and with AI feed optimizers that already deliver +83% lifts, or brands will depend entirely on Google for visibility. For retail media products, the bottleneck to own is unified in-store + onsite + offsite planning and proof. Only 14% get single-brief planning/unified reporting and 46% need ROI to scale in-store — so ship one brief, traffic-validated placement using Placer-style visit/dwell/segmentation, standardized creative (Vistar's top/bottom high-contrast locator + pin, IGD front-of-store/QR/loyalty tactics), and incrementality/visit/purchase reporting that satisfies ANA's ditch-vanity-metrics bar. Support the two monetization expansions actually working: onsite + offsite bundles (Flaconi x Criteo C-Max, Netflix-TTD Kokai's 4M segments, DV360's 50-package bundles) and new vertical networks — QSR app/kiosk/menu-board with franchisee rev-share accounting, travel-signal (Navigator), and non-endemic lifestyle (OneReach Airbnb x Bunnings) — with governance for AI-orchestrated stores competing for screen inventory. Design for transparency as differentiation against Gemini-ranked curation and Smart Bidding exploration. As DV360 centralizes CTV ranking and Target ROAS tolerance explicitly trades ~10%+ efficiency for query diversity, buyers will pay for independent decision logs, scene-level metadata (Coactive's 80% untapped thesis), synthetic-shopper pre-build simulation (CU digital twins), and pDOOH/CTV-to-store measurement that survives OEM disintermediation from launchers. Kevel's >$100M bet on transparent AI is the signal: own the explainable optimization layer that lets retailers keep monetization and lets brands audit what AI chose and why.

Competitive landscape

Google is winning across both domains by owning the agent choke point — UCP, Universal Cart, Gemini Business Agent, Gemini-curated DV360 bundles and Search Ads 360 Home — turning discovery, curation and checkout into Google-mediated steps while retailers keep fulfillment and compliance risk. Walmart/Target Roundel and implicitly Amazon lead U.S. retailer monetization (Roundel growth vs Walmart/Dollar General scale, Walmart Connect/Albertsons in DV360), MercadoLibre leads LatAm's marketplace + Pago + ads flywheel, and Shoprite/Pick n Pay/Takealot are early winners in Africa's purchase-data race. The open challenge comes from independent infra and programmatic commerce: Kevel vs Criteo (Flaconi 3-year onsite+offsite) for retailer stacks, and The Trade Desk + Netflix Kokai vs Google DV360 + LinkedIn CTV for who aggregates commerce audiences for CTV, with TikTok Shop/QVC proving live video can still mint ~1M new customers outside both.

Action items

Trend lines

Contradictions

• In-store is framed as the top incremental opportunity while only 14% achieve single-brief planning or unified reporting, with 46% of brands demanding ROI proof before scaling.
• CTV is scaling as a commerce performance channel via Netflix-Kokai and DV360 Gemini bundles while AI-powered ad blockers simultaneously threaten smart TV OEM ad revenues.
• Google pitches itself as a neutral commerce matchmaker via Universal Commerce Protocol while consolidating discovery, curation and cart control through Gemini, Search and DV360.

Blind Spots

• No in-store pricing, yield or incremental ROAS benchmarks to compare onsite vs in-store vs offsite.
• Missing SMB and long-tail brand perspective; evidence skews to large networks like Target, Kroger, Walmart and CVS.
• No attribution model for AI-agent discovery to checkout; unclear how Universal Cart and Gemini referrals will be measured and monetized.
• Limited privacy and consent detail for in-store audience simulation, foot-traffic tracking and CTV onboarding.

Connections

• Localized foot-traffic, dwell and trade-area income from Placer.ai is becoming shared input linking in-store media placement to assortment and fulfillment planning.
• Google Universal Commerce Protocol and Universal Cart turn AI product discovery into agent-mediated checkout, directly coupling retail media search visibility to commerce conversion infrastructure.
• DV360 Gemini-curated bundles carrying Walmart Connect and Albertsons inventory link CTV programmatic buying to trade-area commerce outcomes.
• QVC TikTok Shop to TV flywheel and Walmart-Disney shoppable holiday show shoppable video collapsing retail media reach and commerce transaction into one loop.

Article index