Shopify reported AI-referred visits and orders tripling year over year in Q2 2026, and those buyers order at roughly twice the new-buyer rate of other channels. It is early, it compounds, and almost nobody has claimed it yet. Getting a client in costs your agency nothing — and every order it earns comes back with a receipt you can put on an invoice.
Keep scrolling. Every step below is the same client — and every step says who does the work.
Agent-ready data generated from your client's catalog. As each product moves through the pipeline we generate the identity, the attributes and the buying questions — then hold them for your team. You approve, edit or reject, and add the category intelligence only you have. Nothing you reject ever ships.
Optimised data for agents. Hover a section to see what each part is for.
Identity is how Google, marketplaces and agents recognise this exact product across every surface. Official GTINs are issued by GS1 — we make the record registration-ready and assign bricks only when confidence is high. Uncertain values are flagged, never shipped. This is the part agencies cannot hand-build at 1,248 variants.
We generate these fields and the Q&A pairs a surface can quote as-is; your review is what makes them true. Approved values ship to every variant. In our ladder experiment, exact-product naming rose from 0% to 45% as the same buying question carried more attributes (p = 0.031, six engines).
Feed hygiene is where in-house AI-commerce projects quietly die: one catalog change and the feed and the storefront start telling different stories. Both are generated from a single source here, checked continuously, and new SKUs are enriched as they appear — so you are never selling a service that decays between renewals.
Every renewal needs a story with a number in it. The weekly report tells you which enrichment patterns are winning so you can apply them to weak products — that is the strategic work you bill for — and the dashboard is the artefact you walk the client through, in your own branding.
What your client gets: entry into more of the queries their budget is already bidding on, plus the free surfaces beside them — Shopping ads, free listings, AI Overviews and AI Mode all read this one record, so one enrichment lifts paid and unpaid together.
Google states it plainly: providing all relevant attributes increases a product's chances of showing for the most relevant queries. Bids decide whether you win the queries you are in — attributes decide which queries you are in.
And a disapproved item is in none of them — it exits the auction until the data cause is fixed. The next section shows how verdicts are watched and repaired.
What your client gets: presence in the assistants Google's feed never reaches — ChatGPT, Perplexity, Claude, Gemini — with nothing to submit and no approval queue. It goes live the moment the page updates, on a domain your client owns outright.
The same facts are published as JSON-LD on the product page, so anything that browses the web can read them there.
What your client gets: the order without the visit. A shopper can find the product and buy it inside the conversation, so your client is sellable to people who never open the site.
The catalog published into Shopify's own agent plumbing — the Shopify Catalog and the store's MCP endpoint — which is what feeds the Shop app, Shopify's global catalog, Microsoft Copilot and ChatGPT's agentic checkout. With no product page in between, the record has to carry the whole answer itself.
What your client gets: entry into more of the queries their budget is already bidding on, plus the free surfaces beside them — Shopping ads, free listings, AI Overviews and AI Mode all read this one record, so one enrichment lifts paid and unpaid together.
Google states it plainly: providing all relevant attributes increases a product's chances of showing for the most relevant queries. Bids decide whether you win the queries you are in — attributes decide which queries you are in.
And a disapproved item is in none of them — it exits the auction until the data cause is fixed. The next section shows how verdicts are watched and repaired.
What your client gets: presence in the assistants Google's feed never reaches — ChatGPT, Perplexity, Claude, Gemini — with nothing to submit and no approval queue. It goes live the moment the page updates, on a domain your client owns outright.
The same facts are published as JSON-LD on the product page, so anything that browses the web can read them there.
What your client gets: the order without the visit. A shopper can find the product and buy it inside the conversation, so your client is sellable to people who never open the site.
The catalog published into Shopify's own agent plumbing — the Shopify Catalog and the store's MCP endpoint — which is what feeds the Shop app, Shopify's global catalog, Microsoft Copilot and ChatGPT's agentic checkout. With no product page in between, the record has to carry the whole answer itself.
Google quietly rejects listings — and a disapproved product exits the auction entirely: no bid can buy back a listing Google refused. Most agencies find out at the renewal meeting. We watch every verdict in real time, name the exact issue and the products it hits, and the same enrichment removes the most common data causes.
Attributes decide which queries a product is in. A rejected listing is in none.
This is the part SEO never gave you. Rankings and sessions were the closest you could get to proving your work; the revenue was always someone else's number. Here every attributable order carries a receipt naming the surface that sent it, so the value of the channel is a figure in your client's own Shopify admin — not a case you have to argue at renewal.
One approval, two billable channels. The record that gets a product into AI answers is the same one Performance Max serves from.
Attributes decide which queries your clients' products are even in. In our ladder experiment, exact-product naming rose from 0% to 45% as the same buying question carried more attributes (p = 0.031, six engines). This is the channel you bill as the new SEO.
“Providing all relevant attributes for a product increases its chances of showing for the most relevant search queries.”
Performance Max Shopping-only campaigns serve from your Merchant Center feed — the same record UCP Fluent enriches.
Performance Max serves from the Merchant Center product data your team just approved. Bids decide whether you win the queries you're in; attributes decide which queries you're in — you can't win an auction you were never entered in. A second line you bill against.
Three structural changes to your commercial position, starting the day you onboard your first merchant.
Your clients pay UCP Fluent directly — and founding-pilot merchants pay nothing: full-catalog enrichment, direct Merchant Center integration, hands-on consulting, in exchange for publishable results. You pay nothing either: no license, no seat fees, no revenue share.
Build the retainer however you want it. Flat monthly, a percentage of what the channel earns, a performance cap, a hybrid — the rate, the floor and the ceiling are yours to write. We never sit between you and your client’s terms, and we never charge your agency anything, so nothing in your pricing has to cover us.
Performance pricing was never really open to you on organic work, because organic could not be measured — you could show rankings and sessions, never revenue. Receipt-level attribution changes what you can put in a contract: a figure sitting in your client’s own Shopify admin that both sides can check. That is what makes a client willing to pay against results.
One application covers both sides: founding commercial terms and direct 1:1 access for your agency, and your selected clients entering through the free founding pilot.
No license, no seat fees, no revenue share.
1:1 with the founding team, not a partner portal.
Selected clients enter free; we publish the measured results.
Tell us roughly how many Shopify clients you run. Every application is read personally.