By 55% of marketers have GEO dollars inside their budgets, according to a Scribewise survey of 205 executive leaders and marketing managers. PMG is recommending pilots at 1.5 to 2x current search spend. And this week, Shopify announced that millions of merchants can now sell directly inside ChatGPT, Google AI Mode, Microsoft Copilot, and the Gemini app through what they're calling Agentic Storefronts.
The pitch is clean. The measurement story is not.
What Shopify actually announced
Starting this week, Shopify merchants get out-of-the-box access to four major AI channels, managed from the Shopify admin. No separate integrations. No additional transaction fees beyond standard processing rates. Products become discoverable in ChatGPT by default. Inventory, pricing, and checkout customizations stay synced.
For brands not on Shopify at all, there's now an Agentic plan — publicly available — that lets any company list products in Shopify Catalog and reach shoppers across those same AI channels.
Shopify confirmed that orders flow into the admin with channel attribution, so merchants can see whether a sale came from ChatGPT versus Copilot versus Gemini. On desktop, ChatGPT links out to the merchant store in a separate browser tab. On mobile, purchases complete via an in-app browser.
Shop Pay is coming to Microsoft Copilot soon. Shopify also co-developed an open standard with Google called the Universal Commerce Protocol (UCP), which already has stated support from Walmart, Target, Etsy, American Express, Mastercard, Stripe, and Visa.
This is real infrastructure, not a beta. The channel distribution is genuinely broad. Shopify built something that works at scale before anyone else did.
The attribution question nobody is answering loudly enough
Here's where I pull out the scorecard.
Shopify says merchants will see "ChatGPT referral attribution" on orders. That tells you the last AI touch before checkout. What it doesn't tell you is anything about the GEO spend, the content investment, the earned media, or the Reddit thread that got your product surfaced by the model in the first place.
I've seen this pitch before. The deck says "full channel visibility." The invoice says you know the referrer but not the journey.
The underlying problem is structural. The attribution stack was built for a click-based world: a user sees an ad, clicks, lands on a page, converts. The pixel fires. The credit flows. Conversational commerce breaks every step of that chain. There's no impression. There's no click in the traditional sense. There's a dialogue between a user and an LLM, and somewhere inside that black box, your product got recommended.
An AirOps analysis of 548,534 pages across 15,000 prompts found that 85% of sources ChatGPT retrieves never make it into the final answer. Your brand can be crawled and discarded before the user ever sees it. You'll never know. There's no impression log for that.
Meanwhile, the metrics are shapeshifting. Digiday reported that clicks are giving way to brand mention rates and salience as the measurement currency of GEO. John Dawson, VP of strategy at Jellyfish, described the current state as "more tolerance for ambiguity." That's a charitable framing. In practice: you're spending real money with speculative attribution.
Matt Allfrey, head of SEO EMEA at PMG, told Digiday plainly: "One of the things it lacks is a clear output or clear ROI." PMG is still recommending those 1.5 to 2x search budget pilots. I respect the honesty. I'm less certain about the math.
Ryan Bouton, VP of growth at Pawco, told Digiday the pet food brand increased GEO spending 10% in Q1 alone to test LLM discovery. Digiday also reported that Pawco allocated 30 to 35% of the budget for its new brand, Genius Dog, to experimental activity. Neither figure came with a performance result attached — because the measurement infrastructure to produce one doesn't fully exist yet.
So the current ledger: brands are spending more. The channel is real. The attribution is channel-level at best, last-touch by default, and completely silent on the investment that influenced the model's recommendation in the first place.
What this means for your Monday morning
You're going to be asked to justify GEO spend. Probably soon, possibly this quarter. Your CFO wants last-click attribution. What you have is a referral source tag on orders that completed inside an AI chat.
That gap is the actual story here, not the shiny new storefront.
A few things worth knowing before that conversation:
Shopify's Help Center states that orders from agentic storefronts display with "channel or referrer attribution" so merchants know where an order originated. Starting point, not a measurement solution. It tells you where the sale closed, not where the purchase decision formed.
The Scribewise survey found that 70% of marketers with GEO budgets reported GEO strategy accounting for 11 to 20% of their marketing spend. That's meaningful dollars going to a channel where your primary KPI might be "brand mention rate" when you present to finance on Friday.
Yahoo's NewFronts pitch this year is instructive. H&R Block embedded tax deadline reminders directly inside users' email planners via Yahoo's new Planner feature — the ad at least produces an engagement signal. Conversational commerce doesn't have that yet. The interaction happens inside a third-party model with no pixel access.
Monday morning action: before you increase your GEO pilot budget, document exactly what measurement you'll use to defend it. Brand mention rate is a proxy metric, not a conversion metric. If your organization runs on CPA or ROAS targets, you need a plan for the gap between what Shopify's admin shows you and what your attribution platform reports. Otherwise you're not just testing GEO. You're testing how much ambiguity your CFO will tolerate.
The channel is real. The infrastructure is moving fast. The measurement is six months behind the spend.
Mira Castellano covers ad tech for The Daily Vibe.
