The Agency Playbook: Adding Agentic Commerce Services in 2026

The short answer: if you run a Shopify or ecommerce agency, your clients now have three concrete reasons to ask you about agentic commerce, and you have three ways to answer. You can refer the work to a specialist, white-label a specialist's delivery under your own brand, or build the practice yourself. Most agencies should start with white-label, sell the work as fixed-scope installs rather than hours, and decide whether to build once their own pipeline proves the demand.

This playbook covers why the questions started, what the work actually is, how to choose between the three options, how to price and staff it, and how to report results. Each section links to the deeper guide. Platform facts are checked against Microsoft's, Shopify's, Amazon's and OpenAI's own announcements and help pages, listed under Sources.

Why are clients asking agencies about agentic commerce now?

Because 2026 turned agentic commerce from a demo into settings in your clients' admin. Three dated changes did it, and each one produces a specific question an agency will hear:

The June change matters most for agencies because it removed the opt-in. Per Shopify's help center, agentic storefronts are active by default for eligible stores and managed under Sales channels → Agentic, with direct checkout inside Copilot, Google AI Mode and Gemini, and Meta. Your clients are already in these channels whether or not anyone configured anything. The question isn't whether to start; it's whether the data those channels read is any good.

One correction worth making in every client conversation: Brand Agents and Copilot Checkout are Microsoft products, not part of Shopify's edition, and none of it requires Shopify Plus. And the in-chat ChatGPT checkout that dominated 2025 headlines is gone: OpenAI wound down Instant Checkout in March 2026, so ChatGPT now sends shoppers to the merchant's own checkout. Getting those facts right is itself a credibility signal. For the full platform map to brief your team from, use the 2026 guide to agentic commerce on Shopify.

What does an agentic commerce service actually include?

It is not one deliverable. It's four bodies of work that share one foundation, the product catalog, and your clients will need some mix of them:

• Off-site channels. Reviewing a client's Sales channels → Agentic settings channel by channel, testing each direct checkout with real edge cases (discount codes, unsupported product types, delivery methods), and fixing the eligibility blockers that keep products out of Shopify Catalog.
• The on-site agent. Installing Microsoft Clarity, joining the Brand Agents beta, rewriting the AI-generated brand identity, adding the privacy disclosure Microsoft requires, and training answers through Shopify's Knowledge Base app.
• Catalog and structured data. Filling attributes, moving prose specs into fields, aligning price, delivery and return facts across product data, schema and feeds. This is the load-bearing work: every channel and the on-site agent read it.
• Commerce GEO and measurement. Tracking whether assistants recommend the client for the prompts that matter, and reporting AI-channel revenue from Shopify's Agentic page and Clarity's agent dashboard.

Commerce GEO

Generative engine optimization for product brands: the work of making a store's products appear, accurately described, when AI assistants such as ChatGPT, Copilot, Gemini and Perplexity answer shopping questions.

Of the four, GEO is the one agencies most often under-scope, because it looks like SEO and isn't. Results are a maintained state, not a delivered artifact, and the measurement is a prompt panel rather than a rank tracker. How agencies package Commerce GEO as a service covers scoping, delivery and the reporting clients actually understand.

How can my agency offer agentic commerce services?

There are three ways, and they form a ladder of increasing commitment and increasing margin. Most agencies should expect to climb it rather than jump to the top rung.

Option one: refer the work out

Honest, fast and free. You keep the relationship healthy and take no delivery risk. The cost is invisible until it isn't: the agentic line item lands on someone else's invoice, and the specialist you referred now has a direct relationship with your client. Referral is right when agentic work is genuinely outside your lane; an agency that only runs paid media shouldn't fake a technical practice. It's wrong if commerce implementation is your core business, because you'd be referring away the center of your own market.

Option two: white-label a specialist

White-label delivery

An arrangement where a specialist firm does the work under the agency's brand and contract; the agency owns the client relationship, strategy and invoice, and the specialist stays invisible to the client.

You sell at your rate, the partner delivers at theirs, and you own the client end to end. It's the option most agencies should take first: delivery capacity now instead of next quarter, a fixed scope you can quote confidently, and a working education in how the delivery runs, which de-risks the build decision later. It's what our Agency Partners program is for.

