Issue 04

The Shelf

Daniel Park almost built another ad platform nobody would use. Then he realized ShopGrid's moat was never the dashboard.

Featuring Daniel Park at ShopGrid
Stories / Retail Media

Every retail media network starts the same way. A marketplace with millions of shoppers looks at its purchase data and thinks: we should sell ads. Then someone draws a roadmap on a whiteboard. Self-service campaign tool. Proprietary audience builder. Custom reporting dashboard. Eventually a DSP.

Daniel Park drew that roadmap. He's VP of retail media at ShopGrid, a marketplace with 200 million monthly shoppers. His board wanted an ad business. He had the data. The path looked obvious.

Daniel reviews a cluttered platform roadmap on a wall screen — dashboard mockups, audience builders, and reporting UIs pile up while he looks skeptical

Six months and 40% of the engineering budget later, Daniel looked up.

The $12 million dashboard

The CPG brands Daniel wanted to attract were already managing campaigns across a dozen retail media platforms. Each with its own API, its own audience taxonomy, its own reporting format. Every mid-size retailer was asking the same brands to integrate yet another proprietary system.

Daniel was building a dashboard to compete on UI quality against companies with 100x his engineering budget.

He pulled the usage numbers on the beta: 23 logins in a month, mostly from his own team testing it. The two brands that had onboarded were running minimum-spend campaigns to "evaluate the platform." Their feedback was polite and damning — nice start, but we already manage eleven of these.

ShopGrid was never going to win that fight.

The Tuesday walk

Daniel walks a retail floor as data streams flow from his tablet to glowing digital screens on end-caps and checkout lanes — the store as a data asset

Daniel walked one of ShopGrid's flagship stores on a Tuesday afternoon and it hit him. He was standing in front of a digital end-cap screen when a loyalty member scanned her app at checkout three aisles over. Real-time, deterministic, matched to a receipt. He could tell you what she bought, what she browsed, what aisle she lingered in, and whether the screen she passed showed a competitor's ad.

No DSP on earth has that signal. No walled garden shares it. ShopGrid's advantage was never going to be its software. It was this:

A self-service platform locks all of that behind a proprietary UI. Only brands willing to learn ShopGrid's tools get access. Daniel needed the opposite: make the data and inventory accessible to the widest possible set of buyers, without asking any of them to learn anything new.

Daniel's pivot echoes a pattern across retail media. The retailers winning ad revenue aren't the ones with the best dashboards — they're the ones whose data is easiest to activate. The dashboard is a cost center. The data is the product.

One storefront, everything on the shelf

Daniel at his desk as retail media product cards radiate outward from his monitor — sponsored products, display, video, in-store, and premium placements being published

Daniel stood up an agentic storefront instead. ShopGrid's entire retail media catalog — sponsored products, on-site display, in-store screens, homepage takeovers — exposed as products that any buyer agent can discover.

The loyalty data powers closed-loop attribution. The store locations are modeled as catalogs with catchment areas for proximity targeting. Daniel defines the products, the pricing, and the rules. Buyer agents handle discovery and execution.

And because ShopGrid runs the server, every transaction flows through their infrastructure. Data stays home.

What ShopGrid's storefront sells
Sponsored product listings   — CPC, catalog-rendered, no creative upload needed
On-site display and video    — CPM, first-party audience targeting
Off-site audience extension  — ShopGrid purchase data on third-party inventory
Premium placements           — Homepage takeovers, category sponsorships, flat-rate
In-store digital screens     — Checkout lanes, end-caps, pharmacy waiting areas
                               2,000+ locations, guaranteed delivery
Each product type is exposed via get_products with pricing, targeting capabilities, and measurement methodology.

Brands that aren't using buyer agents still have basic self-service access. But the storefront is the growth layer — how ShopGrid reaches buyers who would never have integrated a mid-size retailer's proprietary platform.

Sam shows up

Split scene — Sam runs a brief from his agency desk as search beams connect to Daniel's retail media products floating on the right, discovery in action

Sam Adeyemi at Pinnacle Agency is running a cross-retailer campaign for Summit Foods. His buyer agent sends a brief to every retail media storefront in the registry — the same brief, the same protocol, regardless of retailer.

