How Can Brands Prepare for Agentic Commerce?

That reframes what “preparing” actually means — and it’s why the brands moving fastest (Macy’s, Target, Gap, Sephora) aren’t treating this as a technical project bolted onto marketing. They’re treating it as a rebuild of how their brand gets chosen at all.

Why Agentic Commerce Compresses the Marketing Funnel

For as long as digital marketing has existed, the customer journey looked like a funnel with real gaps between stages — a customer might see an ad, research separately over days, compare on a review site, then finally convert on the third or fourth touchpoint. Every stage was a place brands could invest: content for consideration, retargeting for drop-off, service for retention.

An agent doesn’t move through those stages sequentially on a customer’s behalf. It compresses awareness, research, and comparison into the seconds it takes to query your systems and decide whether to recommend you. There’s no second landing page if the agent’s first read of your product data is incomplete. There’s no retargeting an agent that already moved on to a competitor whose catalog answered the query faster.

This is the part most checklists undersell: the technical requirements below aren’t really about “keeping up with a new channel.” They’re about the fact that your one shot at being chosen now happens entirely inside a machine’s evaluation of your systems, not across a multi-touch relationship you get to build over time.

What that Actually Requires, Starting with the Data

Every serious analysis of this space converges on the same starting point: your product catalog has to be genuinely readable by a machine, not just formatted for a human scrolling a page. Complete, structured attributes, real-time accurate inventory, a standardized taxonomy — the basic inputs an agent needs to find, evaluate, and act on a product in that single compressed moment. An agent can't recommend what it can't reliably read, and unlike a human who might forgive a confusing product page and keep scrolling, an agent simply moves to whichever brand answered the query cleanly.

Speed compounds this. Agents query at machine speed, not browsing speed. APIs that respond slowly, or inventory data that's stale by even a few minutes, don't get a second look — they get skipped in favor of a competitor whose systems responded faster.

Glance has effectively had to solve this from the reverse direction: Glance MCP exists specifically so that a brand’s own product data can become legible to Glance’s commerce intelligence — search, style recommendations, virtual try-on — without either side building a custom integration from scratch. It’s the same underlying problem (make your data readable to an agent, fast) solved as infrastructure rather than as a one-off project.

Adopt the Protocols Agents Actually Use

Once the data itself is in shape, the next real decision is which of the emerging agent protocols to support — because these aren’t competing theories, they’re already live infrastructure with real adoption behind them. The wider vendor landscape building toward this — who's an infrastructure provider versus who owns the actual discovery experience — is broken down separately.

The three that matter right now:

Microsoft Copilot Checkout went live around the same time as UCP, built on the same underlying shift. Today, ACP handles discovery and recommendation, then hands off to the merchant's own checkout to close the sale.

Glance's own experience with this is concrete, not theoretical: it's a named launch partner in Mastercard's Agent Connect infrastructure, announced September 2026. That's the same "which protocol do I actually commit to" decision every brand reading this now faces — Glance made a specific bet on where the transaction layer was heading, rather than waiting to see which protocol won before committing to any of them.

Building Trust and Governance for Agent-Initiated Purchases

Letting an agent transact on a customer's behalf raises a question that has to be answered before the interaction happens, not during it: how does a brand know the agent is authorized, and how does the customer stay in control? The practices already emerging as standard:

•  Transparent consent flows

•  Granular permissions for what an agent can and can't do without asking again

•  Action logs a customer or brand can audit

•  Secure payment authorization at the point of transaction

•  Override mechanisms that let a human step back in

Only about 1 in 10 consumers currently say they're willing to let an agent operate with zero oversight, and brands that can't offer clear guardrails risk losing trust before the category even fully matures. This connects directly back to the funnel-compression problem: if trust isn't already built into the single interaction, there's no follow-up conversation where a brand gets to earn it back.

There’s a real tension here: 71% of marketing leaders report that agents have already weakened their ability to connect directly with customers. Preparing for agentic commerce isn’t purely upside — brands are trading some direct relationship-building for machine-mediated discovery, and that trade is genuinely uncomfortable for teams built around owning the customer relationship.

Don’t Ignore Fulfillment — an Agent Can Buy, But it Can’t Deliver

An AI agent can research, compare, and complete a purchase, but it hits a hard physical limit the moment the transaction closes: someone still has to get the product to the customer's door. A brand's fulfillment and delivery operations need to be just as machine-readable as its product catalog — accurate delivery windows, reliable tracking data, infrastructure an agent can verify before it recommends a purchase in the first place. Brands treating this as a marketing-only problem are missing half of what actually determines whether an agent chooses them, because the compressed moment doesn't end at checkout — it includes whatever confidence the agent can verify about what happens after.