Option three: build in-house

Full margin, full control, full risk. Building means hiring or training against the competencies below, developing your own delivery method, absorbing the cost of the first slow projects, and keeping up as the platforms change underneath you. In 2026 alone Microsoft, Amazon, OpenAI and Shopify each changed something material. Build when demand is proven in your own pipeline, not in a headline.

Should agencies white-label or build agentic delivery?

It's a capacity and timing question more than a capability question. Most agencies could build the practice eventually; the issue is what the ramp costs while clients are asking now. Here are the dimensions that actually decide it:

White-label isn't automatically forever. Some agencies will graduate to in-house, and the ones who do it well will have learned the delivery shape from working next to a specialist first. The failure mode is the middle path: half-building while half-referring, so you carry the costs of both and the advantages of neither. White-label vs. in-house delivery has the full decision matrix by agency size and client mix.

How should agencies price agentic commerce services?

Not by the hour. The work repeats across clients, the client is buying an outcome (a configured agent, a machine-readable catalog, measurable AI-channel visibility) rather than your time, and hourly billing shrinks your margin exactly as your team gets faster. We don't publish numbers, and you shouldn't copy anyone's, but the logic transfers.

Productized install

A fixed-scope, fixed-price, fixed-timeline engagement with a named deliverable list, such as a configured Brand Agent, enriched priority SKUs, tested AI-channel checkouts and a measurement baseline, sold as a SKU rather than scoped from scratch per client.

A productized agentic menu usually has three tiers, each priced on a different logic:

1. 01

   The audit (entry SKU, fixed fee)

   Readiness across catalog attribute coverage, structured data, product pages, AI-channel settings and assistant visibility. Cheap to deliver once templated, and every gap it finds becomes a line on the install proposal.

2. 02

   The install (core SKU, fixed fee)

   Agentic storefront controls set and tested, Brand Agent configured, priority SKUs enriched, Knowledge Base answers overridden, baseline measurement in place. Catalog enrichment beyond the priority SKUs is priced by throughput: SKUs enriched to a defined schema per week.

3. 03

   The retainer (recurring)

   Ongoing enrichment as the catalog grows, agent-answer review, Commerce GEO monitoring, and reconfiguration as the platforms change. This is where agency economics live; the install earns the retainer.

Two rules hold regardless of your numbers. Price against the client's alternative, which is hiring a specialist directly or losing AI-channel sales while they wait, not against your costs. And the deliverable list is the contract: name the SKU counts, the channels configured and the reports delivered, because vague scope turns fixed-price work into unpaid hourly work. One more: don't quote your first install at the rate of your fifth; the ramp cost belongs in the price or the timeline. Pricing agentic services: fixed-scope logic for agencies works through fixed fees, throughput pricing and retainer structure.

What skills does an agentic commerce practice need?

Not one hire: five competencies, several of which will live in the same person in a small practice. The useful question for each is which existing role it's closest to, because that tells you what to train versus what to buy.

The role that ties these together is an agentic commerce lead: someone who can talk to a client's merchandising team about attributes in the morning and test a Copilot checkout in the afternoon. You don't need one before your first install if you white-label, but you will before you build. Hiring an agentic commerce lead has the role scorecard, interview signals and a sample job description.

How do you pitch agentic commerce to a client?

Lead with what's already true about their store, not with the technology. Since June 2026 an eligible Shopify store is in AI channels by default, so the first slide isn't "AI shopping is coming"; it's "here is what ChatGPT, Copilot and Gemini can currently see about your products, and here is where it's wrong". That reframes the pitch from speculative to corrective.

A deck that works follows four beats. The shift: the three dated changes above, in one slide, with the vendor named correctly for each. Their gap: findings from a readiness audit and a citation check on their own products and category prompts. The options: what they could do themselves, what you'd do, and what it costs to wait. The close: a fixed-scope install with named deliverables and a date. Keep claims to what the platforms document; a client who later finds that Brand Agents isn't a Shopify feature, or that ChatGPT no longer checks out in the chat, stops trusting the rest of the deck.

Two free tools give you the evidence for the gap slide: the AI Citation Check for the off-site half and the Readiness Score for the store itself. The agency pitch deck for agentic commerce annotates the full deck slide by slide.

How do you report agentic commerce results to clients?