The brief Sam's agent sends via get_products
{
  "buying_mode": "brief",
  "brief": "Organic food brand targeting health-conscious grocery shoppers.
           Looking for sponsored products and in-store screens across
           grocery retailers with deterministic purchase attribution. Q4 2026.",
  "brand": { "domain": "summitfoods.example.com" },
  "filters": {
    "channels": ["sponsored_listing", "digital_out_of_home"],
    "start_date": "2026-10-01",
    "end_date": "2026-12-31"
  }
}
Sam's agent sends this same brief to every storefront in the registry. Each storefront responds with its own matching products.

ShopGrid's agent evaluates the brief, matches Summit's product catalog against ShopGrid's inventory, and returns only the products where Summit has eligible GTINs. Sam's agent sees sponsored products, on-site display, and in-store screens — all with pricing and measurement methodology — through the same protocol it used for the other eight retailers in the campaign.

Sam never logged into ShopGrid's platform. He didn't learn a new UI. He didn't schedule a call with a sales rep.

Sam's agent validates ShopGrid's signals

Before committing budget, Sam's agent calls get_signals to check what targeting data ShopGrid can offer:

ShopGrid's audience signals
{
  "signals": [
    {
      "signal_agent_segment_id": "loyalty_health_shoppers",
      "name": "Health-Conscious Loyalty Members",
      "signal_type": "owned",
      "data_provider": "ShopGrid",
      "coverage_percentage": 38,
      "deployments": [{
        "type": "agent",
        "agent_url": "https://ads.shopgrid.example.com",
        "is_live": true,
        "activation_key": { "type": "segment_id", "segment_id": "sg_health_loyalty" }
      }],
      "pricing_options": [
        { "pricing_option_id": "loyalty_cpm", "model": "cpm", "cpm": 6.50, "currency": "USD" }
      ]
    },
    {
      "signal_agent_segment_id": "organic_purchasers_90d",
      "name": "Organic Category Purchasers (90-day)",
      "signal_type": "owned",
      "data_provider": "ShopGrid",
      "coverage_percentage": 22,
      "deployments": [{
        "type": "agent",
        "agent_url": "https://ads.shopgrid.example.com",
        "is_live": true,
        "activation_key": { "type": "segment_id", "segment_id": "sg_organic_90d" }
      }],
      "pricing_options": [
        { "pricing_option_id": "purchasers_cpm", "model": "cpm", "cpm": 9.00, "currency": "USD" }
      ]
    }
  ]
}
These are signals ShopGrid owns — built from loyalty data and point-of-sale records, not modeled lookalikes. See the Signals protocol for how data providers expose audience segments.

The organic purchaser segment is what closes the deal. ShopGrid doesn't just know who sees the ad — they know who bought the product. That's the data moat Daniel walked past on the store floor.

Sam's agent books the deal

His agent calls create_media_buy to book a combined deal — sponsored products with a 3x ROAS target, plus a homepage takeover — in a single call.

The media buy
{
  "brand": { "domain": "summitfoods.example.com" },
  "start_time": "2026-10-01T00:00:00Z",
  "end_time": "2026-12-31T23:59:59Z",
  "packages": [
    {
      "product_id": "sg_sponsored_grocery",
      "pricing_option_id": "cpc_standard",
      "budget": 45000,
      "targeting": {
        "signal_ids": ["sg_organic_90d"]
      },
      "performance_target": { "metric": "roas", "value": 3.0 }
    },
    {
      "product_id": "sg_homepage_premium",
      "pricing_option_id": "flat_weekly",
      "budget": 15000
    }
  ]
}

From brief to booked. Daniel's team didn't take a call. Sam's team didn't learn a new platform. The agents had a structured conversation, and every response was machine-readable and ready to execute.

The closed loop

Daniel and Sam view a closed-loop attribution flow — impressions flow through clicks, store visits, and purchases in a circular diagram with website, app, and in-store sources

This is commerce media's defining advantage. The same company that shows the ad records the purchase. Not modeled conversions. Not panel-based measurement. Deterministic: loyalty card matched to point-of-sale receipt. Online and in the physical store.