Evaluate your Fit before You Over-Invest

Not every brand needs to prepare for every flow this category will eventually support. Forrester's 2026 Agentic Commerce Framework specifically recommends brands "evaluate their agentic fit" first — assessing whether their actual customers are already experimenting with conversational commerce, and whether their specific products suit non-owned conversational channels at all, before committing significant resources. A practical version: identify the handful of "Do It For Me" journeys your customers are most likely to hand to an agent, test them with real customers, and prioritize based on what you actually learn — not based on which flow sounds most impressive in a pitch deck.

The Market Size and Adoption Numbers Behind This Shift

The numbers behind this shift are large enough that agentic commerce is projected to generate between $3 trillion and $5 trillion globally by 2030, according to McKinsey research. Nearly half of all consumers are expected to use AI agents for brand interactions by the end of 2026, and Shopify has already reported orders from AI-powered search growing 15x year-over-year through 2025. 63% of European shoppers already use AI to compare brands, prices, and reviews, and 55% use it just to learn about a product category — this isn't a future behavior brands are preparing for; it's a current one they're already behind on if they haven't started.

Which Brands Are Already Preparing for Agentic Commerce

Macy's built its own AI shopping agent for curated discovery and virtual try-on. Target partnered directly with OpenAI to bring its catalog into ChatGPT's conversational shopping experience. Gap became one of the first major fashion companies to enable checkout directly inside Google Gemini via UCP. Sephora has been running its own AI-forward path in parallel. None of these are pilots sitting in a lab — they're live, and each is a different bet on which part of the compressed interaction matters most for their specific customer base. 

For brands that don’t want to build that discovery-and-styling layer from scratch, Glance for Shopify adds selfie-based try-on and AI styling chat directly to an existing storefront — the same “meet the compressed moment well” problem Macy’s and Gap solved for themselves, available as infrastructure instead of a custom build.

The Bottom Line

This isn't optional, and it isn't a task to add to an existing marketing checklist. The brands treating 2026 as the year to rebuild for a one-shot, machine-evaluated interaction — not just add agentic commerce as a line item — are the ones positioned to benefit as this category moves from early adoption to default behavior. The ones still planning for a multi-touch funnel that no longer exists in the same way risk discovering that “later” was the moment their competitors became the default recommendation instead of them.

Frequently Asked Questions

How can brands prepare for agentic commerce? 

Start with machine-readable product data — complete, structured, real-time accurate. Then adopt the protocols agents actually use (MCP, UCP, ACP), build trust and governance controls around agent-initiated transactions, and make sure fulfillment infrastructure is verifiable by an agent too. The deeper shift to plan for: agentic commerce compresses the traditional awareness-consideration-conversion funnel into a single interaction, so there’s no second touchpoint to recover from a bad first impression.

What data do brands need to make agent-ready?

Complete and structured product attributes, accurate real-time inventory, and a standardized taxonomy — the baseline an AI agent needs to find, evaluate, and act on a product at all. Fast, well-documented APIs matter too, since agents operate at machine speed and simply move on if a brand’s systems respond too slowly.

Do all brands need to prepare for agentic commerce the same way? No. Forrester’s 2026 Agentic Commerce Framework specifically recommends evaluating your “agentic fit” first — not every brand or product benefits equally from every conversational commerce channel. The more useful exercise is identifying which specific customer journeys are likely to shift to an agent, testing with real customers, and prioritizing from there.

Why does agentic commerce change the marketing funnel, not just add a channel to it? 

Traditional funnels assumed multiple touchpoints — an ad, followed by research, followed by comparison, followed eventually by a purchase — with real opportunities to recover a lost customer at each stage. An agent compresses those stages into one query and one evaluation. If a brand’s data or systems don’t answer that query well the first time, there’s no retargeting campaign that gets a second chance at the same decision.

Does preparing for agentic commerce mean giving up direct customer relationships?

Not entirely, but it’s a real trade-off worth being honest about — 71% of marketing leaders report that agents have already weakened their ability to connect directly with customers. Brands preparing for this shift are trading some of that direct relationship-building for machine-mediated discovery, which is why trust and governance controls (letting a human stay in the loop) matter as much as the technical readiness pieces.

What happens after an agent completes a purchase — is fulfillment part of “readiness” too?

Yes, and it’s frequently overlooked. An agent can research, compare, and complete a transaction, but it can’t physically deliver a product. Brands increasingly need fulfillment operations — delivery windows, tracking accuracy — to be verifiable by an agent before it recommends a purchase in the first place, not just fast in practice.