With first-party numbers, which in 2026 you finally have. Shopify's Agentic page reports sales, orders, online-store sessions and conversion for each AI channel, combining referral sales with direct-checkout sales, plus the search queries the client's products appeared for. Clarity's Brand Agents dashboard reports engagement, conversions and sales impact, and compares agent-assisted sessions with organic traffic. Shopify's Knowledge Base app logs the questions shoppers ask AI agents about the store, answered and unanswered.

A quarterly business review for an agentic engagement has three sections. Metrics: AI-channel revenue and orders by channel, agent-assisted conversion against the site average, and the citation-panel results for the client's priority prompts. What the conversations told us: the unanswered questions and dropped chats, each mapped to a catalog gap or policy fix. Next quarter: the roadmap, including whatever the platforms changed since the last review, which in 2026 has been something every quarter.

The second section is what renews the retainer. Clients understand a list of real shopper questions their store couldn't answer, and they understand a fix with a date next to it. The QBR template for agentic engagements gives you the structure and the metrics section ready to fill.

What can go wrong when an agency sells agentic commerce?

Most of the risk is in promises, not delivery. Four come up repeatedly, and each has a cheap guard.

• Selling yesterday's platform. Plenty of 2025 decks still promise checkout inside ChatGPT; OpenAI wound Instant Checkout down in March 2026. Re-check every platform claim in your proposals each quarter against the vendor's own help pages.
• Overpromising a beta. Microsoft lists Brand Agents as a beta that merchants join through a waitlist. Don't put a go-live date for the agent in a contract; put the work you control (catalog, Knowledge Base answers, checkout tests) on the timeline instead.
• Consent your client didn't notice. Accepting the Brand Agents terms publishes the catalog to Microsoft Merchant Center, and Microsoft requires a privacy-policy disclosure that conversation content is shared with it. Get the client's sign-off in writing before you click accept.
• Reporting what you can't measure. Promise the first-party metrics Shopify and Clarity actually report, not a revenue-lift percentage. A baseline taken before the install is what makes any later number credible.

The underlying discipline is the same one that makes the delivery work: say only what the platforms document, and scope only what you control. For a client-facing version of the readiness checks, the agentic commerce readiness checklist works as a leave-behind.

What does white-label agentic delivery with GigaCommerce look like?

Our Agency Partners program is option two from the ladder. In summary: a dedicated pod works as an extension of your team, in your project tools, with your brand on every report, and never speaks to your client unless you want it to. Pods come in three tiers, from a single-service specialist to a full-operation team, with response-time commitments by tier. Pricing is scoped per partner on a call; we don't publish rates. Non-solicitation of your clients is contractual and runs both ways, and project-scoped work such as a single audit or install is available alongside retainers.

The service families cover the work in this playbook: agentic storefronts and Copilot checkout configuration, Brand Agents, catalog enrichment, Commerce GEO and Amazon listings for Alexa for Shopping. Applying takes a short form, and we reply with a fit assessment within 48 hours. If you're still deciding between white-label and building, the white-label vs. in-house guide is the better next read; it's written to be useful even if you never partner with us.

What should an agency do in its first 60 days?

Small and specific. Not a practice launch: three clients and a repeatable motion.

1. 01

   Days 1–10: pick three clients

   Shopify clients with catalogs deep enough to matter who trust you enough to try something new. Open each one's Sales channels → Agentic page and note what's active and where each channel closes the sale.

2. 02

   Days 10–25: run the audits

   Catalog attribute coverage, structured data, AI-channel settings, a test order in each direct checkout, and assistant visibility for their category prompts. Present findings as findings, not as a pitch.

3. 03

   Days 25–40: sell the first install

   At least one audit will surface gaps the client wants closed. Quote a fixed-scope install. Deliver it white-label if you can't yet deliver it yourself; the client experiences your brand either way.

4. 04

   Days 40–60: write the SKU sheet

   Turn what you sold into a one-page menu (audit, install, retainer) with named deliverables. Then decide, with real pipeline data, whether the build case holds, and revisit it quarterly.

Notice what this plan leaves out: hiring, a rebrand, a webinar. Capability first, marketing second. An agency with one delivered install and a tight SKU sheet will outsell one with a polished "AI Commerce" page and nothing delivered.