When Sam pulls delivery reports, he sees exactly where conversions happened — website, app, and in-store — broken out with attributed revenue and ROAS. For Summit's ketchup campaign, 12% of attributed sales came from in-store purchases by shoppers who saw ads on checkout screens. No other media channel can close that loop.

Closed-loop attribution in delivery reporting
{
  "attribution_window": {
    "post_click": { "interval": 14, "unit": "days" },
    "post_view": { "interval": 1, "unit": "days" }
  },
  "by_package": [{
    "package_id": "pkg_sp_001",
    "impressions": 850000,
    "spend": 10350,
    "conversions": 2100,
    "conversion_value": 28500,
    "roas": 2.75,
    "by_action_source": [
      { "action_source": "website", "count": 1500, "value": 20000 },
      { "action_source": "app", "count": 350, "value": 4800 },
      { "action_source": "in_store", "count": 250, "value": 3700 }
    ]
  }]
}
Different retailers use different attribution windows — 14-day click vs. 7-day click, for example. The protocol makes this transparent, so Sam can normalize ROAS comparisons across retailers without calling each sales rep to ask how they count.

This is what changes the economics. A CPG brand running across ten retailers used to need ten integrations, ten logins, ten reporting views. Now Sam sends one brief and compares ROAS across all of them, because every storefront reports attribution in the same format.

So you launch the storefront. Then what?

Daniel by a retail storefront and Priya by a streaming TV shape flank Sam in the center — different sell-side verticals connecting to one buyer through the same protocol

The moment ShopGrid's agent goes live, its retail media inventory is one prompt away from every buyer agent in the ecosystem.

An agency strategist asks their planning tool: "Find me retail media across grocery with deterministic purchase attribution and in-store screens." ShopGrid shows up — not because Daniel pitched them, but because the storefront is discoverable through the registry.

A mid-market CPG brand without a dedicated retail media team asks their AI assistant: "I sell organic ketchup and I want to promote it across grocery retailers. What's available for $50K?" Daniel's agent responds with sponsored products and in-store screens — a deal that never would have happened through traditional sales channels, because ShopGrid's sales team would never have prioritized a $50K brand.

Daniel's engineering investment goes into what differentiates ShopGrid: loyalty data quality, attribution accuracy, in-store screen coverage, real-time inventory feeds. Not into another dashboard that agencies use reluctantly.

For mid-size retailers, this is the unlock. You don't need to outbuild Amazon's ad platform. You need to make your data accessible through the same protocol Amazon's data is accessible through. The buyer agent doesn't care about your dashboard. It cares about your data.

On a Tuesday afternoon, Daniel walks the same store floor where the idea hit him. The checkout screens are running a Summit Foods campaign. A shopper scans her loyalty card. The attribution closes in real time. Daniel's phone buzzes — a new buyer agent just discovered ShopGrid's storefront and sent a brief. He doesn't need to take the call. The agent already responded.

His team isn't managing campaigns. They're building product.

Share:

Who builds this?

AAO members are already building the tools and infrastructure for agentic advertising. Here's who can help a retail media network like ShopGrid get started.

Ad ops & retail media platforms

Build and manage the sales agent that powers your retail media storefront.

Operative, Mediaocean, Media.net, ResponsiveAds

Measurement & attribution

Prove closed-loop attribution and incremental lift to buyer agents.

iSpot, Adelaide, Skyrise, mFilterIt

Retail technology

In-store screens, digital shelf, and physical inventory infrastructure.

Broadsign, VIOOH, ShowHeroes

Data & identity

Identity resolution, loyalty data activation, and cross-device matching.

LiveRamp, MetaRouter, Utiq, Evorra, AdFixus

Programmatic infrastructure

Connect your storefront to programmatic demand and manage yield.

StackAdapt, Cognitiv, Sabio, Venatus

Creative & catalog rendering

Generate catalog-rendered creatives, dynamic product ads, and in-store content.

Celtra, Connected-Stories, Camcorder AI, HyperMindZ.ai

Characters you met here appear in other stories

Build your storefront

If you run a retail media network — or want to build tools for one — here's where to start.

Discuss with Addie Learn how storefronts work More stories