Target's Matt Drzwicki explains why retailers are becoming AI's most valuable data partners

Full analysis

Target's retail media chief is making a bigger claim than it first appears. On the surface: Roundel built a new outcome-based ad platform (Precision Plus) that pumps Target's shopper data into The Trade Desk, Google, Meta, TikTok — and now ChatGPT — then measures results against actual store sales. Underneath: retailers are positioning their first-party purchase data as the fuel AI ad platforms can't build without. For ad-tech operators, the question is who owns the value when transaction data becomes the targeting layer of the AI era.

This is a Type 1 shift in the making — hard to reverse if it takes hold — but Type 2 for any single operator deciding how to respond this quarter. The forcing function is real: AI shopping traffic is growing fast, OpenAI and Google are building commerce ad products now, and the retailers are lining up as data suppliers.

The Market Analyst — The interesting move isn't Target's; it's the pattern. Every major retail media network will race to become the "preferred data partner" for OpenAI, Google's commerce platform, and the DSPs. That's a land grab, and it favors the retailers with clean transaction data and store-level measurement — Walmart Connect, Amazon (which needs no partner), Kroger, Target. The losers to watch: identity vendors and cookie-based targeting players. If purchase data becomes the signal AI platforms buy against, LiveRamp-style graphs and third-party segments get commoditized. In plain terms: the thing advertisers pay to target is shifting from "people who look like buyers" to "people who actually bought," and only retailers hold the receipts.

The Skeptic — Every headline number here is self-reported by the guy selling the platform. 2,000% AI traffic growth off an unstated base could be 50 visits to 1,000. "Influenced sales" and "3× lift" are Roundel grading its own homework — the exact conflict of interest that incrementality was supposed to kill. The load-bearing assumption is that ChatGPT ads and Google's commerce platform become real ad channels with real budget. They aren't yet; the formats are, in his own words, "still being defined." Retailers announcing they're "first partners" costs nothing and books no revenue. In plain terms: this is a pitch dressed as a trend, and most of the trend hasn't happened.

The Operator — Precision Plus is a clean-room plumbing job, and clean-room plumbing breaks constantly. The meeting notes make the point better than the podcast: Trade Desk data stuck waiting to land in Snowflake, video line items overpacing $60 in an hour, measurement paused on data-engineering blockers. That's the reality of "objective buying" — the incrementality story only works when the pipes reconcile, and they usually don't on the first pass. Any agency or SSP buying the full-funnel dream should budget for months of integration and a QA team, not a slide. In plain terms: the architecture is elegant on paper and messy the Tuesday morning someone tries to run a campaign through it.

The Customer / End User — Two customers here. The brand (an Aura Ring) wants proven net-new sales, and it's getting a metric defined by the seller — better than impressions, but not independent. The agency wants leverage, and this quietly erodes it: if the retailer owns the data, the targeting, and the measurement, the agency becomes a line item. The one asking for none of this is the shopper, who's now being profiled by purchase history piped into a chatbot's ad slot. In plain terms: brands get better proof, agencies get squeezed, and shoppers get targeted in places they didn't expect ads yet.

The CFO — Follow the margin. Retail media is high-margin because the retailer sells its own data against its own (and now others') inventory. Precision Plus extends that take rate off-site — Target now clips a fee on media running through The Trade Desk and Meta. For DSPs and SSPs, that's a new party standing between them and the budget. For publishers with no first-party purchase data, it's a warning: the money is consolidating around whoever can prove a sale, and impression-based inventory keeps getting cheaper. Incrementality-as-currency isn't a measurement upgrade — it's a repricing of everything that can't prove lift. In plain terms: "prove it moved product" is about to become the price of admission, and a lot of inventory can't.

The tensions:

1. Is AI-platform advertising a real channel or a press release? The Analyst sees an inevitable land grab worth positioning for; the Skeptic sees zero booked revenue and undefined formats. Both are right about different time horizons.
2. Does incrementality help advertisers or the retailer? The Customer wants independent proof; the CFO notes the retailer defines and sells the metric. "Incrementality as currency" is real — but whoever controls the currency controls the exchange rate.
3. Does retail data displace identity, or need it? The Analyst says purchase data commoditizes the identity graph; the Operator knows you still need identity resolution to match a shopper across clean rooms. Retail data may be the fuel, but identity vendors are still the plumbing — for now.

What this actually hinges on: two beliefs. First, that AI platforms (ChatGPT, Google's commerce layer) become channels with enough spend to matter — currently unproven. Second, that retailer-defined incrementality becomes an accepted industry currency rather than a vendor talking point. The council leans toward the structural direction being right — purchase data is genuinely becoming the targeting layer, and retail media networks are the natural suppliers — but leans hard skeptical on the timeline and every specific number in this episode.

Before committing anything: demand third-party-audited incrementality (not the seller's "influenced sales"), and treat AI-channel spend as an experiment budget, not a line item, until formats and pricing exist. The operators most exposed are cookie/segment-based targeters and impression-only publishers; the safest are retailers with clean store-level sales data and the measurement firms that can independently verify lift.

Prediction: Through the end of 2026, retail media networks' spend flowing to AI-chat ad placements (ChatGPT and Google's commerce platform) will remain immaterial — under 1% of total retail media budgets — with no top-10 RMN reporting AI-channel ad revenue as a disclosed line in its Q3 or Q4 2026 results.

Confidence: Medium — Formats are undefined by the guest's own admission; budgets follow proven outcomes, not pilots.

Why: Drzwicki himself says the ChatGPT ad format and experience are "still being defined," and states plainly that media dollars only move once platforms "demonstrate performance" — an admission that the proof doesn't exist yet. Retail media budgets are governed by measurable ROAS and incrementality, so dollars won't scale into a channel with no standard measurement or format in a single quarter. The opposite outcome — a sudden budget shift into AI-chat ads by year-end — would require OpenAI and Google to ship measurable, priced ad products and advertisers to reallocate against them within months, which contradicts both the stated maturity of the products and the deliberate, outcome-gated way retail media buyers move.

Revisit by 2026-12-31: We're right if no top-10 retail media network discloses material AI-chat ad revenue and the channel stays a rounding error in budgets. We're wrong if any major RMN reports AI-platform ad spend as a distinct, material revenue line by its Q4 2026 reporting.

The structural bet — that retail purchase data becomes AI's targeting fuel — is probably correct on a multi-year horizon. It just won't show up in the money this year.